Post on 13-Jul-2020
UvA-DARE is a service provided by the library of the University of Amsterdam (http://dare.uva.nl)
UvA-DARE (Digital Academic Repository)
Assessment of pharmacovigilance approaches for monitoring the safety of antimalarial drugsin pregnancy
Dellicour, S.O.M.C.
Link to publication
Citation for published version (APA):Dellicour, S. O. M. C. (2014). Assessment of pharmacovigilance approaches for monitoring the safety ofantimalarial drugs in pregnancy. Alblasserdam: Dutch University Press.
General rightsIt is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s),other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons).
Disclaimer/Complaints regulationsIf you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, statingyour reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Askthe Library: https://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam,The Netherlands. You will be contacted as soon as possible.
Download date: 24 Aug 2020
Assessment of Pharmacovigilance Approaches
for Monitoring the Safety of Antimalarial
Drugs in Pregnancy
Stéphanie Dellicour
Assessm
ent of Pharmacovigilance A
pproaches for Monitoring the Safety of A
ntimalarial D
rugs in PregnancyStéphanie D
ellicour
Assessmentofpharmacovigilanceapproachesformonitoringthesafetyofantimalarial
drugsinpregnancy
2
© Stephanie Dellicour, 2014
All rights reserved
Dutch University Press
Alblasserdam
ISBN 978 90 361 0408 1
NUR 870
ASSESSMENTOFPHARMACOVIGILANCEAPPROACHESFOR
MONITORINGTHESAFETYOFANTIMALARIALDRUGSINPREGNANCY
ACADEMISCH PROEFSCHRIFT
ter verkrijging van de graad van doctor
aan de Universiteit van Amsterdam
op gezag van de Rector Magnificus
prof. dr. D.C. van den Boom
ten overstaan van een door het college voor promoties ingestelde
commissie, in het openbaar te verdedigen in de Aula der Universiteit
op woensdag 22 oktober 2014, te 13:00 uur
door Stéphanie Ondine Marie Cornélie Dellicour
geboren te Uccle, België
4
Promotiecommissie Promotores: Prof. dr F.O. ter Kuile
Prof. dr M. Boele van Hensbroek
Copromotor: Dr M. Desai Overige leden: Prof. dr R.C. Pool Prof. dr F.G.J. Cobelens Dr P.F. Mens Prof. dr M.W. Borgdorff Dr J. Webster Dr M.J. Rijken Faculteit der Geneeskunde
To Per and Lea
6
Contents Page Number Chapter 1: General Introduction 7 Chapter 2: Dellicour S, Hall S, Chandramohan D, Greenwood B. The safety of
artemisinins during pregnancy: a pressing question. Malaria Journal 2007, 6:15.
23
Chapter 3: Dellicour S, Tatem AJ, Guerra CA, Snow RW, ter Kuile FO. Quantifying
the Number of Pregnancies at Risk of Malaria in 2007: A Demographic Study. PLoS Medicine 2010, 7(1):e1000221.
37
Chapter 4: Dellicour S, ter Kuile FO, Stergachis A. Pregnancy Exposure Registries
for Assessing Antimalarial Drug Safety in Pregnancy in Malaria‐Endemic Countries. PLoS Medicine 2008, 5(9):e187.
65
Chapter 5: Dellicour S, Brasseur P, Thorn P, Gaye O, Olliaro P, Badiane M,
Stergachis A, ter Kuile FO. Probabilistic Record Linkage for Monitoring the Safety of Artemisinin‐Based Combination Therapy in the First Trimester of Pregnancy in Senegal. Drug Safety 2013, 36(7):505‐513.
79
Chapter 6: Dellicour S, Desai M, Mason L, Odidi B, Aol G, Phillips‐Howard PA,
Laserson KF, Ter Kuile FO. Exploring Risk Perception and Attitudes to Miscarriage and Congenital Anomaly in Rural Western Kenya. PLoS One 2013, 8(11):e80551.
97
Chapter 7: Riley C, Dellicour S, Ouma P, Kioko U, ter Kuile FO, Ahmeddin O,
Kariuki S, Buff AM, Desai M, Gutman J. Assessment of Knowledge and Adherence to the National Guidelines for Malaria Case Management in Pregnancy among Healthcare Providers and Drug Outlet Dispensers in rural, Western Kenya. To be submitted
113
Chapter 8: Dellicour S, Desai M, Aol G, Oneko M, Ouma P, Bigogo G, Burton D,
Breiman R, Hamel M, Slutsker L, Feikin D, Kariuki S, Odhiambo F, Pandit J, Laserson KF, Calip G, Stergachis A, ter Kuile FO. Risk of inadvertent exposure to artemisinin derivatives in the first trimester of pregnancy and its association with miscarriage: a prospective study in Western Kenya. To be submitted
141
Chapter 9: Discussion and Summary 179 Samenvatting 191 Acknowledgements 197 About the author 199
7
Chapter1
GeneralIntroduction
Chapter 1
8
PharmacovigilanceAll drugs have the potential to cause harm and their risk‐benefit profiles need to be determined to
maximize therapeutic outcomes.[1] Attention to the importance of drug safety monitoring was
brought to light by the thalidomide tragedy in the 1960s.[2,3] Thalidomide was used as an
antiemetic and sedative which was considered safe including during pregnancy. More than 10,000
malformed children with phocomelia (with limbs severely underdeveloped or absent) were born
before a safety signal was identified and thalidomide was withdrawn.[4] This had a profound impact
on drug regulatory processes from registration to post‐marketing surveillance.[5]
Pharmacovigilance, sometimes referred to as drug monitoring or post‐marketing surveillance, is
defined by the World Health Organization (WHO) as “the science and activities relating to the
detection, assessment, understanding, and prevention of adverse reactions or any other possible
drug‐related problems”.[1] Fundamental aims are early detection of unknown safety issues,
quantifying risk and risk factors associated with adverse drug reactions (ADRs) and primarily
preventing unnecessary harm to patients.
Before a drug is marketed, drugs are tested pre‐clinically in‐vitro and in animal studies for toxicity. If
a compound passes these tests it goes through a series of clinical trials (see figure 1). Drug
regulatory authorities make their decision based on evidence regarding efficacy, safety and quality
deemed adequate for a drug to be approved. By the time a drug is marketed, information on safety
and efficacy has been derived from limited number of subjects (between 500‐5000) who are
typically carefully selected and followed under controlled conditions for a relatively short amount of
time.[6] This limits the possibility of detecting rare adverse events (the rule of three suggests that by
the time of registration, safety information is typically only available for adverse events occurring at
frequencies of 0.1 to 1%1[7]), or events with slow onset such as cancer or events occurring in
vulnerable groups such as pregnant women, children and the elderly or people with co‐morbid
conditions and concomitant medications which are usually excluded from pre‐registration clinical
trials.
Passive surveillance through spontaneous reporting is the backbone of post‐marketing surveillance.
This entails identification of a suspected ADR following drug exposure by healthcare professionals,
assessments of severity and causality and reporting according to standard procedures. Passive
surveillance is useful for hypothesis generation and identification of new safety signals but has its
limitations such as under‐reporting, variable quality and completeness of the reported
data, tendency for reporting of known reactions and false causality attribution.[8] Active surveillance
approaches are essential to assess and evaluate the nature and rates of adverse events as they can
provide denominator data. The most common pharmacovigilance active surveillance involves
identifying individuals exposed to a drug of interest and following them systematically to assess for
adverse events (e.g. Prescription Event Monitoring system used in the UK [9]). Other approaches
using pharmacoepidemiological methods, include cohort or case control designs, are used for safety
signal confirmation or characterization.[10,11] Deciding on an approach should be based on the
safety specification of a product and whether the purpose of the investigation is hypothesis or signal
1 The rule of three is commonly used in pharmacovigilance to estimate the sample size needed to detect adverse events. It
is based on the assumption that there is 95% chance of observing one occurence of an event in a population 3 times the size of the event’s rate (e.g. to pick up a rare event occuring at a rate of 1 /10 000, 30 000 individuals will need to be monitored to be 95% certain of detecting this event).
Introduction
9
generating; hypothesis strengthening or a confirmatory study, (i.e. evaluation of a known safety
signal).
Figure 1 Clinical development of medicines adapted from “WHO Policy Perspectives on Medicines — Pharmacovigilance: ensuring the safe use of medicines”.[6]
PharmacovigilanceinresourceconstrainedsettingsThe public health burden of ADRs and medication errors is difficult to quantify but it is likely to be
worse in resource constrained settings because of weak health systems, overstretched and often
inadequately trained healthcare staff, unreliable supply and quality of drugs, polypharmacy (use of
multiple drugs concomitantly), concomitant use of herbal remedies as well as availability of most
prescription drugs from the informal market.[12] Few low and middle‐income countries have
functioning national pharmacovigilance systems. Currently, less than 2% of ADR reports collated at
the Uppsala Monitoring Centre of the WHO Programme for International Drug Monitoring (UMC,
which collates reports from national pharmacovigilance centres globally [13]) come from low or
middle‐income countries.[14] Indeed, most of these countries make national treatment or
prevention policy decisions with reliance on drug safety data from industrialised countries where
pharmacovigilance systems and regulatory authorities are better established. However, this often
comes at the expense of no or limited safety data for drugs targeting tropical diseases which are
seldom encountered in these industrialised settings.
While there is increasing access to medicines to combat HIV, malaria and tuberculosis in the tropics,
systems are needed to assess the risk‐benefit balance of these much‐needed interventions.[15] Drug
safety monitoring cannot rely on passive spontaneous reporting systems for ADRs due to the
limitations mentioned above. Targeted pharmacovigilance approaches are required in situations
where no systematic recording of drug exposure and suspected ADRs exist particularly where health
resources are limited. Passive surveillance has been proposed at selected sentinel sites where
healthcare professionals are trained and incentivised to report suspected severe ADRs to specific
drugs of interest.[16] Building on existing infrastructure and surveillance systems or studies would
provide more cost‐effective ways to collect reliable and timely data. Demographic surveillance sites
Preclinical Animal Experiments
Phase I > Phase II > Phase IIPhase IV
Post Approval
Development
Registration
Post registration
Phase I20‐50 healthy volunteers to
gather preliminary data
Animal experiments for acute toxicity, organ damage, dose dependence,
metabolism, kinetics, carcinogenicity, mutagenicity & teratogenicity
Phase III250‐4000 varied patient groups to determine short term safety and
efficacy
Phase II150‐350 patient with disease to determine safety and dosage
Phase IVPost‐approval studies to
determine specific safety issues
Chapter 1
10
are increasingly being considered for use as platforms for post‐marketing surveillance. Such sites
monitor vital events (births, deaths and migration) through regular censuses of a pre‐defined
population (typically 2 to 4 times per year), most also have links to clinical and treatment data from
local health facilities. They have well‐defined and characterized populations, and are valuable to
monitor public health interventions where health information systems are weak as they provide
denominator data, a framework to identify specific groups (such as pregnant women or children)
outside the healthcare setting as well as human resources and infrastructure for research.[17] Five
African sites that are part of the International Network for the Continuous Demographic Evaluation
of Populations and Their Health (INDEPTH) are conducting Phase IV studies to monitor antimalarial
effectiveness and safety in “real life” settings.[18] Antimalarial safety is being monitored through a
Spontaneous Adverse Events Reporting System (passive surveillance) and Cohort Event Monitoring
(CEM; active surveillance). The spontaneous reporting system is being strengthened in the selected
sites ensuring health workers are sensitised and familiar with the ADR reporting procedures. CEM
protocols involve active follow up of patients prescribed antimalarials on days 3 and 7 following
treatment and systematic recording of all clinical events during that time. This enables the
characterisation of known ADRs in terms of incidence as a denominator is available (which is not the
case for systems based on spontaneous reporting) and identification of risk factors as well as
detection of unknown safety signals. Another approach is to capitalize on planned studies, such as
those conducted by the Malaria in Pregnancy Consortium (MiPc) and the ACT consortium (ACTc)
which have set up a centralized safety database collating ADRs collected across studies using
standardized reporting procedures and tools.[19,20] Such systems could then be extended to other
drugs and expanded to additional sites through new collaborations.
DrugsafetyinpregnancySpecial considerations apply to the assessment of drug safety in pregnancy, particularly for drugs
used to treat diseases that are only endemic in resource constrained settings. Most drugs are
marketed with limited information on their safety when used during pregnancy. Pre‐approval
studies have inherent limitations in determining the safety of drugs used during pregnancy. Animal
reproductive toxicology studies have ambiguous predictive value for human teratogenesis due to
variations in species‐specific effects [21] and pregnant women are routinely excluded from pre‐
licensure clinical trials for fear of harming the mother or the developing fetus.[22,23] Physiological
changes associated with pregnancy limit the inference of pharmacokinetic and pharmacodynamic
data from non‐pregnant subjects.[24] Consequently, most drugs are not recommended for use
during pregnancy due to the lack of information on their risk‐benefit profile. Yet drugs are widely
used by pregnant women. Medication often cannot be avoided for chronic diseases, such as asthma,
epilepsy, malignant diseases, HIV, or other illnesses which may harm the mother and the unborn
baby if left untreated. Furthermore, many women are likely to be inadvertently exposed to drugs at
the most vulnerable time of embryo development early in gestation before pregnancy status is
recognised. Healthcare providers, pregnant women and policy‐makers need valid information in
order to make informed decisions about the use of drugs during pregnancy.
Introduction
11
PharmacovigilanceapproachesfordrugsusedinpregnancyThere are several methodological challenges to the study of safety of drugs in pregnancy, including
those common to overall pharmacovigilance methods such as the need for large sample sizes to
minimise the possibility that the observed association between a drug and a rare outcome occurred
due to chance; the possibility of confounding by indication which implies studies should take into
account the contributing risk of the underlying maternal illness; the general lack of background rates
needed to put signals into context and the potential self‐referral bias introduced by voluntary
reporting.[25,26] Special considerations are also required for assessment of outcome and drug
exposure. Monitoring drug safety in pregnancy necessitates systematic recording of pregnancy
outcomes, examining the newborns and possibly following up the infants to detect anomalies not
detectable at birth. Where deliveries occur outside health facilities, as is the case in many rural
settings in low and middle income countries and particularly for early pregnancy loss, systems need
to be put in place to capture all pregnancy outcomes. Ascertainment of drug exposure requires
reliable information on the drug and gestational age at the time of exposure. This is important as the
impact on the fetus from drugs used in pregnancy depends on the stage of pregnancy at the time of
exposure as different tissues and organs have specific developmental timelines.[27,28] These
methodological considerations are described in detail in chapter 4 based on the case of artemisinin
combination therapies used for the treatment of malaria.
Passive surveillance (i.e. spontaneous reports of suspected ADRs) for detecting drug with potential
embryo‐fetal adverse effects relies on judicious clinicians to make the association between a drug
exposure in pregnancy and an adverse event which will often occur months and sometimes years
later (for congenital anomalies not detectable at birth). In resource constrained settings, such an
approach has limited use due to under‐ascertainment of outcomes such as miscarriages and birth
defects in communities where birth anomalies are considered taboos, as well as the general lack of
awareness around pharmacovigilance and the need to report suspected ADRs.
Pregnancy exposure registries, a type of cohort design, are the most common approach to monitor
drug safety in pregnancy in industrialised countries (e.g. the Antiretroviral Pregnancy Registry[29] or
the Antiepileptic Drug (AED) Pregnancy Registry[30]). The U.S. Food and Drug Administration (FDA)
and the European Medicines Agency (EMA) recommend pregnancy exposure registries for products
that are likely to be used during pregnancy or by women of reproductive age, particularly if there
have been case reports of adverse pregnancy outcomes following exposure, drugs in the same
pharmacological class are known to pose risk during pregnancy or pre‐clinical animal data suggests
potential teratogenic risk.[31,32] A pregnancy exposure registry involves active identification and
recruitment of pregnant women exposed to a drug or class of drug of interest and following these
women throughout pregnancy. This approach has many advantages such as active enrolment and
pregnancy outcome ascertainment to minimise loss to follow up, centralised and prospective
ascertainment of exposed pregnancies which reduces biases. Drawbacks of most pregnancy
registries are the limited sample size to detect moderate effects, the lack of outcome validation and
standardisation of birth defect assessments.[27] To overcome the sample size requirement
pharmacoepidemiologic studies using existing databases are often used in high income countries.
Cohort studies use large databases such as medical insurance claim databases (e.g. the Medicaid
database in the US[33]), electronic medical records (e.g. GPRD in the UK[34]), or data from
teratology information services (e.g. the Motherisk Program in Canada[35]). Birth defect registries
and case‐control surveillance systems are also used to assess associations between medications and
Chapter 1
12
congenital malformations (e.g. the Birth Defects Study by Slone Epidemiology Centre[36]).
Applicability of these methods in low and middle income countries is unknown.
Malariainpregnancy
EpidemiologyandburdenMalaria is a major public health problem affecting an estimated 3.4 billion people worldwide (half of
the world’s population) which live in areas at risk of malaria. In 2012, an estimated 207 million cases
of malaria occurred globally, resulting in over half a million deaths.[37] This life threatening disease
is preventable and treatable through proper use of antimalarial drugs, insecticide treated bednets
and indoor residual spraying of insecticides. Since 2000, a significant decrease in malaria mortality
rates have been observed due to the vast increase in the financing and coverage of malaria control
programmes.[37]
Pregnant women and children under five are most at risk of malaria. Malaria in pregnancy (MiP) can
have devastating consequences for the mother and fetus, including severe maternal anaemia and
mortality, miscarriage, intrauterine growth retardation and low birthweight (LBW), preterm birth
and stillbirth.[38,39] It is estimated to cause as many as 900,000 low birth weight births, and
100,000 infant deaths every year. Furthermore, up to a quarter of maternal mortality in endemic
countries could be attributable to malaria.[39,40,41,42,43]
PreventionguidelinesWHO recommends two approaches for the prevention of MiP in areas with stable malaria
transmission (where the population is exposed to a fairly constant, high rate of malaria infected
mosquito bites [>10 per person per year]): intermittent preventative treatment in pregnancy (IPTp)
with sulfadoxine‐pyrimethamine (SP) at each antenatal clinic visit following quickening and the use
of insecticide‐treated bed nets (ITNs) to protect pregnant women from the bites of infected
mosquitoes.[44,45] Coverage with these prevention strategies remains very low. In 2007 it was
estimated that only about 39% and 25% of pregnant women in sub‐Saharan Africa used ITNs and
received at least two doses of IPTp, respectively.[46]
TreatmentguidelinesPregnant women, when infected, require rapid diagnosis and case management with safe and
effective antimalarial drugs to prevent progression to severe disease or death, or to prevent
asymptomatic infections from becoming chronic leading to fetal growth restriction and malaria‐
related maternal anaemia.[38,39] Established antimalarials like chloroquine and SP are no longer
recommended for the case management of malaria illness due to drug resistance of the parasite.
Artemisinin‐based combination therapies (ACTs) are now the recommended treatment for
uncomplicated malaria in sub‐Saharan Africa.[47] Artemisinin derivatives have been used for
centuries in China but they have only been deployed for use as ACTs since 2001.[47,48] Based on the
level of drug resistance to the partner medicine, the following fixed‐dose combination ACTs are
recommended: artemether plus lumefantrine; artesunate plus amodiaquine; artesunate plus
mefloquine; artesunate plus SP and dihydroartemisinin plus piperaquine.
In the first trimester of pregnancy an ACT is indicated only if an alternative antimalarial is not
immediately available, or if treatment with 7‐day quinine (without, or combined with, clindamycin)
fails or there is uncertainty of compliance with the 7‐day quinine regiment. The recommendation for
treatment of pregnant women with severe malaria includes parenteral artesunate because it is
Introduction
13
faster acting than parenteral quinine and is associated with improved survival in patients with severe
malaria.[49,50,51]
Antimalarialsafetyinpregnancy
AntimalarialsconsideredsafethroughoutpregnancyAntimalarials recommended for pregnant patients must be safe for the mother and her unborn
baby. There is insufficient information on the safety of most antimalarials in pregnancy, particularly
for exposure in the first trimester. A retrospective study in Thailand, reported the deleterious effect
of even a single episode of malaria in the first trimester and its association with an increase in the
risk of miscarriage, emphasising the need for safe and efficacious drugs in this critical period.[52]
This is further supported by a recent modelling study reporting that up to two thirds of placental
infections occur by the end of the first trimester.[43] Antimalarial medicines considered safe in the
first trimester of pregnancy are quinine (with or without clindamycin), chloroquine, proguanil and
more recently mefloquine although this is based on limited evidence. As mentioned above,
chloroquine is no longer recommended for treatment of falciparum malaria. Proguanil and
chlorproguanil although deemed very safe in pregnancy, have limited value in resource constrained
settings as the current formulations available are either too expensive (atovaquone‐proguanil) or
carries the risk of acute haemolysis in patients with glucose‐6‐phosphate dehydrogenase (G6PD)
deficiency (which has a prevalence of up to 30% in some African countries[53]) when in combination
with dapsone. This combination is no longer available since its marketing authorization holder
withdrew it from the market in 2008.[54] Animal studies (in rodents, dogs and primates) did not find
quinine to have embryo‐fetal toxicity except one study which reported congenital malformations in
5% of pups born to rats receiving quinine.[55,56,57] There was no evidence of embryo‐fetal toxicity
in published human data. Quinine is often thought to have abortifacient properties at high dosage.
The author of a commonly cited publication from 1908 reports the beneficial effect of quinine as “It
may be laid down as an almost invariable rule that if in a pregnant woman suffering from malarial
fever uterine action has begun, the quinine will probably hasten the abortion; but that if uterine
action has not begun, it will not start it, but will, on the contrary, probably be the means of saving
the pregnancy”.[58] A recent observational study in Tanzania reported an increased risk of
pregnancy loss in women exposed to quinine in the first trimester [59] however it is not clear
whether the observed effect could be explained by the underlying malaria infection, especially
because of the potential for poor compliance with the 7‐day quinine regimen given the poor
tolerability of quinine due to cinchonism (a syndrome which includes symptoms of ringing of the
ears, blurred vision, headache, nausea and dizziness) and hypoglycaemia.[60,61] Mefloquine has
been recommended as a prophylactic for pregnant travellers by WHO and CDC. Recently after
review of existing evidence the FDA and CDC changed their recommendation for the use of
mefloquine in pregnancy both as a prophylactic and for treatment.[62] Mefloquine has an
acceptable reproductive toxicity profile in animal studies at standard doses and data on 1000 first
trimester exposures suggests no increase in risk to adverse pregnancy outcomes. Although 1
retrospective study found an increase in the risk of stillbirths associated with mefloquine treatment
doses in pregnancy this finding has not been confirmed by subsequent studies.[60,61,63] The result
of a multi‐centre randomised controlled trial in over 4000 second and third trimester pregnant
women did not find an increased risk of stillbirth with two or three 15 mg/kg treatment doses of
mefloquine compared to SP when used as IPTp. However tolerability was low, even when 15 mg/kg
Chapter 1
14
was given as a split dose over two days, with 30% of pregnant women reporting vomiting and
dizziness which could limit the use of mefloquine for IPTp.[64]
AntimalarialscontraindicatedinthefirsttrimesterofpregnancySP, which has limited use due to increasing drug resistance, is still used for the prevention of malaria
in pregnancy through IPTp in most areas with high malaria transmission. As an anti‐folate it is not
recommended in the first trimester of pregnancy due to risk of neural tube defects.[65] Animal
studies found embryotoxicity at high doses including cleft palate in rat models.[66] Documented use
in over two thousand pregnancies in the second and third trimester of pregnancy found no evidence
of embryotoxicity.[60,61] Amodiaquine is considered safe in pregnancy although there are neither
documented data on exposure in the first trimester of pregnancy nor animal reprotoxicology
studies.[67,68] Lumefantrine is only available as an ACT in combination with artemether which is
contraindicated in the first trimester of pregnancy (see below) and data from animal studies did not
show any embryotoxic effect.[60,61] Piperaquine, which is also only available as an ACT, has a safety
profile expected to be similar to that of chloroquine. Animal reprotoxicology studies of piperaquine
found no safety concerns except that gestation was prolonged in exposed rats.[69,70] The few
studies on the use of the combination dihydroartemisinin‐piperaquine in second and third trimesters
of pregnancy reported favourable outcomes for pregnant women.[71,72,73] Further trials are
underway with women in their second and third trimesters.[74,75,76,77,78]
ArtemisininderivativesinearlypregnancyAlthough ACTs are highly effective in treating malaria the safety of artemisinin derivatives during
early pregnancy remains to be determined. Animal reprotoxicology studies showed that artemisinin
derivatives have embryotoxic effects in all species studied (i.e. rat, rabbit and monkey) at low dose
ranges.[79,80] Pre‐clinical studies showed that the mechanism of embryotoxicity was through insult
to immature red blood cells (primitive erythroblasts) causing severe anaemia in the embryo and
leading to either embryolethality or malformations, skeletal (shortened or bent long bones and
scapulae, misshapen ribs, cleft sternebrae and incompletely ossified pelvic bones) and
cardiovascular (ventricular septal and vessel defects).[79,81] These studies also predicted that the
main window for insult to the fetus will occur early in pregnancy (between 4 and 10 weeks post
conception).[79] This is the period when the nucleated, metabolically active primitive erythroblasts
predominate in the blood. However, the exact moment when humans are most sensitive is unknown
as the primitive erythroblasts are gradually replaced by enucleated mature erythrocytes (which are
less sensitive to the effects of artemisinins) over several weeks. Information regarding risks
associated with the use of antimalarials in pregnancy in humans is sparse. Although the limited data
on human exposures is reassuring for first trimester exposures 2 , further information is
required.[61,82,83] With the widespread deployment of ACTs, the potential for inadvertent
exposure early in pregnancy is high when neither the physician nor the patient is aware of the
pregnancy.
It is essential to study the risks of artemisinins and ACT drugs in early pregnancy, and that vigilant
attempts are made to establish systems for the systematic collection of pregnancy‐drug exposure
data. It is unknown how best this can achieved in resource constrained malaria endemic countries.
The specialized nature of the reliable assessment of drug exposures, pregnancy outcome and
2 At the time of the first review in chapter 2 there were 123 documented exposures this number is now close to 760 as discussed in the last chapter of this thesis.
Introduction
15
malformations is not easily achievable from routine surveillance systems such as national
pharmacovigilance programmes, where they exist, and will require sentinel sites with enhanced
active surveillance or dedicated studies, good record keeping and follow‐up systems, and training of
staff to examine newborns.
ThesisAimandOutlineThe aim of this thesis is to develop and evaluate such targeted pharmacovigilance systems to assess
the safety of the ACTs in early pregnancy.
This thesis is divided into two components: desk‐based reviews (chapters 2‐4) and field‐based
studies (chapters 5‐8) with the overall aim to assess pharmacovigilance approaches to provide better
estimates of the risk‐benefit profiles of ACTs used in early pregnancy.
Objectivesandoutline1) Review existing evidence on human exposure to ACTs in pregnancy with the emphasis on
first trimester exposures (chapter 2)
2) Derive global estimates of the number of pregnancies at risk of malaria based on a
contemporary map of malaria transmission and demographic data for pregnancy and
fertility rates to provide an estimate of the scale of the problem posed by MiP (chapter 3)
3) Describe the methodological considerations for setting up antimalarial pregnancy exposure
registries in resource constrained settings (chapter 4)
4) Assess the feasibility of record linkage using routinely collected healthcare data as a
pragmatic means of monitoring antimalarial safety in early pregnancy. A study involving
extraction of data from health records from a dispensary in south‐western Senegal is
reported (chapter 5)
5) Explore community perceptions of miscarriage and congenital anomalies. Findings from 10
focus group discussions carried out in western Kenya are described which provide insight
for studies of pregnancy outcomes in similar rural African settings (chapter 6)
6) Assess prescribing behaviour and knowledge of malaria treatment guidelines for pregnant
women among healthcare providers and drug outlet dispensers in an area of high malaria
transmission. The results from a cross‐sectional survey in rural western Kenya are
presented providing insight on the scale of ACT prescribing in early pregnancy (chapter 7)
7) Assess the risk of miscarriage associated with exposure to ACTs in the first trimester of
pregnancy. The findings from a prospective cohort study of women of childbearing age in
Western Kenya are presented (chapter 8)
8) Draw conclusions on the potential of the proposed pharmacovigilance approaches for drugs
used by woman of childbearing age and pregnant women in resource constrained settings
based on all the evidence presented in this thesis (chapter 9)
Descriptionofstudysites
SenegalThe study described in chapter 5 took place in a private mission dispensary based in Mlomp, a village
of approximately 8,000 inhabitants in the District of Oussouye, Casamance, south‐western Senegal.
The dispensary offers outpatient, antenatal clinic (ANC running once a week), and maternity (since
1968) services. The dispensary is well attended, offers high quality services and is equipped to
perform simple laboratory tests including microscopy evaluations of malaria slides. Dispensary
Chapter 1
16
registers for antenatal care, delivery, child welfare clinics and a general register for outpatient visits
have been meticulously kept since 1993. Nearly all pregnant women attend ANC and health facility
deliveries have increased from 50% in 1961 to 99% in 1999. [84,85] The district hospital is situated in
Oussouye (the closest town) about 10km away with limited public transport option and the regional
hospital is in Ziguinchor (about 50km away).
Malaria occurs year‐round and peaks during the rainy season (July to December) in this area. A
recent study showed that malaria transmission intensity in southern Senegal has been decreasing
significantly in the past 15 years.[86] The area is rural, there is no running water or electricity and
rice cultivation is the main economic activity. Mlomp has been under yearly demographic
surveillance by the French National Institute of Demographic Studies (INED) since 1985.[87] Several
research studies on malaria and malaria chemotherapy have been conducted in the study
area.[88,89,90,91]
This setting, with a relatively enclosed population and comprehensive records on malaria episodes,
treatment and pregnancy outcomes, provided a good opportunity to pilot utilisation of routine
healthcare data for monitoring antimalarial safety in pregnancy.
Figure 2. Study site in south‐western Senegal (study presented in chapter 5).
KenyaStudies described in chapters 6 to 8 were conducted within the Health and Demographic
Surveillance System (HDSS) in western Kenya under a long‐standing collaboration between the
Kenya Medical Research Institute (KEMRI) and the US‐based Centers for Disease Control and
Prevention (CDC).[92,93] The HDSS operates a quarterly survey covering an area of about 700 km2
and 225,000 people living in 385 villages. Data on pregnancies, births, deaths, cause of deaths via
verbal autopsies and migrations are collected.[94] This is an integrated field site designed to manage
the longitudinal follow up of residential units, households and individuals. The field operations also
involve surveillance of paediatric out‐patient visits in peripheral health facilities, and monitoring of
paediatric in‐patient visits at two District Hospitals. In addition, household socioeconomic and
Introduction
17
educational status surveys are conducted annually to complement the morbidity and demographic
data. Extensive laboratory facilities have been established to support diagnostic work in parasitic
and bacterial diseases as well as HIV; basic immunology and molecular biology research in these
areas is also conducted. The centre has conducted a number of large studies and trials (e.g., the
large community‐based, group‐randomised, controlled trial of permethrin‐treated bed nets (ITNs)
carried out between 1996 ‐1999, the Phase 3 trial of the RTS,S malaria vaccine and a Phase 2B trial of
a tuberculosis vaccine, among others).
Malaria transmission is perennial and holo‐endemic, although transmission has been greatly reduced
following provision of free ITNs.[95,96] The prevalence of malaria among individuals over 15 years of
age ranged between 10–20% in the period 2006 to 2008. [KEMRI/CDC, unpublished observations]
There is a high rate of HIV infection (in 2008: 15.4% overall: 20.5% among females and 10.2% among
males while the National HIV prevalence is around 7%).[97] The prevalence of TB in individuals over
15 years of age was 600/100,000 and geohelminth prevalence in pregnant women was recorded to
be as high as 76.2%.[98,99] Consequently, the area has mortality figures that reflect this burden of
infectious diseases ‐ infant mortality rate of 111 per 1,000 live births and a life expectancy at birth of
45 years in 2008.[93] The maternal mortality ratio is 669 per 100,000 live births which is higher than
the national estimate of 488 per 100,000 live births reported by the Kenya Demographic and Health
Survey in 2008/2009 for approximately the same time period.[100]
Nyanza Province
Figure 3. Study sites in western Kenya part of the KEMRI/CDC Health and Demographic Surveillance sites (studies presented in Chapters 6‐8).
References1. WHO (2002) The importance of pharmacovigilance. Safety monitoring of medicinal products. Geneva. 2. Lenz W (1988) A short history of thalidomide embryopathy. Teratology 38: 203‐215. 3. McBride W (1961) Thalidomide and congenital abnormalities. Lancet 278: 1358. 4. Zwingenberger K, Wnendt S (1995) Immunomodulation by thalidomide: systematic review of the literature
and of unpublished observations. J Inflamm 46: 177‐211. 5. Eriksson T (2014) Thalidomide in Mann’s Pharmacovigilance. In: Moore EBAaN, editor. Mann’s
Pharmacovigilance: John Wiley & Sons.
Chapter 1
18
6. WHO (2004) Pharmacovigilance: ensuring the safe use of medicines. Geneva. 7. Eypasch E, Lefering R, Kum CK, Troidl H (1995) Probability of adverse events that have not yet occurred: a
statistical reminder. BMJ 311: 619‐620. 8. Edwards IR (1999) Spontaneous reporting‐‐of what? Clinical concerns about drugs. Br J Clin Pharmacol 48:
138‐141. 9. Mann RD, Wilton LV, Pearce GL, Mackay FJ, Dunn NR (1997) Prescription‐event monitoring (PEM) in 1996‐a
method of non‐interventional observational cohort pharmacovigilance. Pharmacoepidemiol Drug Saf 6 Suppl 3: S5‐11.
10. Mann R, Andrews E (2007) Pharmacovigilance. Chichester: John Wiley & Sons Ltd. 11. Strom BL (2000) Pharmacoepidemiology. Sussex: John Wiley & Sons, Ltd. 12. Isah AO, Pal SN, Olsson S, Dodoo A, Bencheikh RS (2012) Specific features of medicines safety and
pharmacovigilance in Africa. Ther Adv Drug Saf 3: 25‐34. 13. UMC the Uppsala Monitoring Centre. 14. UMC (2014) Nine Million. Uppsala. 15. Pirmohamed M, Atuah KN, Dodoo AN, Winstanley P (2007) Pharmacovigilance in developing countries.
BMJ 335: 462. 16. Talisuna AO, Staedke SG, D'Alessandro U (2006) Pharmacovigilance of antimalarial treatment in Africa: is it
possible? Malar J 5: 50. 17. Chandramohan D, Shibuya K, Setel P, Cairncross S, Lopez AD, et al. (2008) Should data from demographic
surveillance systems be made more widely available to researchers? PLoS Med 5: e57. 18. INDEPTH The INDEPTH Effectiveness and Safety Studies of Anti‐malarial Drugs in Africa (INESS). 19. ACT consortium Collating drug safety data from ACT Consortium studies. 20. MiPc Malaria in Pregnancy Consortium Pharmacogivilance Activities. 21. Warkary J (1974) Problems in applying teratologic observations in animals to man. Paediatrics 53: 820. 22. Mitchell A (1994) Special considerations in studies of drug‐induced birth defects. In: Strom B, editor.
Pharmacoepidemiology. Chichester: Wiley. 23. White NJ, McGready RM, Nosten FH (2008) New medicines for tropical diseases in pregnancy: catch‐22.
PLoS Med 5: e133. 24. Little BB (1999) Pharmacokinetics during pregnancy: evidence‐based maternal dose formulation. Obstet
Gynecol 93: 858‐868. 25. Kallen BA (2005) Methodological issues in the epidemiological study of the teratogenicity of drugs.
Congenit Anom (Kyoto) 45: 44‐51. 26. Khoury MJ, Holtzman NA (1987) On the ability of birth defects monitoring to detect new teratogens. Am J
Epidemiol 126: 136‐143. 27. Mitchell AA (2003) Systematic identification of drugs that cause birth defects‐‐a new opportunity. N Engl J
Med 349: 2556‐2559. 28. Moore, Persaud (1998) The Developing Human: Clinically Oriented Embryology. sixth ed. Philadelphia: W.B.
Saunders Company. pp. 548. 29. Tilson HH, Doi PA, Covington DL, Parker A, Shields K, et al. (2007) The antiretrovirals in pregnancy registry:
a fifteenth anniversary celebration. Obstet Gynecol Surv 62: 137‐148. 30. Holmes LB, Wyszynski DF, Lieberman E (2004) The AED (antiepileptic drug) pregnancy registry: a 6‐year
experience. Arch Neurol 61: 673‐678. 31. CDER, CBER (2002) Guidance for Industry: Establishing Pregnancy Exposure Registries. Rockville: FDA. 32. European Medicines Agency (2005) Guideline on the exposure to medicinal products during pregnancy:
need for post‐authorisation data. 33. Rosa F (1999) Databases in the assessment of the effects of drugs during pregnancy. J Allergy Clin Immunol
103: S360‐361. 34. Charlton RA, Cunnington MC, de Vries CS, Weil JG (2008) Data resources for investigating drug exposure
during pregnancy and associated outcomes: the General Practice Research Database (GPRD) as an alternative to pregnancy registries. Drug Saf 31: 39‐51.
35. Pastuszak A, Pinelli M, Koren G (1993) Pregnancy outcome following fetal exposure to tiaprofenic acid in the first trimester. Am J Perinatol 10: 354‐357.
36. Louik C, Kerr S, Mitchell AA (2014) First‐trimester exposure to bupropion and risk of cardiac malformations. Pharmacoepidemiol Drug Saf.
37. WHO (2013) World Malaria Report: 2013. Geneva: World Health Organisation. 38. Steketee RW, Nahlen BL, Parise ME, Menendez C (2001) The burden of malaria in pregnancy in malaria‐
endemic areas. Am J Trop Med Hyg 64: 28‐35.
Introduction
19
39. Desai M, ter Kuile FO, Nosten F, McGready R, Asamoa K, et al. (2007) Epidemiology and burden of malaria in pregnancy. Lancet Infect Dis 7: 93‐104.
40. Guyatt HL, Snow RW (2001) The epidemiology and burden of Plasmodium falciparum‐related anemia among pregnant women in sub‐Saharan Africa. Am J Trop Med Hyg 64: 36‐44.
41. Brabin B, Verhoeff F (2002) The contribution of malaria. In: Maclean AB, Nielson J, editors. Maternal morbidity and mortality. London: Royal College of Obstetricians and Gynaecologists. pp. 65‐78.
42. Guyatt HL, Snow RW (2001) Malaria in pregnancy as an indirect cause of infant mortality in sub‐Saharan Africa. Trans R Soc Trop Med Hyg 95: 569‐576.
43. Walker PG, Ter Kuile FO, Garske T, Menendez C, Ghani AC (2014) Estimated risk of placental infection and low birthweight attributable to Plasmodium falciparum malaria in Africa in 2010: a modelling study. Lancet Glob Health 2: e460‐e467.
44. WHO (2004) A strategic framework for malaria prevention and control during pregnancy in the Africa region. . Brazzaville.
45. WHO (2012) Updated WHO Policy Recommendation (October 2012): Intermittent Preventive Treatment of malaria in pregnancy using Sulfadoxine‐Pyrimethamine (IPTp‐SP). Geneva: World Health Organization.
46. van Eijk AM, Hill J, Alegana VA, Kirui V, Gething PW, et al. (2011) Coverage of malaria protection in pregnant women in sub‐Saharan Africa: a synthesis and analysis of national survey data. Lancet Infect Dis 11: 190‐207.
47. WHO (2001) Antimalarial drug combination therapy: Report of a Technical Consultation. Geneva: World Health Organization.
48. Mutabingwa TK (2005) Artemisinin‐based combination therapies (ACTs): best hope for malaria treatment but inaccessible to the needy! Acta Trop 95: 305‐315.
49. WHO (2010) Guidelines for the treatment of malaria. Second Edition. Geneva. 50. Dondorp A, Nosten F, Stepniewska K, Day N, White N (2005) Artesunate versus quinine for treatment of
severe falciparum malaria: a randomised trial. Lancet 366: 717‐725. 51. Sinclair D, Donegan S, Isba R, Lalloo DG (2012) Artesunate versus quinine for treating severe malaria.
Cochrane Database Syst Rev 6: CD005967. 52. McGready R, Lee SJ, Wiladphaingern J, Ashley EA, Rijken MJ, et al. (2012) Adverse effects of falciparum and
vivax malaria and the safety of antimalarial treatment in early pregnancy: a population‐based study. Lancet Infect Dis 12: 388‐396.
53. Howes RE, Piel FB, Patil AP, Nyangiri OA, Gething PW, et al. (2012) G6PD deficiency prevalence and estimates of affected populations in malaria endemic countries: a geostatistical model‐based map. PLoS Med 9: e1001339.
54. Luzzatto L (2010) The rise and fall of the antimalarial Lapdap: a lesson in pharmacogenetics. Lancet 376: 739‐741.
55. Savini E, Moulin M, Herrou M (1971) Experimental study of the effects of quinine on the rat, rabbit and dog fetus. Therapie 26: 563‐574.
56. Colley J, Edwards J, Heywood R, Purser D (1989) Toxicity studies with quinine hydrochloride. Toxicology 54: 219‐226.
57. Lapointe G, Nosal G (1979) Saccharin‐ or quinine‐induced changes in the rat pups following prolonged ingestion by the dam. Biol Neonate 36: 273‐276.
58. Maxwell J (1908) The use of quinine during pregnancy, labour, and the puerperium. Transaction of the Royal Society of Tropical Medicine and Hygiene 1: 229‐235.
59. Mosha D, Mazuguni F, Mrema S, Sevene E, Abdulla S, et al. (2014) Safety of artemether‐lumefantrine exposure in first trimester of pregnancy: an observational cohort. Malar J 13: 197.
60. Nosten F, McGready R, d'Alessandro U, Bonell A, Verhoeff F, et al. (2006) Antimalarial drugs in pregnancy: a review. Curr Drug Saf 1: 1‐15.
61. Ward SA, Sevene EJ, Hastings IM, Nosten F, McGready R (2007) Antimalarial drugs and pregnancy: safety, pharmacokinetics, and pharmacovigilance. Lancet Infect Dis 7: 136‐144.
62. CDC (2011) Update: New Recommendations for Mefloquine Use in Pregnancy. Atlanta: Centres for Disease Control and Prevention.
63. Gonzalez R, Hellgren U, Greenwood B, Menendez C (2014) Mefloquine safety and tolerability in pregnancy: a systematic literature review. Malar J 13: 75.
64. WHO (2013) WHO Evidence Review Group on Intermittent Preventive Treatment (IPT) of malaria in pregnancy Geneva
65. Hernandez‐Diaz S, Werler MM, Walker AM, Mitchell AA (2000) Folic acid antagonists during pregnancy and the risk of birth defects. N Engl J Med 343: 1608‐1614.
Chapter 1
20
66. Kato T, Kitagwa S (1973) Production of congenital abnormalities in fetuses of rats and mice with various sulfonamides. Congenital Anomalies 13: 7‐13.
67. Tagbor H, Bruce J, Browne E, Randal A, Greenwood B, et al. (2006) Efficacy, safety, and tolerability of amodiaquine plus sulphadoxine‐pyrimethamine used alone or in combination for malaria treatment in pregnancy: a randomised trial. Lancet 368: 1349‐1356.
68. Tagbor HK, Chandramohan D, Greenwood B (2007) The safety of amodiaquine use in pregnant women. Expert Opin Drug Saf 6: 631‐635.
69. Longo M, Pace S, Messina M, Ferraris L, Brughera M, et al. (2012) Piperaquine phosphate: reproduction studies. Reprod Toxicol 34: 584‐597.
70. Batty KT, Moore BR, Stirling V, Ilett KF, Page‐Sharp M, et al. (2010) Investigation of reproductive toxicity of piperaquine in mice. Reprod Toxicol 29: 206‐213.
71. Poespoprodjo JR, Fobia W, Kenangalem E, Lampah DA, Sugiarto P, et al. (2014) Dihydroartemisinin‐piperaquine treatment of multidrug resistant falciparum and vivax malaria in pregnancy. PLoS One 9: e84976.
72. Rijken MJ, McGready R, Boel ME, Barends M, Proux S, et al. (2008) Dihydroartemisinin‐piperaquine rescue treatment of multidrug‐resistant Plasmodium falciparum malaria in pregnancy: a preliminary report. Am J Trop Med Hyg 78: 543‐545.
73. Tarning J, Rijken MJ, McGready R, Phyo AP, Hanpithakpong W, et al. (2012) Population pharmacokinetics of dihydroartemisinin and piperaquine in pregnant and nonpregnant women with uncomplicated malaria. Antimicrob Agents Chemother 56: 1997‐2007.
74. D'Alessandro U, Van Geertruyden J, T. M, Halidou T (2009) Safe and efficacious artemisinin‐based combination treatments for African pregnant women with malaria (PREGACT). ClinicalTrialsgov: Institute of Tropical Medicine, Belgium.
75. Desai M, Katana A, Gutman J (2012) Efficacy of intermittent screening and treatment or intermittent preventive treatment (IPT) with dihydroartemisinin‐piperaquine versus IPT with sulfadoxine‐pyrimethamine for the control of malaria in pregnancy in Kenya (STOP MiP Kenya). ClinicalTrialsgov: Kenya Medical Research Institute.
76. Nosten F, Cheah PY, McGready R (2010) Randomised Trial of 3 Artemisinin Combination Therapy for Malaria in Pregnancy (DMA). Registered study: University of Oxford.
77. Osarfo J (2010) Efficacy, safety and tolerability of dihydroartemisinin‐piperaquine for treatment of uncomplicated malaria in pregnancy in Ghana. ClinicalTrialsgov: Kwame Nkrumah University of Science and Technology.
78. Ter Kuile F (2013) STOPMiP: Intermittent screening and treatment or intermittent preventive therapy for the control of malaria in pregnancy in Indonesia. MiP Library: Liverpool School of Tropical Medicine.
79. Clark RL (2009) Embryotoxicity of the artemisinin antimalarials and potential consequences for use in women in the first trimester. Reprod Toxicol 28: 285‐296.
80. WHO, TDR (2006) Assessment of the safety of artemisinin compounds in pregnancy: report of two joint informal consultations convened in 2006. Geneva: WHO. WHO/GMP/TDR/Artemisinin/07.1. WHO/GMP/TDR/Artemisinin/07.1.
81. Finaurini S, Ronzoni L, Colancecco A, Cattaneo A, Cappellini MD, et al. (2010) Selective toxicity of dihydroartemisinin on human CD34+ erythroid cell differentiation. Toxicology 276: 128‐134.
82. Dellicour S, Hall S, Chandramohan D, Greenwood B (2007) The safety of artemisinins during pregnancy: a pressing question. Malar J 6: 15.
83. Nosten F, McGready R, Alessandro U, Bonell A, Verhoeff F, et al. (2006) Antimalarial Drugs in Pregnancy: A Review. Current Drug Safety 1: 1‐15.
84. Enel C, Pison G, Lefebvre M (1993) De l'accouchement traditionnel à l'accouchement moderne au Sénégal Cahiers Santé 3: 441‐446.
85. Pison G, Kodio B, Guyavarch E, Etard J‐F (2000) La mortalité maternelle en milieu rural au Sénégal. Population 55: 1003‐1018.
86. Brasseur P, Badiane M, Cisse M, Agnamey P, Vaillant MT, et al. (2011) Changing patterns of malaria during 1996‐2010 in an area of moderate transmission in southern Senegal. Malar J 10: 203.
87. INDEPTH (2011) Mlomp DSS. 88. Agnamey P, Brasseur P, Cisse M, Gaye O, Dumoulin J, et al. (2005) Economic evaluation of a policy change
from single‐agent treatment for suspected malaria to artesunate‐amodiaquine for microscopically confirmed uncomplicated falciparum malaria in the Oussouye District of south‐western Senegal. Trop Med Int Health 10: 926‐933.
Introduction
21
89. Agnamey P, Brasseur P, de Pecoulas PE, Vaillant M, Olliaro P (2006) Plasmodium falciparum in vitro susceptibility to antimalarial drugs in Casamance (southwestern Senegal) during the first 5 years of routine use of artesunate‐amodiaquine. Antimicrob Agents Chemother 50: 1531‐1534.
90. Brasseur P, Agnamey P, Gaye O, Vaillant M, Taylor WR, et al. (2007) Efficacy and safety of artesunate plus amodiaquine in routine use for the treatment of uncomplicated malaria in Casamance, southern Senegal. Malar J 6: 150.
91. Duthé G (2008) Malaria resurgence in Senegal: measuring malaria mortality in Mlomp. Population 63: 198. 92. Sankoh O, de Savigny D, Binka F (2004) Generating Empirical Population and Health Data in Resource
constrained Countries in the Developing World. INDEPTH working paper series 1. 93. Odhiambo FO, Laserson KF, Sewe M, Hamel MJ, Feikin DR, et al. (2012) Profile: the KEMRI/CDC Health and
Demographic Surveillance System‐‐Western Kenya. Int J Epidemiol 41: 977‐987. 94. Adazu K, Lindblade KA, Rosen DH, Odhiambo F, Ofware P, et al. (2005) Health and demographic
surveillance in rural western Kenya: a platform for evaluating interventions to reduce morbidity and mortality from infectious diseases. Am J Trop Med Hyg 73: 1151‐1158.
95. Ouma P, van Eijk AM, Hamel MJ, Parise M, Ayisi JG, et al. (2007) Malaria and anaemia among pregnant women at first antenatal clinic visit in Kisumu, western Kenya. Trop Med Int Health 12: 1515‐1523.
96. ter Kuile FO, Terlouw DJ, Phillips‐Howard PA, Hawley WA, Friedman JF, et al. (2003) Reduction of malaria during pregnancy by permethrin‐treated bed nets in an area of intense perennial malaria transmission in western Kenya. Am J Trop Med Hyg 68: 50‐60.
97. Amornkul PN, Vandenhoudt H, Nasokho P, Odhiambo F, Mwaengo D, et al. (2009) HIV prevalence and associated risk factors among individuals aged 13‐34 years in Rural Western Kenya. PLoS One 4: e6470.
98. van Eijk AM, Lindblade KA, Odhiambo F, Peterson E, Rosen DH, et al. (2009) Geohelminth Infections among pregnant women in rural western Kenya; a cross‐sectional study. PLoS Negl Trop Dis 3: e370.
99. van't Hoog AH, Laserson KF, Githui WA, Meme HK, Agaya JA, et al. (2011) High prevalence of pulmonary tuberculosis and inadequate case finding in rural western Kenya. Am J Respir Crit Care Med 183: 1245‐1253.
100. Desai M, Phillips‐Howard PA, Odhiambo FO, Katana A, Ouma P, et al. (2013) An analysis of pregnancy‐related mortality in the KEMRI/CDC health and demographic surveillance system in western Kenya. PLoS One 8: e68733.
22
23
Chapter2
Thesafetyofartemisininsduringpregnancy:apressingquestion
StephanieDellicour1,SusanHall2,DanielChandramohan1and
BrianGreenwood1
1 Department of Infectious and Tropical Diseases, London School of Hygiene and Tropical
Medicine, London, United Kingdom, 2 Worldwide Epidemiology, GlaxoSmithKline, Research
Triangle Park, United States of America
Malaria Journal 2007, 6:15
Copyright© BioMed Central Ltd
Chapter 2
24
AbstractBackground: An increasing number of countries in sub‐Saharan Africa are changing to
artemisinins combination therapy (ACT) as first or second line treatment for malaria. There is an
urgent need to assess the safety of these drugs in pregnant women who may be inadvertently
exposed to or actively treated with ACTs.
Objectives: To examine existing published evidence on the relationship between artemisinin
compounds and adverse pregnancy outcomes and consider the published evidence with regard to
the safety of these compounds when administered during pregnancy.
Methods: Studies on ACT use in pregnancy were identified via searches of MEDLINE, EMBASE,
Cochrane and Current Contents databases. Data on study characteristics, maternal adverse
events, pregnancy outcomes and infant follow up were extracted.
Results: Fourteen relevant studies (nine descriptive/case reports and five controlled trials) were
identified. Numbers of participants in these studies ranged from six to 461. Overall there were
reports on 945 women exposed to an artemisinin during pregnancy, 123 in the 1st trimester and
822 in 2nd or 3rd trimesters. The primary end points for these studies were drug efficacy and
parasite clearance. Secondary endpoints were birth outcomes including low birth weight, pre‐
term birth, pregnancy loss, congenital anomalies and developmental milestones. While none of
the studies found evidence for an association between the use of artemisinin compounds and
increased risk of adverse pregnancy outcomes, none were of sufficient size to detect small
differences in event rates that could be of public health importance. Heterogeneity between
studies in the artemisinin and comparator drugs used, and in definitions of adverse pregnancy
outcomes, limited any pooled analysis.
Conclusion: The limited data available suggest that artemisinins are effective and unlikely to be
cause of foetal loss or abnormalities, when used in late pregnancy. However, none of these studies
had adequate power to rule out rare serious adverse events, even in 2nd and 3rd trimesters and
there is not enough evidence to effectively assess the risk‐benefit profile of artemisinin compounds
for pregnant women particularly for 1st trimester exposure. Methodologically rigorous, larger
studies and post‐marketing pharmacovigilance are urgently required.
The safety of artemisinins during pregnancy
25
BackgroundEvery year over 50 million pregnancies occur in areas where malaria is endemic, mostly in sub‐
Saharan Africa. Malaria can have serious consequences for the mother and her baby. In regions
where malaria transmission is unstable, pregnant women are at high risk of developing severe
malaria, spontaneous abortion, stillbirth or premature delivery [1]. In high transmission regions,
infected pregnant women are often asymptomatic but parasitaemia can cause maternal anaemia and
low birth weight (LBW), a leading cause of infant morbidity and mortality. About 8–14% of LBWs and
3–8% of infant mortality in sub‐Saharan Africa are attributable to pregnancy‐associated malaria
(PAM)[2].
The WHO recommends a three‐pronged approach to control malaria during pregnancy which includes
effective case management, intermittent preventive treatment (IPT) and insecticide‐treated nets
(ITN). Until recently, WHO recommendations for case management during any trimester of pregnancy
were chloroquine (CQ), or sulphadoxine‐pyrimethamine (SP) in CQ‐resistant areas and, alternatively,
quinine in areas where neither were effective. There is limited information on the safety profile of
most antimalarials used for treatment or prevention of malaria in pregnancy. In a period of increasing
parasite resistance to inexpensive, conventional antimalarials such as chloroquine, SP and quinine,
there is a pressing need to identify drugs which have the most favourable harm‐benefit balance for
these vulnerable patients [3,4]. Artemisinins are a potentially valuable alternative as they are highly
effective, act rapidly and are well‐tolerated. In addition, they have the potential to reduce the
transmission of malaria and to slow development of resistance [5]. In 2002, after a detailed review of
published and unpublished data, a WHO expert committee concluded that artemisinins could be used
during the second or third trimesters if no suitable alternative was available [6] However, treatment
in the first trimester was not recommended unless the life of the woman was at risk because of
concerns raised by animal experiments which suggested that artemisinins might be teratogenic and
cause foetal resorption. Further studies have confirmed the embryotoxic effects of artemisinin and its
derivatives in animals, including primates, with risk being confined to a defined period of gestation
[7,8]. However it remains unknown how these findings translate to man. Because of these safety
'signals' from animal models, there is an urgent need to establish the safety profile of this class of
drugs in pregnancy. By June 2006, 37 countries in Africa had adopted ACTs as the first or second line
treatment policy [9] and consequently pregnant women are likely to be increasingly exposed to ACTs;
many women will be exposed in the first trimester before they are aware of their pregnancy. This
review examines all published and ongoing studies reporting exposure to artemisinins during
pregnancy and discusses their safety during pregnancy.
MethodsElectronic searches were made using MEDLINE/PubMed, EMBASE (1980 – July 2006) databases and
through the Cochrane Central Register of Controlled Trials (CENTRAL). Reference lists from identified
articles and conference abstracts, including those of the Multilateral Initiative on Malaria (MIM)
conference, 2005, and the American Society of Tropical Medicine and Hygiene (ASTMH) annual
meeting, 2005, were reviewed. The National Institute of Health's list of clinical trials was searched.
Studies on exposure of pregnant women with Plasmodium falciparum malaria to any artemisinin
derivative in, irrespective of whether the participants were symptomatic, have been included in this
review. The following data were extracted from reports: setting, year, study design, sample size,
characteristics of study participants, drug regimen, and outcomes of interest. Outcomes of interests
Chapter 2
26
were: (1) maternal serious adverse events (fatal, life‐ threatening or requiring hospitalization); (2)
maternal non‐serious adverse events (3) adverse pregnancy outcomes (miscarriage, stillbirth, preterm
delivery, low birth weight, neonatal death, congenital abnormality, developmental delay).
The internal and external validity of each study was assessed based on reporting comprehensiveness,
the representativness of the study population, how bias and confounding factors were accounted for,
and the power of the study.
DescriptionofstudiesSixty‐nine studies of artemisinin‐related drugs were identified. However, 55 reports were excluded
because they were review papers, animal studies, or because study participants were not pregnant
women. Among the fourteen studies that had reported maternal and foetal outcomes of treatment
with an artemisinin derivative in pregnancy, five were randomized trials (RCT) and nine were
descriptive non‐randomized studies. Overall these represent 1,121 women who had been exposed to
an artemisinin compound during pregnancy (242 in RCTs and 879 in observational studies). Taking
into account inclusion of women in more than one of the studies reported from the Thai‐Burmese
border, reports were found of 945 pregnancies exposed to an artemisinin compound (123 in the 1st
trimester and 822 in 2nd or 3rd trimesters). The design, characteristics of the study population, type
of intervention and outcomes of each study included in the review are shown in Table 1.
RandomizedcontrolledtrialsFive RCTs of artemisinin in pregnant patients have been reported [10‐14]. A randomized trial in
Nigeria studied 45 pregnant women treated with artemether alone or in combination with
mefloquine, but provided only limited information regarding the safety of artemether because of
inconsistent infant follow up [14]. In Thailand, an effectiveness and safety trial compared artesunate
plus mefloquine with quinine in 57 pregnant women. This study excluded women who had taken
antimalarials within the previous 28 days and those who had over 4% parasitized red cells. The study
population comprised predominantly milder cases. Specific neurological evaluation of mothers was
performed weekly using established tests and infants were assessed for physical and neurological
development at 12 or 24 months [10]. Three of the comparative trials were conducted in antenatal
clinics (ANCs) on the Thai Burmese Border. They compared different drug regimens of artemisinin or
its derivatives to quinine in second or third trimester pregnancies. In 2000, a report of 108 pregnant
women randomized to receive quinine or artesunate plus mefloquine was published [12]. This was
followed in 2001 by a report of a randomized trial of artesunate versus quinine plus clindamycin in
129 pregnant women [13]. More recently, the results of a trial of 81 pregnant women randomized to
receive quinine or artesunate‐atovaquone‐proguanil has been published [11]. Women who presented
with a first malaria episode were recruited excluding those with complicated or severe malaria.
Detailed information on outcome was available. A high level of efficacy of the artemisinin derivatives
was reported in each study and the drugs were well tolerated.
DescriptivestudiesThe first descriptive report of the use of artemisinin derivatives during pregnancy came from China.
Wang et al reported the outcomes of six pregnancies exposed to either artemisinin or artemether.
Children of these women were examined 5 to 10 years later for congenital malformation, physical and
neurodevelopmental evaluation. No adverse events were reported for the mother or the infants [15].
The safety of artemisinins during pregnancy
27
McGready et al reported the outcome of case‐series pregnancies exposed to artemisinins on the Thai‐
Burmese border in three succeeding publications [16‐18]. The latest publication of 2001 encompasses
all cases seen between 1992–2000 (461 women with 539 artemisinin‐based treatments) including
women described in previous publications [12,13]. Overall, there was 11% loss to follow up. A
subgroup of 44 women was exposed inadvertently during their first trimester. In addition, two
publications report on the treatment of 27 pregnant patients with atovaquone‐proguanil plus
artesunate [19,20].
The second publication [20], focuses on the pharmacokinetics properties of the drugs in 24 of these
women. This group of studies provides the largest and most detailed record on the use of
artemisinins in pregnancy.
During a mass drug administration campaign in the Gambia, 287 pregnant women were accidentally
exposed to treatment with artesunate plus SP, including 77 who were exposed during the first
trimester of pregnancy [21]. Both active and passive surveillance were used to maximize
ascertainment of pregnancy outcomes. Reliable case ascertainment and outcome measurement was
available for births notified within seven days of delivery, as these newborns were thoroughly
examined by a paediatrician.
In Sudan, Adam et al have reported on two studies [22,23]. The first consisted of the follow‐up of 28
symptomatic pregnant women who received intramuscular artemether after previous treatment
failure with CQ or quinine. Women were assessed for neurological defects. There was one first
trimester exposure which resulted in a normal newborn. The second study described 32
symptomatic pregnant women treated with artesunate plus SP. In both studies, women and their
infants were followed up systematically (at birth and one year of age).
Ashley et al reported accidental exposure to artemisinin compounds during pregnancy in three
clinical trials for the treatment of uncomplicated P. falciparum malaria [24]: they documented the
inadvertent exposures to dihydroartemisinin‐piperaquine of one woman at 11 weeks gestation and
another at 18 weeks gestation [25,26]. In Uganda, four pregnant women were accidentally exposed
in their 1st trimester to artemether‐lumefantrine during a trial [27]. All these women delivered
normal babies.
ResultsMaternal adverse events and pregnancy outcomes in the artemisinin group and comparison groups
(where applicable) are summarized in Tables 2 and 3 respectively.
MaternaladverseeventsOverall, artemisinin derivatives were well tolerated; none of the studies reported serious adverse
event (SAE) attributed to the use of artemisinin. There were seven maternal deaths reported in total:
three women exposed to an artemisinin (two on the Thai‐Burmese border and one during the mass
drug administration campaign in the Gambia), three with unknown exposure (two reports from RCTs
on the Thai‐Burmese border and one in the Gambia) and one not exposed from the Gambia study.
Cause of death was known for only five women [severe malaria and anaemia (1), liver abscess (1),
post‐partum haemorrhage (3)]. Other maternal adverse events were described clearly only in the
studies conducted at the Thai‐Burmese border.
Chapter 2
28
Table 1: Description of the studies included in the review on the use of artemisinin in pregnancy.
Study site, year Reference Study Population Antimalarial treatment Primary Outcomes
Randomized trials*
Nigeria, 1994–1997 [14] Pregnant women infected with P. falciparum malaria after treatment failure§
1) Artemether IM + mefloquine n = 22 2) Artemether IM n = 23 Artemether overall dosage: 3.2–9.6 mg/kg
Parasitaemia at day 28; AEs Foetal outcomes; infant neurodevelopment
Thailand, 1995–1998 [10] Pregnant women infected with P. falciparum malaria§
1)Quinine n = 29 2)Artesunate+ mefloquine n = 28 Artesunate overall dose: 12 mg/kg
Time to parasite & fever clearance; anaemia; AEs; neurological examination Perinatal assessment; birth outcomes; infant physical and neurological assessment
TBB, 1995–1997 [12] Pregnant women with uncomplicated P. falciparum malaria
1)Quinine n = 42 2)Artesunate+mefloquine n = 66 Artesunate overall dose: 12 mg/kg
Parasite clearance; anaemia; AEs; neurological deficit Birth outcomes; infant developmental milestones
TBB, 1997–2000 [13] Pregnant women with uncomplicated P. falciparum malaria
1)Quinine+clindamycin n = 65 2) Artesunate n = 64 Artesunate overall dose: 12 mg/kg
Treatment failure at day 42 or before delivery; anaemia; AEs Birth outcomes; infant developmental milestones
TBB, 2001–2003 [11] Pregnant women with uncomplicated P. falciparum or mixed malaria
1)Quinine n = 42 2) Artesunate atovaquone-proguanil n = 39 Artesunate overall dose: 12 mg/kg
Fever, parasite clearance; treatment failure at day 63; anaemia; AEs Birth outcomes; infant developmental milestones
Case series
China, 1976–1980 [15] Pregnant women infected with P. falciparum or P vivax malaria§
Trimesters (n): 3rd (1), 2nd (5)
Artemisinin IM or Artemether Overall artemisinins dose: 500–900 mg
Therapeutic effects and adverse reactions for mother and child
TBB, 1992–1996 [17] Pregnant women with uncomplicated multi- drug resistant P. falciparum or mixed malarial infection Trimesters (n): 3rd & 2nd (74), 1st (16)
Artesunate alone or with mefloquine; or Artemether+mefloquine Overall artemisinins dose: 12–16 mg/kg
Treatment failure at day 42; anaemia; ADR; neurological deficit Birth outcome; infant neurological development
TBB, 1991–1996 [18] Pregnant women with uncomplicated multi- drug resistant P. falciparum malarial infection after treatment failure Trimesters (n): 3rd & 2nd (74), 1st (16)
Artesunate alone or with mefloquine; or Artemether+mefloquine Overall artemisinins dose: 12–16 mg/kg
Treatment failure at day 42; anaemia; ADR; neurological deficit Birth outcome; infant neurological development
TBB, 1992–2000 [16] Pregnant women with uncomplicated multi- drug resistant P. falciparum or mixed malarial infection Trimesters (n): 3rd (211), 2nd (201), 1st (42)
Artesunate alone or with mefloquine; clindamycin atovaquone-proguanil;or coartemether, artemether IM Overall artemisinins dose: 12–16 mg/kg
Treatment failure at day 42; anaemia; ADR; neurological deficit Birth outcome; infant developmental milestones
TBB 1999–2001 [19] Pregnant women with multi-drug resistant P. falciparum or mixed malarial infection Trimesters (n): 3rd & 2nd (24), 1st (3)
Artesunate+atovaquone-proguanil Artesunate overall dose: 12–14 mg/kg
Treatment failure at day 42; parasite clearance; AEs Birth outcomes
TBB 2000–2001 [20] Pharmacokinetic study: pregnant women with multi-drug resistant P. falciparum or mixed malarial infection Trimesters (n): 3rd (13) & 2nd (11)
Artesunate+atovaquone-proguanil Artesunate overall dose: 12 mg/kg
Pharmacokinetic parameters; AEs (including ECG); parasite clearance Birth outcomes
Gambia, 1999 [41] Pregnant women participating in mass prevention campaign Trimesters (n): 1st (77)
Artesunate + SP Artesunate overall dose: 200 mg
MDA 1° aim: malaria transmission Pregnancy outcomes
Sudan, 1997–2001 [22] Pregnant women infected with P. falciparum malaria after treatment failure§
Trimesters (n): 3rd (15), 2nd (12), 1st (1)
Artemether IM Artemether overall dose: 480 mg
Treatment failure at day 28; anaemia; neurological deficit Birth outcomes
Sudan, 2004–2005 [23] Pregnant women infected with uncomplicated P. falciparum malaria§
Trimesters (n): 3rd (22), 2nd (10)
Artesunate + SP Artesunate overall dose: 200 mg
Treatment failure at day 28; anaemia Birth outcomes
* 2nd and 3rd trimesters only §100% symptomatic/febrile Abbreviation: ADR: adverse drug reaction; AE: adverse event; ECG: electrocardiograph; IM: intramuscular injection; NR: not reported; Q: quinine; A: artemisinin derivative; A+M: artesunate+mefloquine; TBB: Thai-Burmese border
In these studies, adverse events that could potentially be associated with the study drug were defined
as the occurrence of symptoms after treatment not present at baseline and which occurred before a
recrudescence of parasitaemia or a second infection. Data from Thailand and the Thai Burmese
border showed that tinnitus and dizziness were significantly less common in the artemisinin
treatment group (range 9–64%) compared to the quinine arm (45–79%). Although nausea and
vomiting were also significantly less frequent in the artesunate‐mefloquine treatment arm in the
Thailand study, no significant difference was reported in the Thai‐Burmese border trials. Other
adverse events were reported with a similar frequency in each treatment group. The most frequent
complaints according to these studies were dizziness (range 42–45% for the artemisinin group versus
49– 87% for the quinine arm); nausea (range 22–57% and 8– 93% respectively); headache (range 21–
The safety of artemisinins during pregnancy
29
30% and 30–50% respectively); anorexia (range 7–33% and 0–47% respectively) and muscle and joint
pain (range 12–31% and 12– 27% respectively).
All of these symptoms were also reported on presentation (dizziness 57–61%; headache 72–75%;
anorexia 46–62%, nausea 30%–40% and muscle/joint pain 59–68%) and so the apparent adverse
events may have been due to the underlying malarial episode. Specific neurological evaluation of
mothers was performed in four studies [10,12,17,22]. This included Romberg's test, assessment of
heel‐toe ataxia, fine finger dexterity, auditory acuity and an assessment for nystagmus. There was no
report of a woman with an adverse neurological event associated with drug administration. The
studies were too heterogeneous in terms of severity of disease, drug treatment and outcome
measurements, and numbers to few to pool data for a meta‐analysis.
Table 2: Safety outcomes: maternal adverse events and pregnancy outcomes in women exposed to artemisinin compounds. Randomized trials
Study site, year Ref
Antimalarial treatment N (% follow up) Outcomes of interests
Maternal Safety¥ Pregnancy outcomes
Nigeria, 1994–1997 [14] 1) Artemether IM + mefloquine n = 22 2) Artemether IM n = 23
45 (100%) Minimal, only A+M: abdominal discomfort (9%) and dizziness (9%)
Neonatal jaundice (n = 2 A & n = 1 A+M) 6 (13%) followed up to 1 year
Thailand, 1995–1998 [10] 1) Quinine n = 29 2) Artesunate+ mefloquine n = 28
60 (95%) Neurological exam: all normal. Nausea (16%), vomiting (12%), vertigo (12%), tinnitus (18%), and hypoglycaemia (3%) more frequent in Q (p < 0.05). Other no difference: Palpitation (6%), blurring vision (11%).
Neonatal jaundice (n = 5 Q & n = 1 A+M) 46 (81%) followed up to 1 year
TBB, 1995–1997 [12] 1) Quinine n = 42 2) Artesunate+ mefloquine n = 66
108 (85%) Neurological exam 1 maternal death* Tinnitus (15%) and dizziness (42%) more frequent in Q (p < 0.05). Headache (21%), nausea (45%), abdominal pain (28%), vertigo (12%), muscle/joint pain (32%), and anorexia (35%) no difference with Q (p > 0.05).
Abortions (n = 2 A+M) 46 (49%) followed up to 1 year Neonatal deaths (n = 2 A+M & n = 1 Q)
TBB, 1997–2000 [13] 1) Quinine)+ clindamycin n = 65 2) Artesunate n = 64
129 (91%) Tinnitus (9%) more frequent in Q+C (p < 0.05). Headache (30%), dizziness (41%) nausea (25%), vomiting (8%), abdominal pain (18%), rash (9%), contractions (35%), muscle/joint pain (12%), and anorexia (42%) no difference with Q+C (p > 0.05).
Stillbirths (n = 1 A & n = 1 Q+C) Congenital abnormality: midline epidermoid cyst (n = 1 Q+C) Neonatal deaths (n = 1 A & n = 2 Q+C) 72 (62%) followed up to 1 year
TBB, 2001–2003 [11] 1) Quinine n = 42 2) Artesunate atovaquone-proguanil (AAP) n = 39
81 (91%) 1 maternal death** Tinnitus (24%) more frequent in Q (p < 0.05).
Stillbirth (n = 1 not specified maternal death) Congenital abnormalities: polythelia (n = 1 AAP); cleft lip & palate (n = 1 AAP); aural atresia (n = 1 Q) Neonatal deaths (n = 1 A & n = 2 Q+C) 59 (78%) followed up to 1 year Developmentally delayed (n = 1 AAP)
TOTAL No 1st trimester, P. falciparum or mixed malaria 17% to 100% symptomatic 242 artemisinins exposures
423 (94%) 2 maternal deaths Neurological exam in 2 studies
Stillbirths (n = 1 A; n = 1 Q+C & n = 1 unknown) Congenital abnormality: (n = 2 A & n = 2 Q) Neonatal deaths (n = 4 A+M & n = 5 Q) 229 (59%) infant followed up to 1 year
¥ Prevalence of possible ADRs are only reported for the artemisinin treatment groups * cause unrelated to malaria (treatment group NR) ** caused by a ruptured liver abscess (treatment group NR) Abbreviation: ADR: adverse drug reaction; AE: adverse event; IM: intramuscular injection; NR: not reported; Q: quinine; A: artemisinin derivative; A+M: artesunate+mefloquine; TBB: Thai-Burmese border
PregnancyoutcomesNinety‐six percent of the 945 women exposed to an artemisinin in pregnancy were followed up to
delivery. Twenty (2.1%) had miscarriages, 19 (2%) stillbirths and 11 (1.2%) neonatal deaths. Six (0.6%)
congenital abnormalities were reported (one left aural atresia, one polythelia, one epidermoid cyst
Chapter 2
30
and three not described); whereas the expected rate of birth defects in developing countries is about
6% [28]. Of the 214 infants examined up to at least one year of age only one was reported to be
developmentally delayed. The mother of this infant was treated with artesunate‐atovaquone‐
proguanil and he was assessed for motor and neuro‐developmental milestones (no details provided).
Table 3: Safety outcomes: maternal adverse events and pregnancy outcomes in women exposed to artemisinin compounds: Descriptive studies.
Study site, year Ref Antimalarial treatment N (% Follow-up) Outcomes of interests
Maternal Safety Pregnancy outcomes
Overall 1st trimester expo- sures
China, 1976–1980 [15] Artemisinin IM or Artemether 6 (100%) No adverse effect 6 (100%) followed up at 5–9 year
None
TBB, 1992–2000 [16] Artesunate alone or with mefloquine; clindamycin atovaquone-proguanil; coartemether, artemether IM
461 (89%) No ADRs (pruritus) Maternal deaths (n = 2)*
Abortions 4.8% (n = 20) Stillbirth 1.8% (n = 7) Congenital abnormalities 0.8%: anencephaly (n = 1); midline epidermoid cyst (n = 1)
Abortions 23% (n = 10)
TBB 1999–2001 [19] Artesunate+atovaquone- proguanil
27 (100%) Symptoms possible ADRs: Headache (42%); muscle/joint pain (30%); abdominal pain (42%); anorexia (40%); nausea (25%); vomiting (10%); rash (15%); dizziness (70%), tinnitus (24%); contraction (15%) and sleep disturbance (8%)
Neonatal deaths (n = 1) Normal deliveries and healthy newborns
Gambia, 1999 [41] Artesunate + SP 325 (88%) Maternal deaths (n = 1)** Stillbirth (n = 11) Congenital abnormalities: umbilical hernia (n = 1); undescended testis; (n = 1) Neonatal deaths (n = 8)
Not reported
Sudan, 1997–2001 [22] Artemether IM 28 (100%) NR Neonatal deaths (n = 1) Normal delivery and healthy newborn
Sudan, 2004–2005 [23] Artesunate + SP 32 (100%) Giddiness and nausea (13%) Neurological exam.
Neonatal deaths (n = 1) None
TOTAL 123 1st trimester, P. falciparum or mixed malaria 7% to 100% symptomatic 945 artemisinin exposures
879 (90%) 3 maternal deaths Neurological exam in 2 studies
Abortions n = 20 Stillbirths (n = 18) Congenital abnormality: (n = 4) Neonatal deaths (n = 11) 65 (15%) infant followed up to 1 year
Abortions n = 10
* One maternal death was due to severe malaria and anaemia, the other died of causes unrelated to malaria. ** For 1 maternal death the exposure status could not be confirmed; verbal autopsies indicate that the deaths were due to postpartum haemorrhage. Abbreviation: ADR: adverse drug reaction; AE: adverse event; IM: intramuscular injection; NR: not reported; Q: quinine; A: artemisinin derivative; A+M: artesunate+mefloquine; TBB: Thai-Burmese border
DiscussionNone of the studies included in this review revealed an increased risk of serious maternal adverse
effects, adverse birth outcomes, or neuro‐development deficits associated with the use of an
artemisinin drug during pregnancy. However, these studies were not designed to assess safety
endpoints and, although they were sufficiently powered to answer the original study objective, the
studies were under‐powered to detect rare safety outcomes, or small differences in adverse event
rates, between the comparison groups.
The very low prevalence of congenital anomalies reported can partly be explained by the complex
nature of birth defect ascertainment; ideally a dysmorphologist should have assessed all newborns
for abnormalities that would not be detected by an untrained physician. It was also not possible to
assess the expected rate of adverse birth outcomes in any of the study settings due to the lack of
The safety of artemisinins during pregnancy
31
background population data on rates for abortions, stillbirths and congenital abnormalities. Four
studies which looked for neurological damage following artemisinin exposure revealed no indication
of neurotoxicity. However, in the light of recent debates about the potential ototoxicity of the
artemisinins, further studies conducting thorough auditory evaluation before and after treatment
may be needed [29‐31].
The studies reviewed were highly heterogeneous in terms of treatment used, outcomes assessment,
population and follow up rate. Different artemisinin drug regimens were used with varying control
treatments as a comparator and six studies did not have a control group but compared the rates of
adverse birth outcome to community rates derived from separate studies. Most of the studies used
an artemisinin derivative combined with another drug, which adds to the difficulty of teasing out
individual drug effects and restricts comparison between studies. Four studies used artesunate in
combination with mefloquine; use of the latter in pregnancy has caused concern because of an
increased risk of stillbirth in women who received this drug during pregnancy in one study [32]
although this was not found in others [33‐36]. The methods used to monitor adverse event and birth
outcomes also differed widely between the studies. Under‐estimates of adverse outcomes cannot
be ruled out, particularly for early pregnancy loss in the mass drug administration campaign in the
Gambia. There are no statistics on expected rate of early pregnancy loss, and determining the
prevalence of spontaneous abortion is difficult. In western countries, miscarriages occur in 10 to15%
of pregnancies, mostly during the first few weeks. The prevalence of miscarriages is likely to be even
higher in resource poor setting. When assessing maternal AEs, there is the methodological challenge
of differentiating between the effect of malaria and its treatment which is difficult to do.
Furthermore, the severity of malaria episode varied between the different study populations since
three included women who had already experienced a treatment failure whilst nine others included
some asymptomatic women. Furthermore, the study in the Gambia was preventive; as the majority
of the participants in this study are unlikely to have had malaria, lower rates of adverse birth
outcomes would be expected compared to those of studies of malaria case management. The
allocation of interventions could have been influenced by many factors particularly for the non‐
randomized studies, such as prognostic factors (severity or malaria attack rates, parasite resistance,
mother's age, gravidity, other drugs etc.), which could themselves influence birth outcome and
treatment response. There is, therefore, a potential for selection and detection bias. Maternal
malaria has been associated with stillbirth, abortion and LBW and these adverse end‐points are also
considered as possible indicators of an adverse event related to drug administration [1].
Furthermore, none of these studies controlled for other drug use, which could potentially influence
pregnancy outcomes.
Most of the information on artemisinin exposure during pregnancy comes from studies conducted in
Southeast Asia, at the antenatal clinics of the Shoklo Malaria Research Unit among women of the
Karen ethnic minority. The RCTs had robust design and randomly allocated treatment groups,
although the investigators were not blinded to the treatment group. Follow up of infants varied
between these three studies (49 to 80%) and only 46 infants were followed up to one year of age
and examined for developmental abnormalities in the non‐comparative studies. The women
enrolled in these studies are likely to be representative of this population group (Karen ethnic
minority) as over 90% of the women in the studied Thai‐ Burmese border region attend ANCs. These
findings cannot be extrapolated directly to sub‐Saharan Africa considering geographic differences in
parasite species, resistance pattern, transmission intensity and host immunity.
Chapter 2
32
A number of studies of artemisinin treatment in pregnancy are in progress (Table 4), which will
contribute further to knowledge on the safety of these drugs. However, except for the phase IV
study being conducted by the Centres for Disease Control in Tanzania, these studies focus on 2nd
and 3rd trimester pregnancies. The results of animal studies suggest that if there is a safety issue
related to the administration of artemisinins in pregnancy this is likely to occur very early in
pregnancy. Only 123 documented first trimester exposures have been reported and this does not
provide enough evidence to determine safety. Moreover, an "all or nothing rule" seems to apply for
exposure during the first weeks of pregnancy. Animal experiments suggest that congenital
abnormalities occur only after exposure over a narrow dose range and over a limited period of time
and that exposure usually lead to a normal pregnancy outcome or death of the foetus.
Developmental toxicity studies in the monkey confirmed findings from rodent studies; with embryo
death induced at therapeutic dose ranges [37]. The teratogenic effect is thought to involve red blood
cells production, erythropoiesis, which implies the human sensitive period would be within the first
trimester of pregnancy [38]. An effect of ACTs on early pregnancy loss will be difficult to detect,
especially in communities where artemisinins are likely to be used most frequently. More extensive
studies are needed that will be able to detect rarer outcomes and any congenital abnormalities that
might result from exposure in the first trimester of pregnancy. Ethical constraints will prohibit
randomized trials of artemisinins in the first trimester of pregnancy in most communities where
alternatives exist until more information on safety has been obtained. Thus, data will largely have to
come from observational studies.
Unfortunately, most countries in sub‐Saharan Africa do not have the infrastructure and resources for
routine pharmacovigilance and very few have a formal system for routine collection of data on
possible drug related adverse effects [39]. In industrialized countries, a variety of post‐marketing
surveillance techniques is used. Case reports and case series are the first source of information to
detect adverse events in pregnancy. Although these are useful in the generation of hypotheses of a
possible safety signal, specific pharmaco‐epidemiological studies using a cohort or case‐control
approach are required to evaluate teratogenic risk. The main factors impeding the implementation
of pharmacovigilance in poor countries include limited access to healthcare facilities, availability of
most prescription drugs from the informal market, poor labelling of medicines, counterfeit and sub‐
standard pharmaceutical products, a high level of illiteracy, poor record keeping and a shortage of
qualified healthcare professionals. Special pharmaco‐epidemiological studies will be needed to
assess the safety profile of a product's use outside the controlled environment of clinical trials.
Active surveillance systems could use the existing sentinel demographic surveillance sites (DSS),
which already undertake regular household visits to obtain more information on drug usage and
adverse effects during pregnancy. However, these studies are expensive and not sustainable in the
long term and should be restricted to newer products or following identification of new safety
concerns for an older drug. Antenatal clinics (ANC) could make a useful platform for routine
surveillance, as a high proportion of pregnant women attend an ANC at least once during pregnancy
in sub‐Saharan Africa [40]. It is necessary to develop a more pragmatic pharmacovigilance system
that can be linked to a routine health information system and ideally a mix of different approaches
should be used to assess the safety profile of individual drugs for pregnant patients.
The safety of artemisinins during pregnancy
33
ConclusionMalaria in pregnancy is a major public health issue and infected pregnant women need prompt
treatment with effective drugs. Although artemisinin derivatives and combinations have an excellent
efficacy profile, there is very limited data on the safety of artemisinin use during pregnancy
particularly in sub‐Saharan Africa. Although a few studies of the safety and efficacy of artemisinins
during pregnancy are currently underway, these will not produce data on the safety of artemisinins
during the first trimester of pregnancy. Post‐marketing pharmacovigilance of artemisinin use during
pregnancy is needed urgently as ACT is implemented in almost all countries in Africa. Innovative
pharmacovigilance tools, methods and systems are needed to monitor the safety of artemisinins and
other antimalarials.
Authors' contributionsSD carried out the literature review and wrote the first draft of the paper. SH, DC and BG contributed
to the structure and content and were involved in re‐drafting the paper.
AcknowledgementsWe thank Jenny Hill and Feiko ter Kuile, from the MiP working group, for sharing with us the
information collated for the MiP database. We are also grateful to John McArthur for sharing with us
the outline for the phase IV study in Tanzania, Rose McGready for clarifying the overlap between
publications for the studies on the Thai‐Burmese border and Steve Bowling for reviewing the
manuscript. Stephanie Dellicour is supported by GlaxoSmithKline.
Chapter 2
34
Table 4: Ongoing Studies (Information extracted from Malaria in Pregnancy (MiP)* Database [42] on 05/02/2007).
Study Setting Design & Drug Regimen Outcome Status (start date-completion date)
Safety/Efficacy prevention trial in pregnancy
Ifakara, Tanzania CDC/IHRDC-IMPACT Randomised open label, n = 1200 (400 per arm) IPTp Control: SP 2 doses Intervention: 1. SP monthly 2. SP+artesunate monthly
1°: Placental parasitaemia and AEs 2°: Maternal illness and parasitaemia at delivery, birth outcome (BW, Gestational age, foetal and infant health), childhood developmental milestones
Recruitment concluded; ongoing follow up (01/03-ongoing)
Safety/Efficacy Treatment trial in Pregnancy
ANC at Muheza Hospital, Tanzania GMP Randomised open label, target (Phase III) n = 350 2nd or 3rd trimesters Control: SP Intervention: SP+amodiaquine, Amodiaquine+artesunate, Chloroporguanil-dapsone
1°: Treatment failure at day 28; Treatment outcome (parasite/fever clearance, parasite recrudescence) 2°: Foetal viability and birth outcomes (preterm delivery, foetal death, perinatal/ neonatal mortality, neonatal abnormality); maternal AE (hypoglycaemia)
Recruitment completed (01/04–07/06)
Shoklo Malaria Research Unit (SMRU) ANC, Thailand UNICEF-UNDP-World Bank-WHO-TDR
Randomized intervention trial n = 250 2nd or 3rd trimesters Group 1: Artesunate Group 2: Co-artemether (artemether/ lumefantrine)
1°: Treatment outcome at day 42 or at delivery (parasite/fever clearance, parasite recrudescence) 2°:Gametocyte carriage; pharmacokinetic parameters; histo- pathology examination of the placenta
Currently recruiting (06/02/2004–31/12/ 2008)
Bangladesh WHO
Randomised controlled trial n = 684 Control: placebo rectal capsule Intervention: Artesunate rectal capsule
Pregnancy outcomes Currently recruiting (10/11/2003- ongoing)
Malawi; Prof Meshnick, UNC Randomised open label, n = 141 2nd or 3rd trimesters Control: SP Intervention: SP+artesunate or SP+azithromycin
1°: Parasitological failure rates; parasite clearance time; fever clearance times and incidence rate of adverse events 2°: Prevalence rate of abortions; still births; peripheral parasitaemia at delivery; placental malaria and of maternal anaemia
Recruitment completed (09/2003–10/ 2005)
Efficacy/Pharmacokinetic trial in Pregnancy
Mozambique UCT, South Africa
Non-Randomized openLabel, target n = 30 2nd or 3rd trimester pregnant HistoricalControl Intervention: SP+artesunate
1°: Pharmacokinetic parameters 2°: gametocyte carriage, maternal AE & birth outcomes
Currently recruiting (03/2006–09/2008)
Kinshasa, DRC, NIH-NICHD Dose-equivalence trial: part of investigational new drug application n = 60 2nd or 3rd trimester Control: SP Intervention: Artesunate-mefloquine combinations
Pharmacokinetic parameters Recruitment completed (07/2005–12/ 2005)
Pharmacovigilance: Post-marketing surveillance
Tanzania, CDC Pharmacovigilance surveillance system: part of a large ongoing study to look at district wide use of ACTs 1st trimester Control: SP Intervention: SP+Artesunate
Pregnancy outcome and status of child Ongoing (2005–2007)
A partnership between Novartis WHO- TDR and the Government of Zambia. [43]
Pregnancy Registry Prospective active surveillance cohort. Expected n = 1600 Control: SP Intervention: artemether-lumefantrine
Maternal and neonatal outcomes examined
Ongoing (2005)
* MiP is a consortium of experts in the field of malaria funded by the Bill and Melinda Gates Foundation to review current research and develop future research strategy for malaria in pregnancy. One of the key objectives of MiP is to create a database containing all published and unpublished research and a trial registry on malaria in pregnancy. Abbreviations: ACT: artemisinin combination therapy; AE: adverse events; ANC: antenatal care; BW: birthweight; CDC: Centers for Disease Control & Prevention; GMP: Gates Malaria Partenership; DRC: Democratic Republic of the Congo; IHRDC: Ifakara Health Research and Development Centre; IPTp: intermittent presumptive treatment for pregnancy; NIH-NICHD: National Institutes of Health-The National Institute of Child Health and Human Development; RCT: randomized controlled trials; SP: sulphadoxine-pyrmethamine; TBB: Thai-Burmese Border; UNC: University of North Carolina; UCT: University of Cape Town; WHO-TDR: World Health Organization – The Special Programme for Research and Training in Tropical Diseases
The safety of artemisinins during pregnancy
35
References1. Luxemburger C, McGready R, Kham A, Morison L, Cho T, Chongsu‐ phajaisiddhi T, White NJ, Nosten F:
Effects of malaria during pregnancy on infant mortality in an area of low malaria transmission. Am J Epidemiol 2001, 154:459‐465.
2. Steketee RW, Nahlen BL, Parise ME, Menendez C: The burden of malaria in pregnancy in malaria‐endemic areas. Am J Trop Med Hyg 2001, 64:28‐35.
3. Newman RD, Parise ME, Slutsker L, Nahlen B, Steketee RW: Safety, efficacy and determinants of effectiveness of antimalarial drugs during pregnancy: implications for prevention pro‐ grammes in Plasmodium falciparum‐endemic sub‐Saharan Africa. Trop Med Int Health 2003, 8:488‐506.
4. Nosten F, McGready R, d'Alessandro U, Bonell A, Verhoeff F, Menen‐ dez C, Mutabingwa T, Brabin B: Antimalarial Drugs in Preg‐ nancy: A Review. Current Drug Safety 2006, 1:1‐15.
5. WHO: Antimalarial drug combination therapy. In WHO/CDS/ RBM/2001 Geneva: World Health Organization; 2001.
6. WHO: Assessment of the safety of artemisinin compounds in pregnancy. Report of two informal consultations convened by WHO in 2002. In WHO/CDS/MAL/2003.1094 WHO/RBM/TDR/ Artemisinin/03.1 Geneva: WHO; 2003.
7. Clark RL, White TE, S AC, Gaunt I, Winstanley P, Ward SA: Devel‐ opmental toxicity of artesunate and an artesunate combina‐ tion in the rat and rabbit. Birth Defects Res B Dev Reprod Toxicol 2004, 71:380‐394.
8. Longo M, Zanoncelli S, Manera D, Brughera M, Colombo P, Lansen J, Mazue G, Gomes M, Taylor WR, Olliaro P: Effects of the antima‐ larial drug dihydroartemisinin (DHA) on rat embryos in vitro. Reprod Toxicol 2006, 21:83‐93.
9. Antimalarial treatment policies for P. falciparum and P. vivax by country in WHO Africa region [http://www.who.int/malaria/ amdp/amdp_afro.htm]
10. Bounyasong S: Randomized trial of artesunate and mefloquine in comparison with quinine sulfate to treat P. falciparum malaria pregnant women. J Med Assoc Thai 2001, 84:1289.
11. McGready R, Ashley EA, Moo E, Cho T, Barends M, Hutagalung R, Looareesuwan S, White NJ, Nosten F: A randomized comparison of artesunate‐atovaquone‐proguanil versus quinine in treat‐ ment for uncomplicated falciparum malaria during preg‐ nancy. J Infect Dis 2005, 192:846‐853.
12. McGready R, Brockman A, Cho T, Chq D, van Vugt M, Luxemburger C, Chongsuphajaisiddhi T, White N, Nosten F: Randomized com‐ parison of mefloquine‐artesunate combination versus qui‐ nine in treatment of multi‐drug resistant falciparum malaria in pregnancy. Trans R Soc Trop Med Hyg 2000, 94:689.
13. McGready R, Cho T, Leopold Villegas S, Brockman A, van M VI, Looareesuwan S, Whitezg NJ, Nosten F: Randomized comparison of quinine‐clindamycin versus artesunate in the treatment of falciparum malaria in pregnancy. Trans R Soc Trop Med Hyg 2001, 95:651.
14. Sowunmi A, Oduola AM, Ogundahunsi OA, Fehintola FA, Ilesanmi OA, Akinyinka OO, Arowojolu AO: Randomised trial of arte‐ mether versus artemether and mefloquine for the treat‐ ment of chloroquine/sufadoxine‐pyrimethamine‐resistant falciparum malaria during pregnancy. J Obstet Gynaecol 1998, 18:322‐327.
15. Wang TY: Follow‐up observation on the therapeutic effects and remote reactions of artemisinin (Qinghaosu) and arte‐ mether in treating malaria in pregnant woman. J Tradit Chin Med 1989, 9:28‐30.
16. McGready R, Cho T, Keo NK, Twai KL, Villegas L, Looareesuwan S, White N, Nosten F: Artemisinin antimalarials in pregnancy: a prospective treatment study of 539 episodes of multidrug‐resistant Plasmodium falciparum. Clin Infect Dis 2001, 33:2009.
17. McGready R, Cho T, Cho JJ, Simpson JA, Luxemburger C, Dubowitz L, Looareesuwan S, White NJ, Nosten F: Artemisinin derivatives in the treatment of falciparum malaria in pregnancy. Trans R Soc Trop Med Hyg 1998, 92:430‐433.
18. McGready R, Nosten F: The Thai‐Burmese border: drug studies of Plasmodium falciparum in pregnancy. Ann Trop Med Parasitol 1999, 93(Suppl 1):S19.
19. McGready R, Keo NK, Villegas L, White NJ, Looareesuwan S, Nosten F: Artesunate‐atovaquone‐proguanil rescue treatment of multidrug‐resistant Plasmodium falciparum malaria in pregnancy: a preliminary report. Trans R Soc Trop Med Hyg 2003, 97:592‐594.
20. McGready R, Stepniewska K, Ward SA, Cho T, Gilveray G, Looareesuwan S, White NJ, Nosten F: Pharmacokinetics of dihydroartemisinin following oral artesunate treatment of pregnant women with acute uncomplicated falciparum malaria. Eur J Clin Pharmacol 2006, 62:367‐371.
21. Deen JL, von Seidlein L, Pinder M, Walraven GE, Greenwood B: The safety of the combination artesunate and pyrimethamine sulfadoxine given during pregnancy. Trans R Soc Trop Med Hyg
Chapter 2
36
22. 2001, 95:424. 23. Adam I, Elwasila E, Mohammed Ali DA, Elansari E, Elbashir M: Artemether in the treatment of falciparum
malaria during pregnancy in eastern Sudan. Trans R Soc Trop Med Hyg 2004, 98:509. 24. Adam I, Ali DM, Abdalla MA: Artesunate plus sulfadoxine‐pyrimethamine in the treatment of
uncomplicated Plasmodium falciparum malaria during pregnancy in eastern Sudan. Trans R Soc Trop Med Hyg 2006, 100:632‐635.
25. Ashley EA, White NJ: Artemisinin‐based combinations. Curr Opin Infect Dis 2005, 18:531‐536. 26. Ashley EA, Krudsood S, Phaiphun L, Srivilairit S, McGready R, Leowat‐ tana W, Hutagalung R, Wilairatana P,
Brockman A, Looareesuwan S, Nosten F, White NJ: Randomized, controlled dose‐optimization studies of dihydroartemisinin‐piperaquine for the treatment of uncomplicated multidrug‐resistant falciparum malaria in Thailand. J Infect Dis 2004, 190:1773‐1782.
27. Ashley EA, McGready R, Hutagalung R, Phaiphun L, Slight T, Proux S, Thwai KL, Barends M, Looareesuwan S, White NJ, Nosten F: A ran‐ domized, controlled study of a simple, once‐daily regimen of dihydroartemisinin‐piperaquine for the treatment of uncomplicated, multidrug‐resistant falciparum malaria. Clin Infect Dis 2005, 41:425‐432.
28. Piola P, Fogg C, Bajunirwe F, Biraro S, Grandesso F, Ruzagira E, Babi‐ gumira J, Kigozi I, Kiguli J, Kyomuhendo J, Ferradini L, Taylor W, Checchi F, Guthmann JP: Supervised versus unsupervised intake of six‐dose artemether‐lumefantrine for treatment of acute, uncomplicated Plasmodium falciparum malaria in Mbarara, Uganda: a randomised trial. Lancet 2005, 365:1467‐1473.
29. Christianson A, Howson CP, Modell B: March of Dimes: Global report on birth defects. The hidden toll of dying and disabled children. New York: March of Dimes Birth Defects Foundation; 2006.
30. Adjei GO, Goka BQ, Kurtzhals JA: Neurotoxicity of artemisinin derivatives. Clin Infect Dis 2006, 43:1618. 31. McCall MB, Beynon AJ, Mylanus EA, van der Ven AJ, Sauerwein RW: No hearing loss associated with the use
of artemether‐lumefantrine to treat experimental human malaria. Trans R Soc Trop Med Hyg 2006, 100:1098‐1104.
32. Toovey S: A case‐control auditory evaluation of patients treated with artemether‐lumefantrine. Am J Trop Med Hyg 2006, 74:939‐940. author reply 940
33. Nosten F, Vincenti M, Simpson J, Yei P, Thwai KL, de Vries A, Chong‐ suphajaisiddhi T, White N: The effects of mefloquine treatment in pregnancy. Clin Inf Dis 1999, 28:808.
34. Adam I, Ali DA, Alwaseila A, Kheir MM, Elbashir MI: Mefloquine in the treatment of falciparum malaria during pregnancy in Eastern Sudan. Saudi Med J 2004, 25:1400‐1402.
35. Phillips‐Howard PA, Steffen R, Kerr L, Vanhauwere B, Schildknecht J, Fuchs E, Edwards R: Safety of mefloquine and other antimalarial agents in the first trimester of pregnancy. J Travel Med 1998, 5:121‐126.
36. Steketee RW, Wirima JJ, Slutsker L, Khoromana CO, Heymann DL, Breman JG: Malaria treatment and prevention in pregnancy: indications for use and adverse events associated with use of chloroquine or mefloquine. Am J Trop Med Hyg 1996, 55:50‐56.
37. Vanhauwere B, Maradit H, Kerr L: Post‐marketing surveillance of prophylactic mefloquine (Lariam) use in pregnancy. Am J Trop Med Hyg 1998, 58:17‐21.
38. Clark R, Kumemura M, Makori N, Nakata Y, Bernard F, Harrell A, White TEK, Arima A: Artesunate: Developmental Toxicity in Monkeys [Abstract]. 46th annual meeting of the Teratology Society; Tucson, Arizona 2006.
39. Laffan S, James A, Maleeff B, Pagana J, Bushdid P, Clark R, White T: Mitochondrial involvment of artesunate toxicity in rat embryonic erythroblasts. [Abstract]. 46th annual meeting of the Teratology Society; Tucson, Arizona 2006.
40. Lang T, Hughes D, Kanyok T, Kengeya‐Kayondo J, Marsh V, Haaland A, Pirmohamed M, Winstanley P: Beyond registration – measuring the public‐health potential of new treatments for malaria in Africa. Lancet Infect Dis 2006, 6:46‐52.
41. WHO: Antenatal care in developing countries: promises, achievements and missed opportunities: an analysis of trends, levels and differentials, 1990–2001. Geneva: WHO; 2003.
37
Chapter3
QuantifyingtheNumberofPregnanciesatRiskofMalariain2007:ADemographicStudy
StephanieDellicour1,AndrewJ.Tatem2,3,CarlosA.Guerra2,4,RobertW.Snow2,5,FeikoO.terKuile1,6
1 Child and Reproductive Health Group, Liverpool School of Tropical Medicine, Liverpool,
United Kingdom, 2 Malaria Public Health and Epidemiology Group, Centre for Geographic
Medicine, Kenyan Medical Research Institute–University of Oxford–Wellcome Trust
Collaborative Programme, Nairobi, Kenya, 3 Department of Geography and Emerging
Pathogens Institute, University of Florida, Gainesville, Florida, United States of America, 4
Spatial Ecology and Epidemiology Group, Department of Zoology, University of Oxford,
Oxford, United Kingdom, 5 Centre for Tropical Medicine, Nuffield Department of Clinical
Medicine, University of Oxford, CCVTM, Oxford, United Kingdom, 6 Department of
Infectious Diseases, Tropical Medicine & AIDS, Academic Medical Centre, University of
Amsterdam, Amsterdam, The Netherlands
PLoS Medicine 2010, 7:1 Copyright© This is an open‐access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited
Chapter 3
38
AbstractBackground: Comprehensive and contemporary estimates of the number of pregnancies at risk of
malaria are not currently available, particularly for endemic areas outside of Africa. We derived
global estimates of the number of women who became pregnant in 2007 in areas with Plasmodium
falciparum and P. vivax transmission.
Methods and Findings: A recently published map of the global limits of P. falciparum transmission
and an updated map of the limits of P. vivax transmission were combined with gridded population
data and growth rates to estimate total populations at risk of malaria in 2007. Country‐specific
demographic data from the United Nations on age, sex, and total fertility rates were used to
estimate the number of women of child‐bearing age and the annual rate of live births. Subregional
estimates of the number of induced abortions and country‐specific stillbirths rates were obtained
from recently published reviews. The number of miscarriages was estimated from the number of live
births and corrected for induced abortion rates. The number of clinically recognised pregnancies at
risk was then calculated as the sum of the number of live births, induced abortions, spontaneous
miscarriages, and stillbirths among the population at risk in 2007. In 2007, 125.2 million pregnancies
occurred in areas with P. falciparum and/or P. vivax transmission resulting in 82.6 million live births.
This included 77.4, 30.3, 13.1, and 4.3 million pregnancies in the countries falling under the World
Health Organization (WHO) regional offices for South‐East‐Asia (SEARO) and the Western‐Pacific
(WPRO) combined, Africa (AFRO), Europe and the Eastern Mediterranean (EURO/EMRO), and the
Americas (AMRO), respectively. Of 85.3 million pregnancies in areas with P. falciparum transmission,
54.7 million occurred in areas with stable transmission and 30.6 million in areas with unstable
transmission (clinical incidence,1 per 10,000 population/year); 92.9 million occurred in areas with P.
vivax transmission, 53.0 million of which occurred in areas in which P. falciparum and P. vivax co‐
exist and 39.9 million in temperate regions with P. vivax transmission only.
Conclusions: In 2007, 54.7 million pregnancies occurred in areas with stable P. falciparum malaria
and a further 70.5 million in areas with exceptionally low malaria transmission or with P. vivax only.
These represent the first contemporary estimates of the global distribution of the number of
pregnancies at risk of P. falciparum and P. vivax malaria and provide a first step towards a more
informed estimate of the geographical distribution of infection rates and the corresponding disease
burden of malaria in pregnancy.
Quantifying the number of pregnancies at risk of malaria
39
IntroductionMalaria in pregnancy can have devastating consequences to a pregnant woman and the developing
fetus, but comprehensive estimates of the annual number of women who become pregnant each
year in malaria endemic areas and are therefore at risk of malaria are not available, particularly for
Latin America and the Asia‐Pacific regions. These figures are an important first step towards
informing policy makers and for estimating the regional needs for therapeutic and disease
prevention tools for malaria in pregnancy. The most cited global estimate is from the Roll Back
Malaria Partnership, which states that ‘‘each year approximately 50 million women living in malaria
endemic countries throughout the world become pregnant’’ [1]. However, an explanation of the
methods used to derive these estimates is not provided. More comprehensive estimates exist for
Africa and are provided by the Africa Regional Office (AFRO) of the World Health Organization
(WHO) in their widely quoted strategic framework document for malaria prevention and control
during pregnancy in the African region [2]. Their estimate of 24.6 million pregnancies at risk of
malaria (predominantly P. falciparum), is based on the number of live born babies delivered in
malarious areas of Africa in the year 2000 using a combination of malaria risk maps [3] and estimates
of the number of live births from UNICEF [4]. A more recent estimate by the WHO states that ‘‘In
Africa, 30 million women living in malaria endemic areas become pregnant each year’’ [5]. Estimates
for outside of Africa are less clear, particularly for P. vivax. P. vivax is the most widely distributed
human malaria parasite and co‐occurs with P. falciparum in tropical areas but also occurs in
temperate regions outside the limits of P. falciparum transmission. It is the major cause of malaria in
much of Asia and Latin America [6,7], and recent evidence has shown that P. vivax infections are far
from benign and can result in significant morbidity in pregnant women with serious consequences
for maternal and infant health [8–10].
Here we define a global estimate of the number of pregnancies at risk of P. falciparum and P. vivax
malaria in 2007 by combining malaria spatial limits developed by the Malaria Atlas Project (MAP;
www.map.ox.ac.uk), which define the total population at risk of malaria [11], with country‐specific
demographic data on women of childbearing age provided by the United Nations and published data
on induced abortions and spontaneous pregnancy loss.
Methods
DataSourcesThegloballimitsofP.falciparummalaria.
The initial focus of the Malaria Atlas Project has been P. falciparum [12] due to its global
epidemiological significance [13] and better prospects for its control and local elimination [14]. The
global spatial limits of P. falciparum malaria transmission in 2007 have recently been mapped. This
was done by triangulating data on transmission exclusion using biological rules based on
temperature and aridity limits on the bionomics of locally dominant Anopheles vectors, data on
nationally reported case incidence rates, and other medical intelligence [15]. The resulting map
stratifies the malaria endemic world by stable and unstable transmission in 2007 [15]. Unstable
transmission refers to areas where transmission is plausible biologically, but limited, with a clinical
incidence of less than one case per 10,000 population per year. Stable transmission refers to areas
with a minimum of one clinical case per 10,000 population per year [15].
Chapter 3
40
ThegloballimitsofP.vivaxmalaria.
Initial attempts to map the limits of P. falciparum and P. vivax transmission were made by Guerra et
al. [16,17]. The resulting maps and ‘‘masks’’ (mapped areas that are filtered and excluded from
analyses) used were later tested against the Malaria Atlas Project parasite prevalence database to
assess their feasibility [18,19]. This testing revealed that the accuracy to define areas of zero
transmission risk due to very low population densities was limited because of the coarse spatial
resolution of the initial map. Moreover, in the initial mapping [16,17], a high density population
mask was used on the basis of the assumption that no transmission occurs in areas where the
population density is so high that conditions become unsuitable for transmission through the
process of urbanization. However, recent analyses [19] provide evidence suggesting that high
density population masks and urban extent maps should not be used to map zero risk because some
transmission can occur in high density urban areas, although this is significantly lower than in rural
areas [18,19]. Therefore, for the current analyses, the P. vivax limits were redefined using the same
methods as in Guerra et al. [16,17], but without applying the population‐based masks. Also,
previously excluded P. vivax endemic countries have now been added after a more extensive review
of the literature; these include Comoros, Djibouti, Madagascar, and Uzbekistan. This refinement of
the spatial limits of transmission for P. vivax accounted for an approximate 19% increase in the
population at risk (PAR) compared with previous estimates [16,17], principally (18%) due to the
inclusion of major cities.
Griddedpopulationdata.
The Global Rural‐Urban Mapping Project (GRUMP) alpha version provides gridded population counts
and population density estimates for the years 1990, 1995, and 2000, both adjusted and unadjusted
to the United Nations’ national population estimates [11,20]. The adjusted population counts for the
year 2000 were projected to 2007 by applying national, medium variant, intercensal growth rates by
country [21] using methods described previously [22].
Annualnumberofpregnanciespercountry.
The number of pregnancies was calculated as the sum of the number of live births, induced
abortions, and spontaneous pregnancy loss (including miscarriages and stillbirths) in 2007.
Livebirths.
The annual number of live births in 2007 was estimated per country using demographic data on the
proportion of women of childbearing age (WOCBAs) within a population and the total fertility rates.
The data were abstracted from the United Nations’ national population estimates, which provide
publicly accessible demographic information by year, age, sex, and country for Africa, Asia, and the
Americas [23]. The number of WOCBAs in each country, defined as the mid‐year resident number of
women aged between 15 and 49y, was obtained for the years 2005 and 2010 (interim years are not
available), and the number of WOCBAs for 2007 was calculated as the midpoint between the 2005
and 2010 estimates. The fraction of WOCBAs per country was then calculated as the number of
WOCBAs in 2007 divided by the mid‐year resident population at risk in 2007 (available by year).
The total fertility rate (TFR) is an age‐standardised measure of fertility and corresponds to
the total number of children that would be born alive to a woman entering her childbearing
Quantifying the number of pregnancies at risk of malaria
41
years at age 15y if she lived to the end of her childbearing years (age 49y) and if her fertility
during these 35 reproductive years was the same as the average woman of childbearing age.
The total fertility rate divided by 35 is the average number of live births per WOCBA per year and
when multiplied by 1,000 this is expressed as the rate of live births per 1,000 WOCBAs per year.
Inducedabortions,miscarriages,andstillbirthrates.
Subregional data on induced abortion rates were obtained from a recently published review that
calculated the worldwide, regional, and subregional incidence of safe and unsafe abortions in
women of childbearing age in 2003 by use of reports from official national reporting systems,
nationally representative demographic health surveys, hospital data, other surveys, and published
studies [24].
Country‐specific information on stillbirth rates was abstracted from model‐based estimates
published by Stanton et al. [25] that derived data from vital registration, demographic and health
surveys (DHS), and data from study reports integrated into a regression model. Regional estimates
were used for three malaria endemic countries for which country‐specific estimates were not
available (French Guiana, Mayotte, and Timor‐Leste).
Country‐specific data on miscarriages (spontaneous abortions) are not available. To calculate the
proportion of pregnancies resulting in miscarriage, a method was applied that uses multipliers to
work backwards from the (known) number of live births and induced abortions to recover the
(unknown) underlying number of pregnancies that ‘‘produced’’ them, as described in detail
previously [24,26–28]. The method takes account of pregnancies that are terminated voluntarily
during the period of risk for miscarriage and estimates the number of spontaneous pregnancy loss
(stillbirths and miscarriages) as 10% of induced abortions plus 20% of live births. It is based on
clinical studies of rates of pregnancy loss by gestational age that indicate that for each 100 induced
abortions an additional ten clinically recognised pregnancies will have aborted spontaneously prior
to the average gestational age of induced abortions in that population, and that approximately 120
additional clinically recognised pregnancies are required to ‘‘produce’’ 100 live births [27,28]. For
example, in Afghanistan it was estimated that in 2007 1.182 million live births occurred among a
population of 27 million and a further 0.284 million induced abortions occurred. The number of
spontaneous pregnancy losses (the sum of the number of stillbirths and miscarriages) was therefore
estimated at 0.2 x 1,182 plus 0.1 x 0.284 = 0.265 million, and the total number of pregnancies as
1.731 million. The reported number of miscarriages used in this manuscript represents the number
of spontaneous pregnancy losses calculated through the multiplier method as described above,
minus the country‐specific number of stillbirths obtained from the review by Stanton et al. [25]. The
estimates provided in this study refer to clinically recognised pregnancies and do not take into
account the potentially large but unknown rates of embryonic loss that may occur in the first 4–6 wk
of gestation.
Estimatingtheannualnumberofpregnanciesexposedtomalaria.
To obtain the total population at risk, the limits of stable and unstable P. falciparum transmission
and the limits of P. vivax transmission described above were overlaid onto the Global Rural‐Urban
Mapping Project (GRUMP) alpha surface, projected to 2007. For every malaria endemic country of
the world, the population within each set of limits was extracted, following approaches described
Chapter 3
42
Total n of
aSource: United Nations Development Program (in millions). bThe total number of pregnancies is the sum of the number of live‐births, stillbirths, spontaneous, and induced abortions (in millions). cThe total pregnancy rate (TPR) and the annual pregnancy rate per 1,000 WOCBAs are weighted means per region and is for illustration purposes only. The number of pregnancies was derived directly as the sum of the national estimates within each region and globally.
previously [13]. The number of pregnancies at risk of malaria was then calculated from the total
annual number of pregnancies estimated to have occurred in 2007 in the entire country
multiplied by the fraction of the total resident population living within the spatial limits of
malaria transmission in that country.
ResultsTables 1 and 2 provide a summary of the total population living within the global spatial limits of
malaria transmission in 2007, and the corresponding number of total population, pregnancies,
and live births, stratified by species and transmission patterns (within areas of assumed unstable
and stable P. falciparum transmission), globally and by WHO region.
The compiled data showed that, globally, 125.2 million women living in areas with P. falciparum
and/or P. vivax transmission became pregnant in 2007: 77.4 million (61.8%) in the countries that fall
under the regional office of the WHO for the South East Asian (SEARO) and the Western Pacific
Region (WPRO); 30.3 million (24.2%) in AFRO; 13.1 million (10.5%) in the Eastern Mediterranean and
European Region (EMRO and EURO); and only 4.3 million (3.4%) in the American Region (AMRO)
(Table 3). Figures 1 and 2 display the same analysis by species, but depicted by continent rather than
by WHO region. Of the 125.2 million pregnancies, 82.6 million (66.0%) are estimated to result in live
births; 48.8 million (63.0%), 22.1 million (72.7%), 9.0 million (68.8%), and 2.7 million (63.1%) in the
SEARO/WPRO, AFRO, EMRO/ EURO, and AMRO regions, respectively (Table 4). It illustrates that the
proportional distribution of pregnancies at risk resulting in live births is slightly different from the
distribution of total pregnancies at risk, primarily reflecting the differences in the proportion of
pregnancies ending in induced abortions, which is much lower in the AFRO region (11.9%) compared
to the global average in the malaria endemic countries of 19.5% [24].
Table 1. Demographic data for malaria endemic countries
WHORO
Region
n of Total Population
WOCBAsa Total n of Pregnanciesb
TPRc Pregnancy Rate Per 1,000 WOCBAs
Live‐
births
Still‐
births
Spontaneous
Abortions
Induced
Abortions
AFRO 43 755 178 36 7.16 204 72.4% 2.3% 13.3% 11.9%
EMRO/EURO 19 544 142 19 4.76 136 68.8% 2.3% 13.1% 15.8%
AMRO 21 530 143 16 3.81 109 63.2% 0.9% 14.0% 22.0%
SEARO/WPRO 19 3,327 881 91 3.62 103 62.4% 1.6% 13.1% 22.8%
Global 102 5,157 1,343 162 4.23 121 65.5% 1.8% 13.3% 19.5%
Quantifying the number of pregnancies at risk of malaria
43
Similar tables with risk estimates by continent and by pregnancy outcome (live‐birth, induced abortions, stillbirths, and miscarriages) are provided in Tables S1, S2, S3. aIncludes countries where P. falciparum and P. vivax co‐exist. bStable transmission, $1 autochthonous P. falciparum cases per 10,000 people per annum; unstable transmission, ,1 autochthonous P. falciparum cases per 10,000 people per annum [15]. MEC, malaria endemic countries; TPR, total pregnancy rate; WHORO, World Health Organization Regional Office.
Table 2. Total population at risk of P. falciparum and/or P. vivax malaria by WHO regional office in 2007 (in millions) (percent of the population in malaria endemic countries at risk).
P.falciparumMalariaOf the 125.2 million pregnancies defined above, 85.3 million occur in areas with P. falciparum
transmission, 51.8% of them (44.2 million) are in the combined SEARO‐WPRO regions and 35.1%
(30.0 million) in the AFRO region. The remainder live in the EMRO‐EURO (9.6%) and AMRO regions
(3.5%) (Figure 3; Table 3). As expected, the top five ranked countries with the highest number of
pregnancies at risk of P. falciparum malaria were the malaria endemic countries with the largest
overall populations: India (28.2 million), Nigeria (6.5 million), Indonesia (4.4 million), Pakistan (3.7
million), and the Democratic Republic of the Congo (3.3 million). Overall, 64.1% of 85.3 million
pregnancies at risk of P. falciparum malaria live in areas with assumed stable transmission (Figure 3).
However, this varies widely by region; from 98.7% in the AFRO region to none in the EURO region. As
depicted in Figure 3, 55.3% of the 44.2 million pregnancies at risk of P. falciparum in the
WPRO/SEARO region occur in areas of very low and unstable transmission.
P.vivaxMalariaGlobally, an estimated 92.9 million pregnancies occurred in areas endemic for P. vivax in 2007
(including in areas where both P. falciparum and P. vivax co‐exist) (Figure 2). The top five ranked
countries include: India (32.9 million), China (21.2 million), Indonesia (6.3 million), Pakistan (5.8
million), and Bangladesh (4.7 million). In the WPRO/SEARO region, where the majority of the
populations at risk of P. vivax live (Figure 4), approximately 98.2% of those pregnancies in malaria
endemic countries occur in areas with P. vivax transmission (alone or combined with P. falciparum).
By contrast this was only 11.9% for the AFRO region where P. vivax transmission is principally
restricted to the horn of Africa region, Madagascar, and the Comoros islands (Figure 4).
The country‐specific demographic data and population at risk estimates (Table S1), as well as
total pregnancies at risk and by specific pregnancy outcomes (live births, induced abortions,
stillbirths, and miscarriages; Table S2) and summary estimates by other regional categories
(continents instead of WHO regions; Table S3), are provided as supplemental information. In
addition, information is provided illustrating which countries are included in the different WHO
WHORO Region
P. falciparum Transmissiona P. vivax
Transmissiona
Any Species
Stable
Transmissionb
Unstable
Transmissionb
Overall Overall Overall
AFRO 599.9 (79.4) 8.4 (1.1) 607.8 (80.5) 73.2 (9.7) 615.4 (81.5)
EMRO/EURO 89.8 (16.5) 101.7 (18.7) 190.9 (35.1) 285.1 (52.4) 343.9 (63.2)
AMRO 41.2 (7.8) 50.2 (9.5) 91.4 (17.2) 96.2 (18.2) 138.2 (26.1)
SEARO/WPRO 654.9 (19.7) 824.9 (24.8) 1479.3 (44.5) 2,722.3 (81.8) 2,770.1 (83.3)
Global 1,385.8 (26.9) 985.1 (19.1) 2,369.4 (45.9) 3,176.9 (61.6) 3,867.6 (75.0)
Chapter 3
44
Similar tables with risk estimates by continent and by pregnancy outcome (live‐birth, induced abortions, stillbirths, and miscarriages) are provided in Tables S1, S2, S3. aIncludes countries where P. falciparum and P. vivax co‐exist. bStable transmission, $1 autochthonous P. falciparum cases per 10,000 people per annum; unstable transmission, ,1 autochthonous P. falciparum cases per 10,000 people per annum [15]. MEC, malaria endemic countries; TPR, total pregnancy rate; WHORO, World Health Organization Regional Office.
regions (also see Figure S1) [29]. In brief, all malaria endemic countries on the African continent
fall under the Africa Regional Office (AFRO), with the exception of Djibouti, Somalia, and Sudan,
which fall under the EMRO office.
Table 3. Number of pregnancies at risk of P. falciparum and/or P. vivax malaria by WHO regional office in 2007 (in millions) (column %).
WHORO Region
P. falciparum Transmissiona P. vivax
Transmissiona
Any Species
Stable
Transmissionb
Unstable
Transmissionb
Overall Overall Overall
AFRO 29.6 (54.1) 0.4 (1.2) 30.0 (35.1) 3.6 (3.9) 30.3 (24.2)
EMRO/EURO 4.0 (7.3) 4.2 (13.7) 8.2 (9.6) 10.4 (11.2) 13.1 (10.5)
AMRO 1.4 (2.5) 1.6 (5.2) 3.0 (3.5) 2.9 (3.1) 4.3 (3.4)
SEARO/WPRO 19.7 (36.1) 24.5 (79.9) 44.2 (51.8) 76.0 (81.8) 77.4 (61.8)
Global 54.7 30.6 85.3 92.9 125.2
DiscussionThis is the first time, to our knowledge, that contemporary species‐specific estimates of the annual
number of pregnancies at risk of malaria globally have been made. Our findings suggest that in
2007 approximately 125 million pregnancies occurred in areas with P. falciparum and/or P. vivax
transmission, resulting in 83 million live births; representing approximately 60% of all pregnancies
globally. Approximately 85 million pregnancies occurred in areas with P. falciparum transmission
and 93 million in areas with transmission of P. vivax transmission, of which about 53 million
occurred in areas where both species co‐exist. The pregnancies at risk estimates for P. falciparum
and P. vivax in Africa (32 million [30 million in the WHO‐AFRO region]) are consistent with the
previous estimates by WHO (25–30 million). By contrast, the numbers at risk outside Africa are
much higher (95 million) than previously estimated (25 million). Comparisons between the
estimates produced in this study and the previous WHO estimates are made difficult because
details of the methodology used by the WHO is not provided and it is not clear if they included all
transmission areas or only areas with stable malaria transmission. Inclusion of only those areas
with stable P. falciparum transmission in our study resulted in global risk estimates of just less
than 55 million pregnancies, 31 million in Africa and 23 million in the other regions, i.e., very
similar to the previous WHO estimates.
However, the numbers of pregnancies at risk outside Africa increase almost 4‐fold if areas with
unstable P. falciparum transmission are included (clinical incidence, 1 per 10,000
population/year) (30 million) and areas situated in the temperate regions outside the limits of P.
falciparum transmission that have P. vivax transmission only (40 million) are also included. It is also
not clear if the previous WHO estimates included pregnancies resulting in live births only or
Quantifying the number of pregnancies at risk of malaria
45
included adjustments for induced abortions or spontaneous pregnancy loss. Since only
approximately two‐thirds of all pregnancies result in live births, estimates that include all
pregnancies are about one‐third higher than estimates based on live births only.
Figure 1. Malaria risk map for P. falciparum and corresponding number of pregnancies in each continent in 2007. doi:10.1371/journal.pmed.1000221.g001
Figure 2. Malaria risk map for P. vivax and corresponding number of pregnancies in each continent in 2007. doi:10.1371/journal.pmed.1000221.g002
Although risk estimates are widely quoted figures, it is important to place them in perspective.
The estimates provided here merely define the global distribution of pregnancies that occur
within the global spatial limits of malaria transmission. These estimates therefore represent ‘‘any
risk’’ of exposure to malaria during pregnancy, and do not represent the distribution of actual
Chapter 3
46
Similar tables with risk estimates by continent and by pregnancy outcome (live‐birth, induced abortions, stillbirths, and miscarriages) are provided in Tables S1, S2, S3. aIncludes countries where P. falciparum and P. vivax co‐exist. bStable transmission, $1 autochthonous P. falciparum cases per 10,000 people per annum; unstable transmission, ,1 autochthonous P. falciparum cases per 10,000 people per annum [15]. MEC, malaria endemic countries; TPR, total pregnancy rate; WHORO, World Health Organization Regional Office.
incidence or health burden on mothers and unborn babies, which is beyond the scope of this
paper. More than half (71 million) of the 125 million pregnancies occur in areas with unstable P.
falciparum transmission (31 million) or with transmission of P. vivax only (40 million), and the risk of
acquiring malaria in these areas is extremely low.
Table 4. Number of live‐births born to pregnancies at risk of at risk of P. falciparum and/or P. vivax malaria by WHO regional office in 2007 (in millions) (column %).
WHORO Region
P. falciparum Transmission a P. vivax
Transmissiona
Any Species
Stable b
Unstable Overall Overall Overall
AFRO 599.9 (79.4) 8.4 (1.1) 607.8 (80.5) 73.2 (9.7) 615.4 (81.5)
EMRO/EURO 89.8 (16.5) 101.7 (18.7) 190.9 (35.1) 285.1 (52.4) 343.9 (63.2)
AMRO 41.2 (7.8) 50.2 (9.5) 91.4 (17.2) 96.2 (18.2) 138.2 (26.1)
SEARO/WPRO 654.9 (19.7) 824.9 (24.8) 1479.3 (44.5) 2,722.3 (81.8) 2,770.1 (83.3)
Global 1,385.8 (26.9) 985.1 (19.1) 2,369.4 (45.9) 3,176.9 (61.6) 3,867.6 (75.0)
Thus, although these 71 million pregnancies represent more than 50% of the global number of
pregnancies at risk, they may only contribute a small proportion to the number of infections in
pregnancy. For example, if the actual incidence of malaria infection in these very low
transmission areas is 1 in 10,000 per person‐year (52 wk), and if the average pregnancy resulting
in a live birth takes 38 wk from fertilisation to term, then 71 million pregnancies at risk may result
in only 5,188 actual malaria infections, whereas in areas with infection rates of 1.36 or higher per
person‐year, all term pregnancies have been potentially exposed to malaria. Furthermore, the
definition of stable transmission for P. falciparum used included all areas with more than one
clinical case per 10,000 population per year. This included almost all pregnancies at risk in the
AFRO Region (99% of the 30 million pregnancies at risk) and 25 million of the 95 million (26%)
pregnancies in the other WHO regions. However, these stable transmission strata cover a very
wide range of transmission intensities and the actual risk of infection to the 55 million
individuals and the impact on maternal and infant health varies enormously within this range.
At the higher end of the transmission spectrum, the majority of malaria infections in pregnancy
remain asymptomatic or pauci‐symptomatic, yet are a major cause of severe maternal anaemia
and preventable low birth weight, especially in the first and second pregnancies. In areas with
stable, but low transmission, and certainly in areas with unstable and exceptionally low
transmission, infections can become severe in all gravidae groups because most women of
childbearing age in these regions have low levels of pre‐pregnancy and pregnancy‐specific
protective immunity to malaria [30]. The most recent version of the World Malaria Map [28]
from the Malaria Atlas Project shows that 89% of the population in stable P. falciparum areas
outside Africa live in areas characterised by low malaria endemicity (defined as P. falciparum
parasite rate in children 2–10 y of age of <5%). This total includes all of the stable P. falciparum
Quantifying the number of pregnancies at risk of malaria
47
transmission areas in the Americas, and 88% of the populations at risk in the Central and South‐
East Asia‐Pacific region [31]. Our estimates do not take seasonality into account and include all
pregnancies occurring throughout the year, whereas those pregnancies that occur outside of the
transmission season may be at no risk, or very low risk of exposure. Our risk estimates for P. vivax
are likely to be less accurate than those for P. falciparum because of greater uncertainties about
the basic biology of transmission and clinical epidemiology. For example, the climatic constraints
on P. vivax transmission are less well defined, the accuracy of clinical reporting of P. vivax in areas
with coincidental P. falciparum is poor, and the untreated hypnozoite stage of P. vivax, which can
remain dormant in infected liver cells for months or years, provides an additional challenge to the
interpretation of prevalence and incidence data [15]. We used a refined P. vivax risk map that
resulted in a 19% increase over previous population at risk estimates (adjusted for population
growth) [18,19], principally resulting from the removal of the population density masks and
thereby the inclusion of many large cities. In most of these cities, pregnancies will be at low or
very low risk of autochthonous infections. Imported malaria associated with travel to rural areas
may be a greater risk factor in these cities. We did not consider infections with P. ovale or P.
malariae, as their distribution is not well described and the adverse effects on maternal health and
the newborn infant are unknown.
Figure 3. Distribution of the number of pregnancies in areas with P. falciparum malaria in 2007 by WHO regions and the corresponding proportion living under stable versus unstable transmission. Blue, SEARO and WPRO; green, AFRO; orange, EMRO; red, AMRO. doi:10.1371/journal.pmed.1000221.g003
Chapter 3
48
Figure 4. Distribution of the number of pregnancies in malaria endemic areas in 2007 by WHO regions and by species (P. vivax transmission only, P. falciparum transmission only or transmission of both species). Blue, SEARO and WPRO; green, AFRO; orange, EMRO; red, AMRO. Pv, P. vivax; Pf, P. falciparum. doi:10.1371/journal.pmed.1000221.g004
In the current analysis we used the map of the global spatial limits of P. falciparum malaria, which
stratifies the malaria endemic world by stable and unstable transmission published in 2008 [15].This
map uses a simple divide between very low risk and higher transmission intensities and a crude
proxy to account for the corresponding levels of acquired immunity in women of childbearing age.
As a next step, we will examine the burden of malaria in pregnancy in terms of health impact on the
pregnant women (e.g., febrile episodes, impact on maternal anaemia and maternal mortality), the
newborn baby (e.g., impact on the frequency of preterm births and low birth‐weight) and the infant
(e.g., susceptibility to malaria). For this project, we will use the more refined P. falciparum
transmission intensity model of risk within the defined stable limits which was developed recently by
the Malaria Atlas Project [31], allowing disease impact calculations across multiple transmission
strata to be made. It is also important to take the different pregnancy outcomes into account in
these further burden estimates. Of the 125 million pregnancies, one in five are estimated to be
terminated voluntarily during the period of risk for miscarriage, and only about two‐thirds (82.6
millions) are expected to result in live births. Although malaria in pregnancy is associated with
miscarriages and stillbirths [30], the majority of the health and economic burden is likely through the
impact on pregnancies that result in live births by increasing the risk of preterm births and low birth‐
weight [30] and by modifying the susceptibility to malaria in the infant [32–34].
Quantifying the number of pregnancies at risk of malaria
49
Most of the existing research and policy guidance for malaria control in pregnancy has focused
on P. falciparum in the stable transmission regions of sub‐Saharan Africa. The results of this study
are consistent with the previous WHO‐RBM risk estimates for areas with stable P. falciparum
malaria in Africa, but our work offers advancement on the existing risk estimates for malaria
endemic countries outside Africa. In these regions, the burden of malaria in pregnancy is less
well defined, both in terms of the number of pregnancies and its actual impact on health. Policy
guidelines for malaria control in pregnancy are also less well developed for these regions.
These estimates of the number of pregnancies at risk of malaria provide a first step towards a
spatial map of the burden of malaria in pregnancy and a more informed platform with which to
estimate the associated disease and economic impact and its geographical distribution. Such
global estimates provide guidance in terms of priority setting for resource allocation for both
research and policy for the control of malaria in pregnancy. This project provides a dynamic
framework that allows risk estimates to be updated when new risk maps of P. falciparum and P.
vivax become available as the world attempts to move towards malaria elimination and
eradication.
Acknowledgments
We acknowledge the support of the Kenya Medical Research Institute. We would like to thank
Richard Steketee for helpful comments on the draft manuscript.
AuthorContributions
ICMJE criteria for authorship read and met: SD AJT CAG RWS FOtK. Agree with the manuscript’s
results and conclusions: SD AJT CAG RWS FOtK. Designed the experiments/the study: SD RWS FOtK.
Analyzed the data: SD AJT RWS FOtK. Collected data/did experiments for the study: AJT CAG FOtK.
Wrote the first draft of the paper: SD. Contributed to the writing of the paper: SD AJT CAG RWS
FOtK.
References
1. RBM Pregnant women and Infants, Available:http://malaria.who.int/ pregnantwomenandinfants.html Geneva: WHO, Accessed 5 January 2010.
2. World Health Organization (2004) A strategic framework for malaria prevention and control during pregnancy in the African region. Brazzaville: World Health Organization: Regional Office for Africa, AFR/MAL/04/01 AFR/MAL/04/01.
3. Snow RW, Craig M, Deichmann U, Marsh K (1999) Estimating mortality, morbidity and disability due to malaria among Africa’s non‐pregnant population. Bull World Health Organ 77: 624–640.
4. United Nations Children’s Fund (1998) The state of the world’s children. New York: Oxford University Press.
5. World Health Organization (2003) Available: http://www.who.int/features/ 2003/04b/en/ Geneva: WHO, Accessed 5 January 2010.
6. Baird JK (2007) Neglect of Plasmodium vivax malaria. Trends Parasitol 23: 533–539. 7. Hay SI, Guerra CA, Tatem AJ, Noor AM, Snow RW (2004) The global distribution and population at
risk of malaria: past, present, and future. Lancet Infect Dis 4: 327–336. 8. Singh N, Shukla MM, Sharma VP (1999) Epidemiology of malaria in pregnancy in central India. Bull
World Health Organ 77: 567–572. 9. Nosten F, McGready R, Simpson JA, Thwai KL, Balkan S, et al. (1999) Effects of Plasmodium vivax
malaria in pregnancy. Lancet 354: 546–549.
Chapter 3
50
10. Poespoprodjo JR, Fobia W, Kenangalem E, Lampah DA, Warikar N, et al. (2008) Adverse pregnancy outcomes in an area where multidrug‐resistant plasmodium vivax and Plasmodium falciparum infections are endemic. Clin Infect Dis 46: 1374–1381.
11. Balk DL, Deichmann U, Yetman G, Pozzi F, Hay SI, et al. (2006) Determining global population distribution: methods, applications and data. Adv Parasitol 62: 119–156.
12. Hay SI, Snow RW (2006) The Malaria Atlas Project: developing global maps of malaria risk. PLoS Med 3: e473. doi:10.1371/journal.pmed.0030473.
13. Snow RW, Guerra CA, Noor AM, Myint HY, Hay SI (2005) The global distribution of clinical episodes of Plasmodium falciparum malaria. Nature 434: 214–217.
14. Hay SI, Smith DL, Snow RW (2008) Measuring malaria endemicity from intense to interrupted transmission. Lancet Infect Dis 8: 369–378.
15. Guerra CA, Gikandi PW, Tatem AJ, Noor AM, Smith DL, et al. (2008) The limits and intensity of Plasmodium falciparum transmission: implications for malaria control and elimination worldwide. PLoS Med 5: e38. doi:10.1371/ journal.pmed.0050038.
16. Guerra CA, Snow RW, Hay SI (2006) Mapping the global extent of malaria in 2005. Trends Parasitol 22: 353–358.
17. Guerra CA, Snow RW, Hay SI (2006) Defining the global spatial limits of malaria transmission in 2005. Adv Parasitol 62: 157–179.
18. Guerra CA (2007) Mapping the contemporary global distribution limits of malaria using empirical data and expert opinion. Oxford: University of Oxford. 258 p.
19. Tatem AJ, Guerra CA, Kabaria CW, Noor AM, Hay SI (2008) Human population, urban settlement patterns and their impact on Plasmodium falciparum malaria endemicity. Malar J 7: 218.
20. Center for International Earth Science Information Network (CIESIN)/ Columbia University; International Food Policy Research Institute (IFPRI)/ the World Bank/and Centro Internacional de Agricultura Tropical (CIAT) (2007) Global Rural Urban Mapping Project (GRUMP) alpha: Gridded Population of the World; version 2 with urban reallocation (GPW‐UR). Palisades (New York): CIESIN; Columbia University, Available: http://sedac. ciesin.columbia.edu/gpw. Accessed 5 January 2010.
21. United Nations Population Division; Department of Economic and Social Affairs (2007) World population prospects; the 2006 revision; ST/ESA/SER.A/ 261/ES. New York: United Nations.
22. Hay SI, Noor AM, Nelson A, Tatem AJ (2005) The accuracy of human population maps for public health application. Trop Med Int Health 10: 1073–1086.
23. UNPD (2006) World population prospects: population database. Available: http://esa.un.org/unpp/. Accessed 5 January 2010.
24. Sedgh G, Henshaw S, Singh S, Ahman E, Shah IH (2007) Induced abortion: estimated rates and trends worldwide. Lancet 370: 1338–1345.
25. Stanton C, Lawn JE, Rahman H, Wilczynska‐Ketende K, Hill K (2006) Stillbirth rates: delivering estimates in 190 countries. Lancet 367: 1487–1494.
26. Leridon H (1977) Human fertility: the basic components (Table 4.20). Chicago: University of Chicago Press.
27. Hammerslough CR (1992) Estimating the probability of spontaneous abortion in the presence of induced abortion and vice versa. Public Health Rep 107: 269–277.
28. Henshaw SK, Binkin NJ, Blaine E, Smith JC (1985) A portrait of American women who obtain abortions. Fam Plann Perspect 17: 90–96.
29. World Health Organization (2009) WHO regional offices, Available: http:// www.who.int/about/regions/en/ Geneva: World Health Organisation, Accessed 5 January 2010.
30. Desai M, ter Kuile FO, Nosten F, McGready R, Asamoa K, et al. (2007) Epidemiology and burden of malaria in pregnancy. Lancet Infect Dis 7: 93–104.
31. Hay SI, Guerra CA, Gething PW, Patil AP, Tatem AJ, et al. (2009) A world malaria map: Plasmodium falciparum endemicity in 2007. PLoS Med 6: e1000048. doi:10.1371/journal.pmed.1000048.
32. Mutabingwa TK, Bolla MC, Li JL, Domingo GJ, Li X, et al. (2005) Maternal malaria and gravidity interact to modify infant susceptibility to malaria. PLoS Med 2: e407. doi:10.1371/journal.pmed.0020407.
33. Le Hesran JY, Cot M, Personne P, Fievet N, Dubois B, et al. (1997) Maternal placental infection with Plasmodium falciparum and malaria morbidity during the first 2 years of life. Am J Epidemiol 146: 826–831.
34. Schwarz NG, Adegnika AA, Breitling LP, Gabor J, Agnandji ST, et al. (2008) Placental malaria increases malaria risk in the first 30 months of life. Clin Infect Dis 47: 1017–1025.
51
Chapter3Appendix&SupportingInformation
Figure S1 Map of the WHO Regions (http://www.who.int/ about/regions/en/index.html).
Found at: doi:10.1371/journal.pmed.1000221.s001 (1.07 MB TIF)
Table S1 Demographic characteristics and total population at risk of P. falciparum and/or
P. vivax malaria by malaria endemic country and by WHO regional office in 2007 (in
millions).
Found at: doi:10.1371/journal.pmed.1000221.s002 (0.38 MB PDF)
Table S2 Total number of pregnancies by pregnancy outcome in areas with P. falciparum
and/or P. vivax transmission by continent in 2007 (in millions).
Found at: doi:10.1371/journal.pmed.1000221.s003 (0.54 MB PDF)
Table S3 Total population, number of pregnancies, and number of live‐births born to
pregnancies in malaria endemic countries by continent in 2007 (in millions).
Found at: doi:10.1371/journal.pmed.1000221.s004 (0.28 MB PDF)
Chapter 3 Appendix
52
Figure S1. Map of the WHO Regions (http://www.who.int/about/regions/en/index.html).
doi:10.1371/journal.pmed.1000221.s001
Quantifying the number of pregnancies at risk of malaria
53
Table S1: Demograp
hic characteristics an
d total population at risk of P.falciparum and/or P.vivax malaria by malaria endemic country an
d by WHO regional office in
2007 (in m
illions)
Table S1
2007
Demograp
hic data
Total population at risk*
UN National Population Estim
ates
Total
Population
both
sexes*
WOCBAs*
Number of
pregnancies*†
TFR
TPR
Pregnancy
rate per
1000
WOCBAs
Percentage pregnan
cies ending in:
P. falciparum#
P. vivax#
Any
species
Countries
Live‐
births
Still‐
births
Induced
Abortions
Spontaneo
us
Abortions
Stable¶
Unstable¶
Overall
Overall
Overall
AFR
O 1
755
178
35
5.18
7.16
199
72.4%
2.3%
11.9%
13.3%
600
8
608
73
615
Angola
17.02
3.93
0.96
6.43
8.56
245
75.1%
2.5%
9.0%
13.4%
15.41
0.28
15.70
0.00
15.70
Ben
in
9.03
2.08
0.44
5.42
7.34
210
73.9%
2.3%
10.3%
13.5%
7.66
0.00
7.66
0.00
7.66
Botswana
1.88
0.50
0.06
2.90
4.25
121
68.3%
1.3%
16.4%
14.0%
0.89
0.00
0.89
0.00
0.89
Burkina Faso
14.78
3.39
0.79
6.00
8.12
232
73.9%
2.0%
10.3%
13.8%
14.23
0.00
14.23
0.00
14.23
Burundi
8.51
2.01
0.56
6.80
9.66
276
70.4%
2.5%
14.1%
13.0%
5.66
0.00
5.66
0.00
5.66
Cam
eroon
18.55
4.47
0.73
4.31
5.74
164
75.1%
2.1%
9.0%
13.8%
16.95
0.00
16.95
0.00
16.95
Cape Verde
0.53
0.14
0.02
3.37
4.56
130
73.9%
1.2%
10.3%
14.6%
0.00
0.25
0.25
0.00
0.25
Cen
tral African
Republic
4.34
1.03
0.18
4.58
6.10
174
75.1%
2.2%
9.0%
13.7%
4.15
0.00
4.15
0.00
4.15
Chad
10.78
2.41
0.57
6.20
8.26
236
75.1%
2.6%
9.0%
13.3%
9.76
0.16
9.92
0.00
9.92
Comoros
0.84
0.21
0.04
4.30
6.11
174
70.4%
1.9%
14.1%
13.6%
0.59
0.00
0.59
0.64
0.64
Congo
3.77
0.91
0.16
4.49
5.98
171
75.1%
2.2%
9.0%
13.8%
3.33
0.00
3.33
0.00
3.33
Dem
. Rep
. of the Congo
62.64
13.90
3.55
6.70
8.92
255
75.1%
2.6%
9.0%
13.3%
57.97
0.00
57.97
0.00
57.97
Côte d'Ivoire
19.26
4.58
0.79
4.46
6.04
173
73.9%
2.6%
10.3%
13.2%
17.80
0.00
17.80
0.00
17.80
Equatorial Guinea
0.51
0.12
0.02
5.36
7.14
204
75.1%
2.0%
9.0%
13.9%
0.51
0.00
0.51
0.00
0.51
Eritrea
4.85
1.20
0.25
5.05
7.17
205
70.4%
2.0%
14.1%
13.5%
3.33
0.95
4.28
4.80
4.82
Ethiopia
83.10
19.51
4.19
5.29
7.51
215
70.4%
2.6%
14.1%
12.9%
46.08
1.50
47.59
49.05
53.13
Gabon
1.33
0.35
0.04
3.06
4.08
116
75.1%
1.4%
9.0%
14.5%
1.33
0.00
1.33
0.00
1.33
Gam
bia
1.71
0.40
0.07
4.70
6.36
182
73.9%
2.1%
10.3%
13.7%
1.51
0.00
1.51
0.00
1.51
Ghana
23.48
5.84
0.87
3.84
5.20
149
73.9%
1.8%
10.3%
14.0%
22.21
0.00
22.21
0.00
22.21
Guinea
9.37
2.14
0.45
5.44
7.37
210
73.9%
2.2%
10.3%
13.6%
9.23
0.00
9.23
0.00
9.23
Guinea
‐Bissau
1.70
0.37
0.10
7.07
9.57
274
73.9%
2.7%
10.3%
13.1%
1.42
0.00
1.42
0.00
1.42
Ken
ya
37.54
9.14
1.84
4.96
7.04
201
70.4%
3.3%
14.1%
12.2%
25.62
0.18
25.80
0.00
25.80
Liberia
3.75
0.84
0.22
6.77
9.17
262
73.9%
2.4%
10.3%
13.4%
3.43
0.00
3.43
0.00
3.43
Chapter 3 Appen
dix
54
Table S1
2007
Demograp
hic data
Total population at risk*
UN National Population Estim
ates
Total
Population
both
sexes*
WOCBAs*
Number of
pregnancies*†
TFR
TPR
Pregnancy
rate per
1000
WOCBAs
Percentage pregnan
cies ending in:
P. falciparum#
P. vivax#
Any
species
Countries
Live‐
births
Still‐
births
Induced
Abortions
Spontaneo
us
Abortions
Stable¶
Unstable¶
Overall
Overall
Overall
Madagascar
19.68
4.63
0.90
4.78
6.79
194
70.4%
2.1%
14.1%
13.4%
17.28
0.00
17.28
18.74
18.74
Malaw
i 13.93
3.11
0.70
5.59
7.94
227
70.4%
2.9%
14.1%
12.6%
13.45
0.00
13.45
0.00
13.45
Mali
12.34
2.80
0.71
6.52
8.83
252
73.9%
1.8%
10.3%
14.0%
12.34
0.47
12.34
0.00
12.34
Mauritania
3.12
0.76
0.13
4.37
5.92
169
73.9%
2.3%
10.3%
13.5%
0.93
0.40
1.33
0.00
1.33
Mayotte*
0.21
0.05
0.01
5.65
8.02
229
70.4%
2.3%
14.1%
13.1%
0.00
0.21
0.21
0.00
0.21
Mozambique
21.40
5.10
1.06
5.11
7.26
207
70.4%
2.3%
14.1%
13.2%
21.06
0.00
21.06
0.00
21.06
Nam
ibia
2.07
0.54
0.07
3.19
4.67
133
68.3%
1.3%
16.4%
14.0%
1.25
0.38
1.64
0.00
1.64
Niger
14.23
3.04
0.85
7.19
9.74
278
73.9%
2.9%
10.3%
12.9%
13.19
0.62
13.81
0.00
13.81
Nigeria
148.09
34.57
7.11
5.32
7.20
206
73.9%
2.3%
10.3%
13.5%
134.60
0.00
134.60
0.00
134.60
Rwanda
9.73
2.47
0.59
5.92
8.41
240
70.4%
2.2%
14.1%
13.3%
5.03
0.00
5.03
0.00
5.03
Sao Tome and Principe
0.16
0.04
0.01
3.85
5.13
147
75.1%
1.9%
9.0%
14.0%
0.13
0.00
0.13
0.00
0.13
Senegal
12.38
2.96
0.54
4.69
6.35
181
73.9%
2.0%
10.3%
13.8%
10.82
0.00
10.82
0.00
10.82
Sierra Leo
ne
5.87
1.36
0.34
6.47
8.76
250
73.9%
2.8%
10.3%
13.0%
5.50
0.00
5.50
0.00
5.50
South Africa
48.58
12.98
1.43
2.64
3.86
110
68.3%
1.2%
16.4%
14.1%
3.44
2.95
6.39
0.00
6.39
Swaziland
1.14
0.30
0.04
3.45
5.05
144
68.3%
1.5%
16.4%
13.8%
0.23
0.00
0.23
0.00
0.23
Tanzania
40.45
9.38
1.96
5.16
7.33
209
70.4%
2.1%
14.1%
13.4%
39.84
0.00
39.84
0.00
39.84
Togo
6.59
1.58
0.29
4.80
6.50
186
73.9%
2.0%
10.3%
13.8%
5.45
0.00
5.45
0.00
5.45
Uganda
30.88
6.65
1.74
6.46
9.17
262
70.4%
2.3%
14.1%
13.2%
27.03
0.00
27.03
0.00
27.03
Zambia
11.92
2.71
0.57
5.18
7.36
210
70.4%
2.2%
14.1%
13.3%
11.84
0.00
11.84
0.00
11.84
Zimbabwe
13.35
3.42
0.44
3.19
4.53
129
70.4%
1.6%
14.1%
13.9%
7.44
0.00
7.44
0.00
7.44
Quantifying the number of pregnancies at risk of malaria
55
Table S1
2007
Demograp
hic data
Total population at risk*
UN National Population Estim
ates
Total
Population
both
sexes*
WOCBAs*
Number of
pregnancies*†
TFR
TPR
Pregnancy
rate per
1000
WOCBAs
Percentage pregnan
cies ending in:
P. falciparum#
P.
vivax#
Any
species
Countries
Live‐
births
Still‐
births
Induced
Abortions
Spontaneo
us
Abortions
Stable¶
Unstable¶
Overall
Overall
Overall
EMRO
409
104
16
3.65
5.31
172
68.8%
2.5%
15.8%
12.8%
90
98
188
262
320
Afghanistan 2
27.15
5.85
1.73
7.07
10.35
296
68.3%
3.2%
16.4%
12.1%
4.56
12.53
17.10
16.60
20.43
Djibouti 1
0.83
0.22
0.03
3.95
5.61
160
70.4%
2.5%
14.1%
13.0%
0.02
0.41
0.43
0.71
0.71
Iran
2
71.21
20.69
1.77
2.04
2.99
85
68.3%
0.7%
16.4%
14.6%
0.15
2.72
2.87
47.46
47.67
Iraq
2
28.99
7.11
1.25
4.26
6.14
176
69.3%
1.9%
15.3%
13.5%
0.00
0.00
0.00
9.21
9.21
Oman
2
2.60
0.63
0.08
3.00
4.33
124
69.3%
0.7%
15.3%
14.7%
0.00
0.00
0.00
0.04
0.04
Pakistan 2
163.90
41.23
6.07
3.52
5.15
147
68.3%
2.9%
16.4%
12.4%
30.74
68.30
99.04
155.73
155.73
Saudi A
rabia 2
24.74
6.00
0.83
3.35
4.83
138
69.3%
0.8%
15.3%
14.6%
0.72
1.22
1.94
13.00
14.39
Somalia 1
8.70
2.04
0.50
6.04
8.58
245
70.4%
3.3%
14.1%
12.2%
8.70
0.55
8.70
8.70
8.70
Sudan
1
38.56
9.35
1.62
4.23
6.05
173
69.9%
3.9%
14.7%
11.5%
28.99
6.84
35.83
2.98
35.93
Syrian
Arab Rep
. 2
19.93
5.35
0.68
3.08
4.44
127
69.3%
0.8%
15.3%
14.6%
0.00
0.00
0.00
5.88
5.88
Yemen
2
22.39
5.18
1.17
5.50
7.93
227
69.3%
2.5%
15.3%
12.9%
15.93
5.72
21.65
1.88
21.74
EURO
135
37
3
2.23
3.24
91
68.9%
1.1%
15.7%
14.2%
0
3
3
23
23
Arm
enia 6
3.00
0.87
0.05
1.39
2.00
57
69.3%
0.8%
15.3%
14.6%
0.00
0.00
0.00
0.35
0.35
Azerbaijan 6
8.47
2.56
0.19
1.82
2.62
75
69.3%
0.4%
15.3%
15.0%
0.00
0.00
0.00
0.22
0.22
Geo
rgia 6
4.40
1.17
0.07
1.41
2.03
58
69.3%
1.0%
15.3%
14.4%
0.00
0.00
0.00
0.69
0.69
Kyrgyzstan 2
5.32
1.47
0.15
2.48
3.63
104
68.3%
1.3%
16.4%
14.0%
0.00
1.20
1.20
1.62
1.66
Tajikistan 2
6.74
1.78
0.25
3.35
4.90
140
68.3%
1.7%
16.4%
13.6%
0.00
2.16
2.16
4.41
4.87
Turkey 6
74.88
20.51
1.81
2.14
3.09
88
69.3%
1.0%
15.3%
14.4%
0.00
0.00
0.00
13.99
13.99
Turkmen
istan 2
4.97
1.42
0.15
2.50
3.66
105
68.3%
1.3%
16.4%
14.0%
0.00
0.00
0.00
1.35
1.35
Uzbekistan 2
27.37
0.01
0.79
2.49
3.65
104
68.3%
1.3%
16.4%
14.0%
0.00
0.00
0.00
0.29
0.29
Chapter 3 Appen
dix
56
Table S1
2007
Demograp
hic data
Total population at risk*
UN National Population Estim
ates
Total
Population
both sexes*
WOCBAs*
Number of
pregnancies*†
TFR
TPR
Pregnancy
rate per
1000
WOCBAs
Percentage pregnan
cies ending in:
P. falciparum#
P.
vivax#
Any
species
Countries
Live‐
births
Still‐
births
Induced
Abortions
Spontaneo
us
Abortions
Stable¶
Unstable¶
Overall
Overall
Overall
AMRO
530
143
16
2.41
3.81
124
63.2%
0.9%
22.0%
14.0%
41
50
91
96
138
Argen
tina 5
39.53
9.95
1.03
2.25
3.64
104
61.8%
0.7%
23.5%
14.0%
0.00
0.00
0.00
3.02
3.02
Belize 4
0.29
0.07
0.01
2.93
4.35
124
67.3%
1.1%
17.5%
14.1%
0.00
0.17
0.17
0.25
0.26
Bolivia 5
9.53
2.37
0.38
3.50
5.66
162
61.8%
1.3%
23.5%
13.4%
0.22
2.61
2.83
4.05
4.06
Brazil 5
191.79
52.64
5.48
2.25
3.64
104
61.8%
0.8%
23.5%
13.9%
12.79
16.69
29.47
23.07
34.48
Colombia 5
46.16
12.71
1.30
2.22
3.59
103
61.8%
0.8%
23.5%
13.9%
5.26
7.74
13.00
17.32
21.44
Costa Rica 4
4.47
1.21
0.11
2.10
3.12
89
67.3%
0.7%
17.5%
14.5%
0.00
0.00
0.00
1.03
1.03
Dominican
Rep
ublic 4
9.76
2.52
0.34
2.81
4.67
133
60.2%
0.9%
25.3%
13.7%
1.41
2.83
4.24
0.00
4.24
Ecuador 5
13.34
3.44
0.41
2.58
4.17
119
61.8%
1.0%
23.5%
13.7%
4.12
1.65
5.77
3.73
5.85
El Salvador 4
6.86
1.82
0.21
2.68
3.98
114
67.3%
1.1%
17.5%
14.1%
0.00
0.00
0.00
1.10
1.10
Guatem
ala 4
13.35
3.24
0.01
3.27
5.29
151
61.8%
0.8%
23.5%
13.9%
0.14
0.00
0.14
0.20
0.20
Guyana 5
0.74
0.18
0.57
4.15
6.17
176
67.3%
2.3%
17.5%
12.9%
1.02
5.27
6.30
4.07
7.30
Haiti 4
9.60
2.47
0.02
2.33
3.77
108
61.8%
1.0%
23.5%
13.7%
0.14
0.52
0.66
0.74
0.74
Honduras 4
7.11
1.81
0.42
3.54
5.88
168
60.2%
1.5%
25.3%
13.0%
8.60
0.00
8.60
0.00
8.60
Mexico 4
106.54
29.58
0.25
3.31
4.92
141
67.3%
1.3%
17.5%
13.9%
0.88
2.63
3.51
5.46
6.47
Nicaragua 4
5.60
1.48
2.78
2.21
3.28
94
67.3%
0.5%
17.5%
14.7%
0.00
0.00
0.00
15.82
15.82
Panam
a 4
3.34
0.88
0.17
2.76
4.10
117
67.3%
1.3%
17.5%
13.9%
1.57
2.12
3.70
2.21
4.71
Paraguay 5
6.13
1.55
0.10
2.56
3.80
109
67.3%
0.9%
17.5%
14.3%
0.90
0.00
0.90
0.19
1.06
Peru 5
27.90
7.45
0.22
3.08
4.98
142
61.8%
1.0%
23.5%
13.7%
0.00
0.00
0.00
1.52
1.52
Surinam
e 5
0.46
0.12
0.86
2.51
4.06
116
61.8%
0.9%
23.5%
13.8%
3.87
1.70
5.57
8.14
8.33
Ven
ezuela 5
27.66
7.33
0.01
2.42
3.92
112
61.8%
0.9%
23.5%
13.8%
0.01
0.05
0.06
0.04
0.06
Fren
ch Guiana 5
0.20
0.05
0.86
2.55
4.13
118
61.8%
0.8%
23.5%
13.9%
0.22
6.24
6.46
4.29
7.89
Quantifying the number of pregnancies at risk of malaria
57
Table S1
2007
Demograp
hic data
Total population at risk*
UN National Population Estim
ates
Total
Population
both sexes* WOCBAs*
Number of
pregnancies*
† TFR
TPR
Pregnancy
rate per
1000
WOCBAs
Percentage pregnan
cies ending in:
P. falciparum#
P. vivax#
Any species
Countries
Live‐
births
Still‐
births
Induced
Abortions
Spontaneo
us
Abortions
Stable¶
Unstable¶
Overall
Overall
Overall
SEARO
1,770
458
52
2.62
3.94
122
66.3%
1.9%
18.5%
13.2%
564
712
1,276
1,632
1,639
Bangladesh 2
158.67
41.02
4.86
2.83
4.14
118
68.3%
2.7%
16.4%
12.6%
15.12
47.99
63.11
154.08
154.08
Bhutan 2
0.66
0.17
0.02
2.19
3.21
92
68.3%
1.7%
16.4%
13.6%
0.66
0.47
0.66
0.66
0.66
Burm
a 2
48.80
13.97
1.40
2.07
3.51
100
59.0%
1.3%
26.5%
13.2%
42.88
1.91
44.79
47.47
48.80
India 2
1,169.02
294.81
34.65
2.81
4.11
118
68.3%
2.1%
16.4%
13.2%
414.53
535.59
950.12
1109.03
1109.03
Indonesia 2
231.63
64.13
6.77
2.18
3.70
106
59.0%
0.7%
26.5%
13.7%
68.59
81.93
150.52
215.66
217.60
Rep
ublic of Korea 2
48.22
13.10
0.80
1.21
2.13
61
56.8%
1.0%
29.0%
13.2%
0.00
0.00
0.00
12.39
12.39
Nep
al 2
28.20
7.18
0.98
3.28
4.80
137
68.3%
3.9%
16.4%
11.4%
3.40
6.15
9.54
18.01
18.92
Sri Lanka 2
19.30
5.29
0.42
1.88
2.75
79
68.3%
0.7%
16.4%
14.6%
1.75
7.53
9.28
10.46
12.84
Thailand 2
63.88
17.85
1.60
1.85
3.14
90
59.0%
0.6%
26.5%
13.9%
16.53
30.53
47.06
63.88
63.88
Timor‐Leste 2
1.16
0.26
0.08
6.53
11.07
316
59.0%
0.8%
26.5%
13.7%
0.96
0.00
0.96
0.63
0.96
WPRO
1,558
423
40
1.88
3.27
139
57.3%
1.3%
28.4%
13.0%
91
113
203
1,090
1,131
Cam
bodia 2
14.44
3.87
0.60
3.18
5.39
154
59.0%
1.2%
26.5%
13.2%
10.77
2.55
13.33
12.65
14.44
China 2
1,328.63
361.61
31.48
1.73
3.05
87
56.8%
1.4%
29.0%
12.9%
17.13
20.32
37.45
896.81
915.87
Lao 2
5.86
1.52
0.24
3.21
5.44
155
59.0%
1.2%
26.5%
13.2%
5.30
0.01
5.31
2.32
5.32
Malaysia 2
26.57
7.06
0.89
2.60
4.41
126
59.0%
0.5%
26.5%
14.0%
6.29
16.17
22.46
10.15
22.66
Papua New
Guinea 3
6.33
1.60
0.25
3.78
5.45
156
69.3%
1.2%
15.3%
14.2%
4.11
0.00
4.11
3.90
4.46
Philippines 2
87.96
22.52
3.52
3.23
5.47
156
59.0%
0.7%
26.5%
13.7%
26.95
20.41
47.35
78.55
82.01
Solomon Islands 3
0.50
0.12
0.02
3.87
5.58
159
69.3%
0.9%
15.3%
14.4%
0.43
0.00
0.43
0.38
0.49
Vanuatu 3
0.23
0.06
0.01
3.74
5.39
154
69.3%
1.0%
15.3%
14.4%
0.22
0.00
0.22
0.20
0.22
Viet Nam
2
87.38
24.82
2.57
2.14
3.63
104
59.0%
0.7%
26.5%
13.8%
19.31
53.31
72.62
85.08
85.49
TOTA
L 5,157
1,343
162
2.77
4.23
159
65.5%
1.8%
19.5%
13.3%
1,386
985
2,369
3,177
3,868
* In m
illions; † The total number of pregnancies is the sum of the number of live‐births, stillbirths, spontaneous and induced abortions; # Includes countries where P.falciaprum
and P.vivaxco‐exist; ¶
Stable
transm
ission: >= 0.1 autochthonous P. falciparum cases per 1,000 peo
ple per annum; u
nstable transm
ission <0.1 autochthonous P. falciparum cases per 1,000 people per annum [15]; Abbreviation: TFR:
Total Fertility Rate; TPR: Total Pregnancy Rate; UN: U
nited
Nations; W
OCBA: W
omen
of Childbearing Age (15‐49 years of age); PWAR: Pregnant Women at Risk; Note: The regional and total estim
ates for
TFR, Stillbirth rate, TPR and Annual PR are weighted m
ean
and is for illustration purposes only. The number of pregnancies at risk was derived by adding national estim
ates within regions and should not be
calculated using regional or global estim
ates for pregnancy rates; C
ontinent inform
ation: A
frica is den
oted by 1; A
sia by 2; O
ceania by 3; N
orth America 4 , South America 5 and Europe 6.
Chapter 3 Appen
dix
58
Table S2: Number of pregn
ancies total and by pregn
ancy outcomes at risk of P.falciparum and/or P.vivax malaria by malaria endemic countries by WHO regional office in
2007 (in m
illions)
Table S2
2007
Number of pregn
ancies at risk of malaria*§
Number of live‐births born to
pregn
ancies at risk of malaria* §
Number of stillbirths born to
pregn
ancies at risk of malaria* §
Number of miscarriages to
pregn
ancies at risk of malaria* §
Number of induced abortions to
pregn
ancies at risk of malaria* §
P. falciparum#
P.
vivax#
Any
species
P. falciparum#
P.
vivax#
Any
species
P. falciparum#
P.
vivax#
Any
species
P. falciparum#
P.
vivax#
Any
species
P. falciparum#
P.
vivax#
Any
species
Countries
Stable¶
Unstable¶
Overall
Overall
Overall
Stable¶
Unstable¶
Overall
Overall
Overall
Stable¶
Unstable¶
Overall
Overall
Overall
Stable¶
Unstable¶
Overall
Overall
Overall
Stable¶
Unstable¶
Overall
Overall
Overall
AFR
O 1
29.62
0.36
29.95
3.60
30.33
21.55
0.25
21.79
2.53
22.05
0.70
0.01
0.71
0.09
0.72
3.95
0.05
3.99
0.47
4.04
3.42
0.05
3.46
0.51
3.52
Angola
0.87
0.02
0.89
0.00
0.89
0.65
0.01
0.67
0.00
0.67
0.02
0.00
0.02
0.00
0.02
0.12
0.00
0.12
0.00
0.12
0.08
0.00
0.08
0.00
0.08
Ben
in
0.37
0.00
0.37
0.00
0.37
0.27
0.00
0.27
0.00
0.27
0.01
0.00
0.01
0.00
0.01
0.05
0.00
0.05
0.00
0.05
0.04
0.00
0.04
0.00
0.04
Botswana
0.03
0.00
0.03
0.00
0.03
0.02
0.00
0.02
0.00
0.02
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Burkina Faso
0.76
0.00
0.76
0.00
0.76
0.56
0.00
0.56
0.00
0.56
0.01
0.00
0.01
0.00
0.01
0.10
0.00
0.10
0.00
0.10
0.08
0.00
0.08
0.00
0.08
Burundi
0.37
0.00
0.37
0.00
0.37
0.26
0.00
0.26
0.00
0.26
0.01
0.00
0.01
0.00
0.01
0.05
0.00
0.05
0.00
0.05
0.05
0.00
0.05
0.00
0.05
Cam
eroon
0.67
0.00
0.67
0.00
0.67
0.50
0.00
0.50
0.00
0.50
0.01
0.00
0.01
0.00
0.01
0.09
0.00
0.09
0.00
0.09
0.06
0.00
0.06
0.00
0.06
Cape Verde
0.00
0.01
0.01
0.00
0.01
0.00
0.01
0.01
0.00
0.01
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Cen
tral African
Rep.
0.17
0.00
0.17
0.00
0.17
0.13
0.00
0.13
0.00
0.13
0.00
0.00
0.00
0.00
0.00
0.02
0.00
0.02
0.00
0.02
0.02
0.00
0.02
0.00
0.02
Chad
0.51
0.01
0.52
0.00
0.52
0.39
0.01
0.39
0.00
0.39
0.01
0.00
0.01
0.00
0.01
0.07
0.00
0.07
0.00
0.07
0.05
0.00
0.05
0.00
0.05
Comoros
0.03
0.00
0.03
0.03
0.03
0.02
0.00
0.02
0.02
0.02
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Congo
0.14
0.00
0.14
0.00
0.14
0.10
0.00
0.10
0.00
0.10
0.00
0.00
0.00
0.00
0.00
0.02
0.00
0.02
0.00
0.02
0.01
0.00
0.01
0.00
0.01
Dem
. Rep
. of Congo
3.28
0.00
3.28
0.00
3.28
2.46
0.00
2.46
0.00
2.46
0.09
0.00
0.09
0.00
0.09
0.44
0.00
0.44
0.00
0.44
0.30
0.00
0.30
0.00
0.30
Côte d'Ivoire
0.73
0.00
0.73
0.00
0.73
0.54
0.00
0.54
0.00
0.54
0.02
0.00
0.02
0.00
0.02
0.10
0.00
0.10
0.00
0.10
0.08
0.00
0.08
0.00
0.08
Equatorial Guinea
0.02
0.00
0.02
0.00
0.02
0.02
0.00
0.02
0.00
0.02
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Eritrea
0.17
0.05
0.22
0.24
0.24
0.12
0.03
0.15
0.17
0.17
0.00
0.00
0.00
0.00
0.00
0.02
0.01
0.03
0.03
0.03
0.02
0.01
0.03
0.03
0.03
Ethiopia
2.32
0.08
2.40
2.47
2.68
1.63
0.05
1.69
1.74
1.88
0.06
0.00
0.06
0.07
0.07
0.30
0.01
0.31
0.32
0.34
0.33
0.01
0.34
0.35
0.38
Gabon
0.04
0.00
0.04
0.00
0.04
0.03
0.00
0.03
0.00
0.03
0.00
0.00
0.00
0.00
0.00
0.01
0.00
0.01
0.00
0.01
0.00
0.00
0.00
0.00
0.00
Gam
bia
0.06
0.00
0.06
0.00
0.06
0.05
0.00
0.05
0.00
0.05
0.00
0.00
0.00
0.00
0.00
0.01
0.00
0.01
0.00
0.01
0.01
0.00
0.01
0.00
0.01
Ghana
0.82
0.00
0.82
0.00
0.82
0.61
0.00
0.61
0.00
0.61
0.01
0.00
0.01
0.00
0.01
0.12
0.00
0.12
0.00
0.12
0.08
0.00
0.08
0.00
0.08
Guinea
0.44
0.00
0.44
0.00
0.44
0.33
0.00
0.33
0.00
0.33
0.01
0.00
0.01
0.00
0.01
0.06
0.00
0.06
0.00
0.06
0.05
0.00
0.05
0.00
0.05
Guinea
‐Bissau
0.08
0.00
0.08
0.00
0.08
0.06
0.00
0.06
0.00
0.06
0.00
0.00
0.00
0.00
0.00
0.01
0.00
0.01
0.00
0.01
0.01
0.00
0.01
0.00
0.01
Ken
ya
1.26
0.01
1.26
0.00
1.26
0.88
0.01
0.89
0.00
0.89
0.04
0.00
0.04
0.00
0.04
0.15
0.00
0.15
0.00
0.15
0.18
0.00
0.18
0.00
0.18
Quantifying the number of pregnancies at risk of malaria
59
Table S2
2007
Number of pregn
ancies at risk of malaria*§
Number of live‐births born to
pregn
ancies at risk of malaria* §
Number of stillbirths born to
pregn
ancies at risk of malaria* §
Number of miscarriages to
pregn
ancies at risk of malaria* §
Number of induced abortions to
pregn
ancies at risk of malaria* §
P. falciparum#
P.
vivax#
Any
species
P. falciparum#
P.
vivax#
Any
species
P. falciparum#
P.
vivax#
Any
species
P. falciparum#
P.
vivax#
Any
species
P. falciparum#
P.
vivax#
Any
species
Countries
Stable¶
Unstable¶
Overall
Overall
Overall
Stable¶
Unstable¶
Overall
Overall
Overall
Stable¶
Unstable¶
Overall
Overall
Overall
Stable¶
Unstable¶
Overall
Overall
Overall
Stable¶
Unstable¶
Overall
Overall
Overall
Liberia
0.20
0.00
0.20
0.00
0.20
0.15
0.00
0.15
0.00
0.15
0.00
0.00
0.00
0.00
0.00
0.03
0.00
0.03
0.00
0.03
0.02
0.00
0.02
0.00
0.02
Madagascar
0.79
0.00
0.79
0.86
0.86
0.56
0.00
0.56
0.60
0.60
0.02
0.00
0.02
0.02
0.02
0.11
0.00
0.11
0.11
0.11
0.11
0.00
0.11
0.12
0.12
Malaw
i 0.68
0.00
0.68
0.00
0.68
0.48
0.00
0.48
0.00
0.48
0.02
0.00
0.02
0.00
0.02
0.09
0.00
0.09
0.00
0.09
0.10
0.00
0.10
0.00
0.10
Mali
0.71
0.03
0.71
0.00
0.71
0.52
0.02
0.52
0.00
0.52
0.01
0.00
0.01
0.00
0.01
0.10
0.00
0.10
0.00
0.10
0.07
0.00
0.07
0.00
0.07
Mauritania
0.04
0.02
0.05
0.00
0.05
0.03
0.01
0.04
0.00
0.04
0.00
0.00
0.00
0.00
0.00
0.01
0.00
0.01
0.00
0.01
0.00
0.00
0.01
0.00
0.01
Mayotte*
0.00
0.01
0.01
0.00
0.01
0.00
0.01
0.01
0.00
0.01
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Mozambique
1.04
0.00
1.04
0.00
1.04
0.73
0.00
0.73
0.00
0.73
0.02
0.00
0.02
0.00
0.02
0.14
0.00
0.14
0.00
0.14
0.15
0.00
0.15
0.00
0.15
Nam
ibia
0.04
0.01
0.06
0.00
0.06
0.03
0.01
0.04
0.00
0.04
0.00
0.00
0.00
0.00
0.00
0.01
0.00
0.01
0.00
0.01
0.01
0.00
0.01
0.00
0.01
Niger
0.78
0.04
0.82
0.00
0.82
0.58
0.03
0.61
0.00
0.61
0.02
0.00
0.02
0.00
0.02
0.10
0.00
0.11
0.00
0.11
0.08
0.00
0.08
0.00
0.08
Nigeria
6.47
0.00
6.47
0.00
6.47
4.78
0.00
4.78
0.00
4.78
0.15
0.00
0.15
0.00
0.15
0.88
0.00
0.88
0.00
0.88
0.67
0.00
0.67
0.00
0.67
Rwanda
0.31
0.00
0.31
0.00
0.31
0.22
0.00
0.22
0.00
0.22
0.01
0.00
0.01
0.00
0.01
0.04
0.00
0.04
0.00
0.04
0.04
0.00
0.04
0.00
0.04
S. Tome &Principe
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Senegal
0.47
0.00
0.47
0.00
0.47
0.35
0.00
0.35
0.00
0.35
0.01
0.00
0.01
0.00
0.01
0.06
0.00
0.06
0.00
0.06
0.05
0.00
0.05
0.00
0.05
Sierra Leo
ne
0.32
0.00
0.32
0.00
0.32
0.24
0.00
0.24
0.00
0.24
0.01
0.00
0.01
0.00
0.01
0.04
0.00
0.04
0.00
0.04
0.03
0.00
0.03
0.00
0.03
South Africa
0.10
0.09
0.19
0.00
0.19
0.07
0.06
0.13
0.00
0.13
0.00
0.00
0.00
0.00
0.00
0.01
0.01
0.03
0.00
0.03
0.02
0.01
0.03
0.00
0.03
Swaziland
0.01
0.00
0.01
0.00
0.01
0.01
0.00
0.01
0.00
0.01
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Tanzania
1.93
0.00
1.93
0.00
1.93
1.36
0.00
1.36
0.00
1.36
0.04
0.00
0.04
0.00
0.04
0.26
0.00
0.26
0.00
0.26
0.27
0.00
0.27
0.00
0.27
Togo
0.24
0.00
0.24
0.00
0.24
0.18
0.00
0.18
0.00
0.18
0.00
0.00
0.00
0.00
0.00
0.03
0.00
0.03
0.00
0.03
0.03
0.00
0.03
0.00
0.03
Uganda
1.53
0.00
1.53
0.00
1.53
1.07
0.00
1.07
0.00
1.07
0.04
0.00
0.04
0.00
0.04
0.20
0.00
0.20
0.00
0.20
0.21
0.00
0.21
0.00
0.21
Zambia
0.57
0.00
0.57
0.00
0.57
0.40
0.00
0.40
0.00
0.40
0.01
0.00
0.01
0.00
0.01
0.08
0.00
0.08
0.00
0.08
0.08
0.00
0.08
0.00
0.08
Zimbabwe
0.25
0.00
0.25
0.00
0.25
0.17
0.00
0.17
0.00
0.17
0.00
0.00
0.00
0.00
0.00
0.03
0.00
0.03
0.00
0.03
0.03
0.00
0.03
0.00
0.03
Chapter 3 Appen
dix
60
Table S2
2007
Number of pregn
ancies at risk of malaria*§
Number of live‐births born to
pregn
ancies at risk of malaria* §
Number of stillbirths born to
pregn
ancies at risk of malaria* §
Number of miscarriages to
pregn
ancies at risk of malaria* §
Number of induced abortions to
pregn
ancies at risk of malaria* §
P. falciparum#
P.
vivax#
Any
species
P. falciparum#
P.
vivax#
Any
species
P. falciparum#
P.
vivax#
Any
species
P. falciparum#
P.
vivax#
Any
species
P. falciparum#
P.
vivax#
Any
species
Countries
Stable¶
Unstable¶
Overall
Overall
Overall
Stable¶
Unstable¶
Overall
Overall
Overall
Stable¶
Unstable¶
Overall
Overall
Overall
Stable¶
Unstable¶
Overall
Overall
Overall
Stable¶
Unstable¶
Overall
Overall
Overall
EMRO
4.01
4.07
8.05
9.79
12.51
2.78
2.79
5.55
6.71
8.60
0.13
0.12
0.25
0.25
0.34
0.49
0.50
0.99
1.25
1.58
0.62
0.66
1.27
1.58
1.99
Afghanistan 2
0.29
0.80
1.09
1.06
1.30
0.20
0.55
0.74
0.72
0.89
0.01
0.03
0.03
0.03
0.04
0.04
0.10
0.13
0.13
0.16
0.05
0.13
0.18
0.17
0.21
Djibouti 1
0.00
0.02
0.02
0.03
0.03
0.00
0.01
0.01
0.02
0.02
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Iran
2
0.00
0.07
0.07
1.18
1.18
0.00
0.05
0.05
0.80
0.81
0.00
0.00
0.00
0.01
0.01
0.00
0.01
0.01
0.17
0.17
0.00
0.01
0.01
0.19
0.19
Iraq
2
0.00
0.00
0.00
0.40
0.40
0.00
0.00
0.00
0.27
0.27
0.00
0.00
0.00
0.01
0.01
0.00
0.00
0.00
0.05
0.05
0.00
0.00
0.00
0.06
0.06
Oman
2
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Pakistan 2
1.14
2.53
3.67
5.77
5.77
0.78
1.73
2.51
3.94
3.94
0.03
0.07
0.11
0.17
0.17
0.14
0.31
0.45
0.71
0.71
0.19
0.41
0.60
0.95
0.95
Saudi A
rabia 2
0.02
0.04
0.06
0.44
0.48
0.02
0.03
0.05
0.30
0.33
0.00
0.00
0.00
0.00
0.00
0.00
0.01
0.01
0.06
0.07
0.00
0.01
0.01
0.07
0.07
Somalia 1
0.50
0.03
0.50
0.50
0.50
0.35
0.02
0.35
0.35
0.35
0.02
0.00
0.02
0.02
0.02
0.06
0.00
0.06
0.06
0.06
0.07
0.00
0.07
0.07
0.07
Sudan
1
1.22
0.29
1.50
0.12
1.51
0.85
0.20
1.05
0.09
1.05
0.05
0.01
0.06
0.00
0.06
0.14
0.03
0.17
0.01
0.17
0.18
0.04
0.22
0.02
0.22
Syrian
Arab Rep
. 2
0.00
0.00
0.00
0.20
0.20
0.00
0.00
0.00
0.14
0.14
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.03
0.03
0.00
0.00
0.00
0.03
0.03
Yemen
2
0.83
0.30
1.13
0.10
1.14
0.58
0.21
0.79
0.07
0.79
0.02
0.01
0.03
0.00
0.03
0.11
0.04
0.15
0.01
0.15
0.13
0.05
0.17
0.02
0.17
EURO
0.00
0.11
0.11
0.62
0.64
0.00
0.08
0.08
0.43
0.44
0.00
0.00
0.00
0.01
0.01
0.00
0.02
0.02
0.09
0.09
0.00
0.02
0.02
0.10
0.10
Arm
enia 6
0.00
0.00
0.00
0.01
0.01
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Azerbaijan 6
0.00
0.00
0.00
0.01
0.01
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Geo
rgia 6
0.00
0.00
0.00
0.01
0.01
0.00
0.00
0.00
0.01
0.01
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Kyrgyzstan 2
0.00
0.03
0.03
0.05
0.05
0.00
0.02
0.02
0.03
0.03
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.01
0.01
0.00
0.01
0.01
0.01
0.01
Tajikistan 2
0.00
0.08
0.08
0.16
0.18
0.00
0.05
0.05
0.11
0.12
0.00
0.00
0.00
0.00
0.00
0.00
0.01
0.01
0.02
0.02
0.00
0.01
0.01
0.03
0.03
Turkey 6
0.00
0.00
0.00
0.34
0.34
0.00
0.00
0.00
0.23
0.23
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.05
0.05
0.00
0.00
0.00
0.05
0.05
Turkmen
istan 2
0.00
0.00
0.00
0.04
0.04
0.00
0.00
0.00
0.03
0.03
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.01
0.01
0.00
0.00
0.00
0.01
0.01
Uzbekistan 2
0.00
0.00
0.00
0.01
0.01
0.00
0.00
0.00
0.01
0.01
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Quantifying the number of pregnancies at risk of malaria
61
Table S2
2007
Number of pregn
ancies at risk of malaria*§
Number of live‐births born to
pregn
ancies at risk of malaria* §
Number of stillbirths born to
pregn
ancies at risk of malaria* §
Number of miscarriages to
pregn
ancies at risk of malaria* §
Number of induced abortions to
pregn
ancies at risk of malaria* §
P. falciparum#
P.
vivax#
Any
species
P. falciparum#
P.
vivax#
Any
species
P. falciparum#
P.
vivax#
Any
species
P. falciparum#
P.
vivax#
Any
species
P. falciparum#
P.
vivax#
Any
species
Countries
Stable¶
Unstable¶
Overall
Overall
Overall
Stable¶
Unstable¶
Overall
Overall
Overall
Stable¶
Unstable¶
Overall
Overall
Overall
Stable¶
Unstable¶
Overall
Overall
Overall
Stable¶
Unstable¶
Overall
Overall
Overall
AMRO
1.36
1.60
2.96
2.90
4.32
0.84
1.01
1.85
1.84
2.73
0.02
0.02
0.03
0.03
0.04
0.18
0.22
0.40
0.40
0.60
0.32
0.35
0.67
0.62
0.95
Argen
tina 5
0.00
0.00
0.00
0.08
0.08
0.00
0.00
0.00
0.05
0.05
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.01
0.01
0.00
0.00
0.00
0.02
0.02
Belize 4
0.00
0.01
0.01
0.01
0.01
0.00
0.00
0.00
0.01
0.01
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Bolivia 5
0.01
0.10
0.11
0.16
0.16
0.01
0.06
0.07
0.10
0.10
0.00
0.00
0.00
0.00
0.00
0.00
0.01
0.02
0.02
0.02
0.00
0.02
0.03
0.04
0.04
Brazil 5
0.37
0.48
0.84
0.66
0.98
0.23
0.29
0.52
0.41
0.61
0.00
0.00
0.01
0.01
0.01
0.05
0.07
0.12
0.09
0.14
0.09
0.11
0.20
0.15
0.23
Colombia 5
0.15
0.22
0.37
0.49
0.61
0.09
0.14
0.23
0.30
0.37
0.00
0.00
0.00
0.00
0.00
0.02
0.03
0.05
0.07
0.08
0.03
0.05
0.09
0.11
0.14
Costa Rica 4
0.00
0.00
0.00
0.02
0.02
0.00
0.00
0.00
0.02
0.02
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Dominican
Rep
ublic 4
0.05
0.10
0.15
0.00
0.15
0.03
0.06
0.09
0.00
0.09
0.00
0.00
0.00
0.00
0.00
0.01
0.01
0.02
0.00
0.02
0.01
0.02
0.04
0.00
0.04
Ecuador 5
0.13
0.05
0.18
0.11
0.18
0.08
0.03
0.11
0.07
0.11
0.00
0.00
0.00
0.00
0.00
0.02
0.01
0.02
0.02
0.02
0.03
0.01
0.04
0.03
0.04
El Salvador 4
0.00
0.00
0.00
0.03
0.03
0.00
0.00
0.00
0.02
0.02
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.01
0.01
Guatem
ala 4
0.01
0.00
0.01
0.01
0.01
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Guyana 5
0.04
0.23
0.27
0.17
0.31
0.03
0.15
0.18
0.12
0.21
0.00
0.01
0.01
0.00
0.01
0.01
0.03
0.03
0.02
0.04
0.01
0.04
0.05
0.03
0.05
Haiti 4
0.00
0.01
0.02
0.02
0.02
0.00
0.01
0.01
0.01
0.01
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Honduras 4
0.37
0.00
0.37
0.00
0.37
0.22
0.00
0.22
0.00
0.22
0.01
0.00
0.01
0.00
0.01
0.05
0.00
0.05
0.00
0.05
0.09
0.00
0.09
0.00
0.09
Mexico 4
0.03
0.09
0.13
0.20
0.23
0.02
0.06
0.08
0.13
0.16
0.00
0.00
0.00
0.00
0.00
0.00
0.01
0.02
0.03
0.03
0.01
0.02
0.02
0.03
0.04
Nicaragua 4
0.00
0.00
0.00
0.41
0.41
0.00
0.00
0.00
0.28
0.28
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.06
0.06
0.00
0.00
0.00
0.07
0.07
Panam
a 4
0.05
0.07
0.11
0.07
0.15
0.03
0.04
0.08
0.05
0.10
0.00
0.00
0.00
0.00
0.00
0.01
0.01
0.02
0.01
0.02
0.01
0.01
0.02
0.01
0.03
Paraguay 5
0.03
0.00
0.03
0.01
0.03
0.02
0.00
0.02
0.00
0.02
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.01
Peru 5
0.00
0.00
0.00
0.05
0.05
0.00
0.00
0.00
0.03
0.03
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.01
0.01
0.00
0.00
0.00
0.01
0.01
Surinam
e 5
0.12
0.05
0.17
0.25
0.26
0.07
0.03
0.11
0.16
0.16
0.00
0.00
0.00
0.00
0.00
0.02
0.01
0.02
0.03
0.04
0.03
0.01
0.04
0.06
0.06
Ven
ezuela 5
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Fren
ch Guiana 5
0.01
0.20
0.20
0.13
0.25
0.00
0.12
0.12
0.08
0.15
0.00
0.00
0.00
0.00
0.00
0.00
0.03
0.03
0.02
0.03
0.00
0.05
0.05
0.03
0.06
Chapter 3 Appen
dix
62
Table S2
2007
Number of pregn
ancies at risk of malaria*§
Number of live‐births born to
pregn
ancies at risk of malaria* §
Number of stillbirths born to
pregn
ancies at risk of malaria* §
Number of miscarriages to
pregn
ancies at risk of malaria* §
Number of induced abortions to
pregn
ancies at risk of malaria* §
P. falciparum#
P.
vivax#
Any
species
P. falciparum#
P.
vivax#
Any
species
P. falciparum#
P.
vivax#
Any
species
P. falciparum#
P.
vivax#
Any
species
P. falciparum#
P.
vivax#
Any
species
Countries
Stable¶
Unstable¶
Overall
Overall
Overall
Stable¶
Unstable¶
Overall
Overall
Overall
Stable¶
Unstable¶
Overall
Overall
Overall
Stable¶
Unstable¶
Overall
Overall
Overall
Stable¶
Unstable¶
Overall
Overall
Overall
SEARO
16.64
20.95
37.57
47.97
48.17
11.02
14.01 25.02
31.88
32.00
0.31
0.40
0.71
0.91
0.91
2.21
2.78
4.98
6.35
6.38
3.11
3.76
6.86
8.84
8.88
Bangladesh 2
0.46
1.47
1.93
4.72
4.72
0.32
1.00
1.32
3.22
3.22
0.01
0.04
0.05
0.13
0.13
0.06
0.19
0.24
0.59
0.59
0.08
0.24
0.32
0.77
0.77
Bhutan 2
0.02
0.01
0.02
0.02
0.02
0.01
0.01
0.01
0.01
0.01
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Burm
a 2
1.23
0.05
1.29
1.36
1.40
0.73
0.03
0.76
0.80
0.83
0.02
0.00
0.02
0.02
0.02
0.16
0.01
0.17
0.18
0.18
0.33
0.01
0.34
0.36
0.37
India 2
12.29
15.88
28.16
32.87
32.87
8.39
10.84 19.24
22.45
22.45
0.26
0.33
0.58
0.68
0.68
1.62
2.10
3.72
4.35
4.35
2.01
2.60
4.62
5.39
5.39
Indonesia 2
2.00
2.39
4.40
6.30
6.36
1.18
1.41
2.60
3.72
3.75
0.01
0.02
0.03
0.05
0.05
0.28
0.33
0.60
0.86
0.87
0.53
0.64
1.17
1.67
1.69
Rep
ublic of Korea 2
0.00
0.00
0.00
0.20
0.20
0.00
0.00
0.00
0.12
0.12
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.03
0.03
0.00
0.00
0.00
0.06
0.06
Nep
al 2
0.12
0.21
0.33
0.63
0.66
0.08
0.15
0.23
0.43
0.45
0.00
0.01
0.01
0.02
0.03
0.01
0.02
0.04
0.07
0.08
0.02
0.04
0.05
0.10
0.11
Sri Lanka 2
0.04
0.16
0.20
0.23
0.28
0.03
0.11
0.14
0.15
0.19
0.00
0.00
0.00
0.00
0.00
0.01
0.02
0.03
0.03
0.04
0.01
0.03
0.03
0.04
0.05
Thailand 2
0.41
0.76
1.18
1.60
1.60
0.24
0.45
0.70
0.94
0.94
0.00
0.00
0.01
0.01
0.01
0.06
0.11
0.16
0.22
0.22
0.11
0.20
0.31
0.42
0.42
Timor‐Leste 2
0.07
0.00
0.07
0.04
0.07
0.04
0.00
0.04
0.03
0.04
0.00
0.00
0.00
0.00
0.00
0.01
0.00
0.01
0.01
0.01
0.02
0.00
0.02
0.01
0.02
WPRO
3.11
3.52
6.62
28.03
29.27
1.84
2.06
3.91
16.08
16.81
0.03
0.03
0.06
0.35
0.36
0.42
0.48
0.90
3.66
3.83
0.81
0.94
1.76
7.93
8.27
Cam
bodia 2
0.44
0.11
0.55
0.52
0.60
0.26
0.06
0.32
0.31
0.35
0.01
0.00
0.01
0.01
0.01
0.06
0.01
0.07
0.07
0.08
0.12
0.03
0.15
0.14
0.16
China 2
0.41
0.48
0.89
21.25
21.70
0.23
0.27
0.50
12.06
12.32
0.01
0.01
0.01
0.30
0.30
0.05
0.06
0.11
2.73
2.79
0.12
0.14
0.26
6.15
6.28
Lao 2
0.21
0.00
0.21
0.09
0.21
0.13
0.00
0.13
0.06
0.13
0.00
0.00
0.00
0.00
0.00
0.03
0.00
0.03
0.01
0.03
0.06
0.00
0.06
0.02
0.06
Malaysia 2
0.21
0.54
0.75
0.34
0.76
0.12
0.32
0.44
0.20
0.45
0.00
0.00
0.00
0.00
0.00
0.03
0.08
0.11
0.05
0.11
0.06
0.14
0.20
0.09
0.20
Papua New
Guinea 3
0.16
0.00
0.16
0.15
0.17
0.11
0.00
0.11
0.11
0.12
0.00
0.00
0.00
0.00
0.00
0.02
0.00
0.02
0.02
0.02
0.02
0.00
0.02
0.02
0.03
Philippines 2
1.08
0.82
1.90
3.15
3.28
0.64
0.48
1.12
1.86
1.94
0.01
0.01
0.01
0.02
0.02
0.15
0.11
0.26
0.43
0.45
0.29
0.22
0.50
0.84
0.87
Solomon Islands 3
0.02
0.00
0.02
0.01
0.02
0.01
0.00
0.01
0.01
0.01
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Vanuatu 3
0.01
0.00
0.01
0.01
0.01
0.01
0.00
0.01
0.01
0.01
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Viet Nam
2
0.57
1.57
2.14
2.51
2.52
0.34
0.93
1.26
1.48
1.48
0.00
0.01
0.01
0.02
0.02
0.08
0.22
0.29
0.34
0.35
0.15
0.42
0.57
0.67
0.67
TOTA
L 54.74
30.61
85.28
92.90
125.24 38.03
20.20 58.19
59.47
82.63
1.18
0.58
1.76
1.63
2.38
7.25
4.04
11.29
12.22
16.51
8.27
5.78
14.04 19.58 23.71
*In m
illions; # Includes countries where P.falciaprum and P.vivax co‐exist; ¶
Stable transm
ission: >= 0.1 autochthonous P. falciparum cases per 1,000 peo
ple per annum; u
nstable transm
ission <0.1 autochthonous P. falciparum cases per
1,000 peo
ple per annum; §
The number of pregnant women
at risk was then
derived
as follow: P
regA
R= PARxFWOCBAxAnnualised
_PR, w
here: FWOCBA=fraction of W
OCBA in
2007; A
nnualised
_PR = the average number of pregnancies
per year; Note: The regional and total estim
ates for TFR, Stillbirth rate, TPR and Annual PR are weighted m
ean and is for illustration purposes only. The number of pregnancies at risk was derived
by adding national estim
ates within regions
and should not be calculated using regional or global estim
ates for pregnan
cy rates; C
ontinen
t inform
ation: A
frica is den
oted by 1; A
sia by 2; O
ceania by3; N
orth America 4 , South America 5 and Europe 6.
Quantifying the number of pregnancies at risk of malaria
63
Table S3: Total population, n
umber of pregnan
cies and number of live‐births born to pregnan
cies in m
alaria endemic countries by continent in 2007 (in m
illions)
Continents
Demograp
hic data*
Total population at risk
(% of the population in
Malaria Endem
ic Countries at risk)
(United Nations National Population Estim
ates)
Malaria Endem
ic Countries (M
EC)
P. falciparum transm
ission#
P. vivax
transm
ission#
Any species
Number
of
Malaria
Endem
ic
Countrie
s
Total
Popul
ation
(both
sexes
)*
WOCBA
*
Total
number of
Pregnancies
*¥
TPR§
Pregnancy
rate per
1000
WOCBAs§
Percentage pregnan
cies ending in§:
Stable
transm
ission¶
Unstable
transm
ission
¶
Overall
Overall
Overall
Live‐
births
Still‐
births
Spontan
e‐ous
Abortion
s
Induced
Abortion
s
Africa1
46
803
190
38.5
7.12
198
72.1%
2.4%
13.3%
12.2%
637.6 (79.4)
16.2 (2.0)
652.8 (81.2)
85.6 (10.7)
660.8 (82.2)
Asia2
28
3,726
984
105.8
3.76
135
62.5%
1.6%
13.1%
22.8%
702.2 (18.8)
918.7 (24.7)
1620.5 (43.5)
2975.3 (79.9)
3,048.2 (81.8)
Oceania
3
3
7
2
0.3
5.46
156
69.3%
1.2%
14.2%
15.3%
4.8 (67.5)
0.0 (0.0)
4.8 (67.5)
4.5 (63.4)
5.2 (73.3)
North America4
10
167
45
4.9
3.84
127
66.5%
0.8%
14.3%
18.4%
14.4 (8.6)
13.0 (7.8)
27.4 (16.4)
30.1 (18.1)
50.6 (30.3)
South America
5
11
363
98
10.6
3.79
122
61.8%
0.8%
13.9%
23.5%
26.8 (7.4)
37.2 (10.2)
64.0 (17.6)
66.1 (18.2)
87.6 (24.1)
Europe6
4
91
25
2.1
2.95
70
69.3%
0.9%
14.5%
15.3%
0.0 (0.0)
0.0 (0.0)
0.0 (0.0)
15.3 (16.8)
15.3 (16.8)
Global
102
5,157
1,343
162.3
4.23
159
65.5%
1.8%
13.3%
19.5%
1385.8 (26.9)
985.1 (19.1)
2369.4 (45.9)
3176.9 (61.6)
3,867.6 (75.0)
Continen
ts
Number of pregn
ancies at risk f malaria (column %)
Number of live‐births born to pregn
ancies at risk of malaria (column %)
P. falciparum transm
ission#
P. vivax
transm
ission#
Any species
P. falciparum transm
ission#
P. vivax
transm
ission#
Any species
Stable
transm
ission¶
Unstable
transm
ission¶
Overall
Overall
Overall
Stable
transm
ission¶
Unstable
transm
ission¶
Overall
Overall
Overall
Africa1
31.3(57.3)
0.7 (2.3)
32.0(37.5)
4.3(4.6)
32.4(25.8)
22.8(59.8)
0.5(2.4)
23.2(39.9)
3.0(5.0)
23.5 (28.4)
Asia2
21.9(39.9)
28.3 (92.5)
50.2(58.8)
85.2(91.7)
88.0(70.3)
14.3(37.6)
18.7(92.6)
33.0(56.7)
54.3(91.2)
56.0 (67.8)
Oceania
3
0.2(0.3)
0.0 (0.0)
0.2(0.2)
0.2(0.2)
0.2(0.2)
0.1(0.3)
0.0(0.0)
0.1(0.2)
0.1(0.2)
0.1 (0.2)
North America4
0.6(1.0)
0.5 (1.6)
1.1(1.2)
0.9(1.0)
1.7(1.4)
0.4(0.9)
0.3(1.6)
0.7(1.2)
0.6(1.0)
1.1 (1.4)
South America
5
0.8(1.4)
1.1 (3.6)
1.9(2.2)
2.0(2.1)
2.6(2.1)
0.5(1.3)
0.7(3.4)
1.2(2.0)
1.2(2.1)
1.6 (1.9)
Europe6
0.0(0.0)
0.0 (0.0)
0.0(0.0)
0.4(0.4)
0.4(0.3)
0.0(0.0)
0.0(0.0)
0.0(0.0)
0.2(0.4)
0.2 (0.3)
Global
54.7
30.6
85.3
92.9
125.2
38.0
20.2
58.2
59.5
82.6
* Source: United
Nations Developmen
t Program
; ¥ The total number of pregnancies is the sum of the number of live‐births, stillbirths, spontaneo
us and induced abortions; § The total pregnancy rate (TPR) and the annual pregnancy
rate per 1000 W
OCBAs are weighted m
eans per region and is for illustration purposes only. The number of pregnancies at risk was derived
directly as the sum of the national estim
ates within each region and globally. They differ
slightly from sim
ilar estimates obtained
indirectly by use of the weighted regional or global estim
ates for pregnancy rates; # Includes countries where P. falciparum and P. vivax co‐exist; ¶
Stable transm
ission: >= 1 autochthonous P.
falciparum cases per 10,000 peo
ple per annum; Unstable transm
ission <1 autochthonous P. falciparum cases per 10,000 peo
ple per annum; A
bbreviation: N. A
merica: North America; S. A
merica: South America; M
EC: M
alaria
Endem
ic Countries; TPR: Total Pregnancy Rate; W
OCBA: W
omen
of Childbearing Age (15‐49 years of age). Continent inform
ation: Africa is den
oted by 1; A
sia by 2; O
ceania by3; N
orth America 4 , South America 5 and Europe 6
Regions inform
ation: Th
e Africa and Europe regions are defined
as the respective continen
ts; A
mericas is defined
as ‘North America’ plus 'South America' countries and Asia region defined
as Asia‐Pacific and Oceania combined
64
65
Chapter4
PregnancyExposureRegistriesforAssessingAntimalarialDrugSafetyinPregnancyinMalaria‐EndemicCountries
StephanieDellicour1,FeikoO.terKuile1,2,AndyStergachis3
1 Child and Reproductive Health Group, Liverpool School of Tropical Medicine, Liverpool,
United Kingdom, 2 Division of Infectious Diseases, Tropical Medicine & AIDS, Academic
Medical Centre Amsterdam, Amsterdam, The Netherlands, 3 Departments of
Epidemiology and Global Health, School of Public Health & Community Medicine,
University of Washington, Seattle, Washington, United States of America
PLoS Medicine 2008, 5:9
Copyright© This is an open‐access article distributed under the terms of the Creative
Commons Attribution License, which permits unrestricted use, distribution, and
reproduction in any medium, provided the original author and source are credited
Chapter 4
66
SummaryPoints There is an urgent need to develop targeted pharmacovigilance systems to assess the safety
of antimalarials in early pregnancy.
The artemisinins are effective antimalarials increasingly deployed in malaria‐endemic
countries; however, they have been shown to be embryo‐ toxic in animal models, and their
safety in early human pregnancies remains uncertain.
Modelling suggests that the probability an embryo will encounter artemisinins during the
critical six‐ week period (at week four to week ten of gestation) through accidental exposure is
12% for areas where adults receive on average one treatment with three days of artemisinin‐
based combination therapy per year.
Most of the approaches used in industrialised countries to evaluate a drug’s embryo‐foetal
toxicity have limited application in resource‐poor countries. Establishing an international
antimalarial pregnancy exposure registry would enable a targeted prospective
pharmacovigilance approach and timely assessment of the risk–benefit profile of
antimalarials.
Here we discuss methodological considerations for the systematic prospective assessment of
pregnancy outcomes and congenital malformations in women exposed to antimalarials early
in pregnancy, including approaches to capture drug exposure information in resource‐poor
settings, choice of comparison groups, and sample size considerations.
Antimalarial pregnancy exposure registries
67
Because pregnant women are routinely excluded from pre‐ licensure clinical trials for fear of harming
the mother or the developing foetus [1], most drugs are marketed with limited information on their
safety during pregnancy and therefore are not recommended for use by pregnant women. Yet drugs
are widely used by pregnant women, and medication often cannot be avoided in chronic diseases such
as epilepsy and HIV or other acute illness that harm the mother and the unborn child if left untreated.
Passive mechanisms of spontaneous reporting of adverse drug effects are inadequate for detecting
drug‐induced foetal risks or lack of such risks [2]. The US Food and Drug Administration and the
European Medicine Agency recommend active surveillance, such as the use of pregnancy exposure
registries (PERs), for products that are likely to be used during pregnancy or by women of
childbearing age (WOCBAs), particularly if there have been case reports of adverse pregnancy
outcome following exposure, if drugs in the same pharmacological class are known to pose risk
during pregnancy, or if pre‐clinical animal data suggest potential teratogenic risk [3,4]. In
industrialised countries, this information can be derived from medical records and automated
databases, including medical or pharmacy insurance claims. Such approaches are challenging in
developing countries where resources for routine pharmacovigilance are rare and automated data
sources generally do not exist [5–8]. Thus, nearly all developing countries rely on drug safety data
from industrialised countries. However, there are often no or limited safety data in pregnancy for
drugs targeting tropical diseases, as these are not widely used in the countries with more robust
pharmacovigilance systems [9].
Antimalarials are a good example [9]. Malaria can have devastating consequences for the mother and
foetus [10,11], and pregnant women require prompt treatment with safe and effective antimalarial
drugs when infected. The artemisinins are among the most effective and rapidly acting antimalarials
to date, providing life‐ saving benefits to children, adults, and pregnant women [12]. The limited
information regarding their safety is reassuring [13], and the World Health Organization (WHO) now
recommends the use of artemisinin combination therapies (ACTs) in the second and third but not yet
in the first trimester (unless alternatives are not available) [12], as uncertainty remains about their
safety in early pregnancy (Box 1) [14–16]. ACTs are rapidly being rolled out and may soon become
among the most widely used antimalarial drugs. Because there are no specific risk management
precautions to exclude WOCBAs from using ACTs, the potential for inadvertent exposure to
artemisinins early in pregnancy is high and in many cases unavoidable (Figure 1). Health care
providers, pregnant women, and policy makers urgently need valid information to make informed
decisions about the risks and benefits of ACTs for WOCBAs.
This paper describes the use of PERs as a targeted pharmacovigilance approach for assessing the safety
of antimalarial drugs used during early pregnancy in resource‐constrained malaria‐endemic countries.
AntimalarialPregnancyExposureRegistries
PERs are the most common approach used to monitor drug safety in pregnancy and provide
reassurance on the potential risk associated with certain drugs. They can serve both to generate
hypotheses and to evaluate suspected risks or risk factors that may have been identified during pre‐ or
post‐marketing phases [2]. In industrialised countries, 32 PERs are registered with the Food and Drug
Administration [17]. There is some variation in design, but they all use prospective approaches and
identify and follow exposed women until the end of pregnancy (i.e., before the outcome is known).
Chapter 4
68
The systematic prospective ascertainment of pregnancy outcomes has several major advantages over
case‐control designs and passive surveillance. This design reduces selection bias (for example, due to
self‐reporting) and recall bias, has the potential to use standardised methods to assess outcome,
and—because of the availability of both numerator and denominators—allows calculations of risk
estimates that can then be compared against comparison groups or background population rates [2].
One other attractive feature of PERs is that they can be time‐limited and terminated once the target
sample size to rule out a pre‐defined risk is reached.
AssessmentofDrugExposureandRecordLinkage
The design for reliably capturing the occurrence and timing of inadvertent drug exposure to ACTs in
early pregnancy requires special consideration. Firstly, the critical period occurs around the time
when many women may not yet be aware of their pregnancy (our current understanding from animal
models of the mechanism of embryotoxicity of the artemisinins suggests that in humans the sensitive
drug exposure time window is between week four to week ten of gestation [Box 1]). Secondly,
retrospective determination of the precise timing of exposure is challenging since a typical treatment
course is short (three days). Another difficulty is the accurate assessment of the gestational age at
the time of exposure. Lastly, malaria treatment is often home‐based or unsupervised and
antimalarials can be obtained from a variety of providers, often over‐the‐ counter. In contrast,
antiretroviral and anti‐tuberculosis drugs are typically provided by formal health services, which are
more likely to keep records. Furthermore, exposures are often long‐ term and continuous, making it
easier to determine if and when a woman was exposed to antiretroviral or anti‐tuberculosis
medication than with the short course of antimalarials.
Although most exposures to artemisinins in early pregnancy will be unintentional, deliberate
exposures can occur where the benefit is perceived to outweigh the potential risk, as recommended
by WHO (such as for severe life‐threatening malaria) [12]. Either way, reliable ascertainment of drug
exposure will require record linkage. This can be done using prospective approaches by linking
datasets containing drug dispensing information (e.g., malaria treatment records from out‐patient
Box1.MechanismofArtemisininToxicityinEarlyPregnancyAnimal reproductive toxicology studies show that artemisinin derivatives all have embryo‐toxic effects at
low‐dose ranges in all species studied (i.e., mice, rat, rabbit, frog, and primate models) [31–34].The
embryo‐toxic mechanism is thought to occur through depletion of embryonic erythroblasts (primitive
erythrocytes), which is associated with severe anaemia leading to cell damage and death due to hypoxia
[35]. In humans, the most sensitive time window may be between week four and week ten, when
erythroblasts circulate and have not yet been fully replaced by definitive erythrocytes [36]. In addition to
the window of sensitivity, the duration of exposure is also important. Rodents have a synchronous clonal
expansion of metabolically active erythroblasts, making them particularly vulnerable during a three‐ to
four‐day window early in pregnancy. In primates (and most likely also in humans), this may not be the case,
as different generations of erythroblasts co‐exist and are progressively replaced by definitive erythrocytes
over a period of weeks [35]. In cynomolgus monkeys, no embryo lethality or malformations were observed
with three‐day exposures (the typical duration of treatment with ACTs) or with seven‐day exposures
[31,32,35,36]. The predictive value of the animal models for humans is unclear, particularly because the
duration of daily exposure is likely to be short (hours) as the artemisinins are rapidly eliminated and limited
to three or at maximum seven days.
Antimalarial pregnancy exposure registries
69
departments) to datasets that capture newly identified pregnancies (e.g., from antenatal clinics or
demographic surveillance systems). This can determine whether a WOCBA might have been pregnant
at the time of treatment. Alternatively, records of pregnant women can be linked retrospectively
with their earlier treatment records. A disadvantage of recruiting pregnant women rather than
WOCBAs is that miscarriages will be missed, as the pregnancy may not be sustained long enough for
women to attend antenatal care. The pregnant woman’s drug history should be taken to verify the
record linkage and to capture information on any drug use not dispensed through formal
pharmacies.
Figure 1. Probability that an Embryo Will Encounter Artemisinins Inadvertently During the Critical Six‐Week Period of Its Development (Week Four to Week Ten), According to the Average Number of ACT Treatments Received Per Year. doi:10.1371/journal.pmed.0050187.g001
In the figure, χ = number of treatments per year, t = embryo‐sensitive period in days (set as 42 days or six weeks), and p = period of treatment and persistence of drug (set as three days because ACTs are normally deployed as a three‐day regimen and artemisinins are eliminated within hours after each dose). The inadvertently exposed group will consist of women taking ACTs for confirmed malaria and for presumed malaria. It has been estimated that over 70% of malaria episodes in rural Africa and about 50% in urban areas are self‐treated without consulting trained professionals [39]. Thus, many of these will be presumptive treatments without involvement of the formal health services, diagnostic confirmation of malaria, or screening for potential pregnancy. Even if more women seek treatment at health facilities with the deployment of more expensive ACTs and rapid diagnostic tests, antimalarials are often administered disregarding any diagnostic test. Studies in Africa indicated that between 30% and 50% of patients with a negative diagnostic test (microscopy or rapid diagnostic test) were still prescribed antimalarial drugs [40,41]. These proportions are likely to increase further when successful malaria control reduces malaria exposure. (Adapted from [16].)
PregnancyOutcomeAssessment
The primary outcome of interest is a decisive factor for the choice of study design, study population,
11.6%
21.9%
31.0%
39.1%
46.2%
52.6%
58.2%
63.1%
67.5%
71.4%
0%
10%
20%
30%
40%
50%
60%
70%
80%
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 7 7.5 8 8.5 9 9.5 10
Ris
k of
Exp
osur
e (%
)
Number of ACT treatment per year
Probability of exposure during embryo-sensitive period: 1-(1-x/365) (t+p) *
1.2%
2.4%
3.6%
4.8%
6.0%
7.1%
8.3%
9.4%
10.5%11.6%
0%1%2%3%4%5%6%7%8%9%
10%11%12%
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Chapter 4
70
and target data sources for outcome ascertainment and needs to be defined a priori. Although pre‐
approval animal reproductive toxicology studies have ambiguous predictive value for human
embryo‐foetal toxicity, due to variations in species‐specific effects [18], the current data from animal
models suggest that the effects are not species‐specific and that exposure early in the first trimester
might cause birth defects and/or early embryo/foetal death with subsequent miscarriages or foetal
resorption. Most of the existing PERs monitor all pregnancy outcomes (i.e., live births, still births, and
miscarriages), but the design and sample size calculation focus on capturing birth defects [19]. Foetal
resorption and early miscarriages are very difficult to assess reliably; most will go unnoticed clinically
as they occur before eight to nine weeks, with the majority occurring before three weeks [20]. Only
repeated pregnancy testing with a switch from positive to negative tests may suggest objectively
early loss of pregnancy [21]. This is unlikely to be feasible or culturally acceptable in many malaria‐
endemic countries, and the frequent use of pregnancy testing itself reduces the probability of
inadvertent exposures in that population. We may thus have to accept that early pregnancy loss
cannot be captured reliably in sufficient numbers, in contrast to later miscarriages and stillbirths.
For birth defects, the duration of follow‐up of the infant needs to be considered carefully, since only
about half of major structural and functional defects in children can be detected or classified at birth
[22,23]. The prevalence also varies with the specific defect inclusion and exclusion criteria and
whether the case definition includes developmental, functional, or other types of congenital
disorders (e.g., non‐structural genetic disorders) [24]. Assessment of congenital malformation
requires careful examination by dedicated staff trained to examine newborns using a standard tool
and scoring system. Suspected birth defects could be reviewed by a centralised committee that
included dysmorphologists and other specialists (e.g., using digital photographs). Complimentary
visiting specialists could study additional outcomes such as cardiovascular and neuro‐developmental
defects and other potential long‐term effects in a selected sample later in infancy.
ComparisonGroups
Assessing the teratogenic potential of a drug requires comparison of the frequency of birth defects
against other groups to put a signal into context. These comparison groups can be external (i.e., from
peripheral sources) or internal (i.e., generated from within the same study or system). External
population data from national health statistics centres and/ or birth defects surveillance systems are
commonly used as sources to calculate background event rates. This type of external comparison data
is not currently available in most malaria‐ endemic countries. Furthermore, these comparisons need
to be interpreted with caution as many confounding factors or potential effect modifiers of risk may
differ from the exposed group of interest [2,25].
Internal comparison groups can consist of women with the same conditions who are unexposed to
drugs (in which case the possibility of confounding by indication should be taken into account), exposed
to a different drug with established safety, or exposed to the same investigational drug, but only outside
the critical period (e.g., in the second or third trimesters).
Antimalarial pregnancy exposure registries
71
0
5,000
10,000
15,000
20,000
25,000
30,000
35,000
40,000
45,000
50,000
1 1.5 2 2.5 3 3.5 4 4.5 5
Num
ber E
xpos
ed
Miscarriage 15% Stillbirth 3% Major Malformations 2% Specific Malformations 0.1%
Relative Risk
0500
1,0001,5002,0002,5003,0003,5004,0004,5005,000
1 1.5 2 2.5 3 3.5 4 4.5 5
Figure 2. Sample Size Calculation for Pregnancy Exposure Registry by Defect Frequency and Detectable Difference. doi:10.1371/journal.pmed.0050187.g002
Exposed to comparison group ratio = 1:1, power = 80%, and one‐sided α = 0.05. Based on the formula for cohort design described
in Strom’s Pharmacoepidemiology [25]: N = 1/[p(1 − R)]2 ×[Z √((1 + 1/k)U(1 − U)) + Z√(pR(1 − Rp) + (P(1 − P))/k)]2 where p is the incidence of disease in unexposed; R is the minimum relative risk to detect; k is the ratio of unexposed controls to exposed, and U = (Kp + pR)/(k + 1).
SampleSizeCalculationThe main determinants of sample size are the degree of the teratogenic effect to be excluded
(relative risk) and the expected frequency of the endpoint of interest in the non‐exposed group
(Figure 2). A third factor is the type and number of potential controls. For example, with an
exposed/unexposed ratio of 1:4, approximately 522 exposed women and 2,090 unexposed women
are needed to exclude a 2‐fold increase in major malformations detectable at birth when the
predicted rate in the comparison group is 2% (power 80%, alpha 0.05). This would be 10,748/42,992
exposed/unexposed women for birth defects that occur at a frequency of one in 1,000 (such as cleft
lip/palate). Such numbers will only be achievable using several sentinel sites over several years. The
sample sizes will also need to account for loss to follow‐ up and the fact that not all births can be
examined for birth defects (e.g., foetal loss with discarding of expelled foetus prior to examination by
study staff). The rate of recruitment will depend on the likelihood of accidental exposure. This
depends on the fertility rate and frequency of drug exposure in the population. For example, in areas
where pregnancy testing is not available, the average number of ACT treatments is one per woman
per year, and the total fertility rate is 5.5, the probability is only 2.5% (or one in 40 women) (Box 2).
Chapter 4
72
ConcomitantMedicationAlthough the registry could be set up initially to address the specific question of the safety of
antimalarials in pregnancy, it is essential to capture concomitant diseases and medications such as
antiretrovirals because of potential drug interactions, confounding, and effect modification. As such,
these additional data could contribute to PERs for other diseases such as the Antiretroviral Pregnancy
Registry [26,27].
DataSourcesThere are many methodological challenges to designing PERs for antimalarials, including those
common to most pharmacovigilance methods in resource‐poor countries [28,29]. The specialised
nature of the reliable assessment of drug exposure and congenital malformations is not easily
achievable from routine pharmacovigilance surveillance systems (where they exist). Such an effort
will require dedicated sentinel sites that are capable of following WOCBAs and linking antenatal care
records with treatment records, such as sites with demographic health surveillance systems or sites
with captive populations where health care is provided centrally and well recorded (e.g., industrial
and agricultural estates or long‐term refugee camps).
While the primary source of information is prospective and observational, data from clinical trials
and other studies involving pregnant women, and retrospective case series (i.e., pregnancies with a
known outcome at the time of reporting) could be included as secondary data and analysed
separately, as is currently done with some of the existing PERs [30].
Box 2: Probability that a Woman of Childbearing Age Treated for Malaria at an Out‐Patient Clinic Had an
Undetected Pregnancy of Four to Ten Weeks Gestation
An approximation of the probability that a WOCBA attending an out‐patient clinic has an early pregnancy can be
indirectly estimated from published total fertility rates. For sub‐Saharan Africa, total fertility rate was 5.5 in
2004 [37] and is defined as the number of live‐born children an average women would have, assuming that
she lives her full reproductive lifetime of 35 years (1,820 weeks, from 15 to 49 years). The total pregnancy
rate (6.7) was then calculated as the total fertility rate (5.5) multiplied by a factor of 1.22 (1 / [1.0 − 0.15 −
0.03]) to take into account 15% pregnancy loss due to miscarriages (a conservative estimate) and 3% due to
stillbirths (the average rate of stillbirths observed in developing countries [38]). Thus, of 1,820 reproductive
weeks, a woman is pregnant for 268 weeks (6.7 × 40 weeks); of which 40.2 weeks (6.7 × 6 weeks) are during
the sensitive six‐week time window from week four to week ten of gestation. Under these conditions, 14.7%
(268 of 1,820) of WOCBAs are pregnant at any time (i.e., one in 6.8), and 2.2% (40.2 of 1,820) or one in 45
are pregnant between week four and week ten.
If accidental exposure is defined as unintentional treatment in early pregnancy only, than the risk of
accidental exposure is slightly higher than 2.2%, as later pregnancy weeks do not contribute to the
denominator. The average time for women in Africa to recognise and report a pregnancy is not well
described in settings where pregnancy testing is not readily available. If it is assumed that this is during the
first ten weeks of pregnancy, then the denominator is 1,619 weeks (the 1,552 weeks that she is not pregnant
[1,820 − 268] plus the 67 weeks of early pregnancy (6.7 pregnancies × 10 weeks), and the risk of accidental
exposure in the four‐ to ten‐week period is 40.2 out of 1,607 weeks or 2.5% (one in 40 women).
This assumes that the probability of getting clinical malaria is the same in these first ten weeks of pregnancy
as in non‐pregnant women.
Antimalarial pregnancy exposure registries
73
ConclusionThe establishment of an international antimalarial pregnancy exposure registry, using specialised
sentinel sites to provide reliable exposure and outcome data for the primary data collection, is a
potentially cost‐effective targeted approach. Central collation of the information would enable
evaluation of the risk–benefit profile of antimalarials in a timely manner, and over time would allow
the detection of rare adverse drug reactions that could not be detected by any single study. New
levels of collaboration between pharmacovigilance programmes, antimalarial drug developers,
research groups, regulatory authorities, and WHO will be essential. This international multi‐product,
multi‐ sponsor approach will require good governance structures, such as those used by the
Antiretroviral Pregnancy Registry, and if successful could serve as a pathfinder for other PERs to
capture much‐needed safety information on other drugs used for tropical diseases [9].
References1. Mitchell A (1994) Special considerations in studies of drug‐induced birth defects. In: Strom B, editor.
Pharmacoepidemiology. Chichester: Wiley. 2. Kennedy DL, Uhl K, Kweder SL (2004) Pregnancy exposure registries. Drug Saf 27: 215‐228. 3. Center for Drug Evaluation and Research, Center for Biologics Evaluation and Research (2002) Guidance for
industry: Establishing pregnancy exposure registries. US Food and Drug Administration. Available: http://www. fda.gov/cber/gdlns/pregexp.htm. Accessed 10 August 2008.
4. European Medicine Agency (2005) Guideline on the exposure to medicinal products during pregnancy: Need for post‐authorisation data. Available: http://www.emea.europa.eu/pdfs/ human/phvwp/31366605en.pdf. Accessed 10 August 2008.
5. Lang T, Hughes D, Kanyok T, Kengeya‐ Kayondo J, Marsh V, et al. (2006) Beyond registration—Measuring the public‐health potential of new treatments for malaria in Africa. Lancet Infect Dis 6: 46‐52.
6. Staedke S (2005) Pharmacovigilance and antimalarial treatment in Africa: The practical challenges. The Newsletter of The East African Network for Monitoring Antimalarial Treatment 8: 2. Available: http://www.eanmat.org/Newsletters/Nwsltt21.doc. Accessed 10 August 2008.
7. Talisuna AO, Staedke SG, D’Alessandro U (2006) Pharmacovigilance of antimalarial treatment in Africa: Is it possible? Malar J 5: 50.
8. Pirmohamed M, Atuah KN, Dodoo AN, Winstanley P (2007) Pharmacovigilance in developing countries. BMJ 335: 462.
9. White NJ, McGready RM, Nosten FH (2008) New medicines for tropical diseases in pregnancy: Catch‐22. PLoS Med 5: e133. doi:10.1371/journal.pmed.0050133
10. Steketee RW, Nahlen BL, Parise ME, Menendez C (2001) The burden of malaria in pregnancy in malaria‐endemic areas. Am J Trop Med Hyg 64: 28‐35.
11. Desai M, ter Kuile FO, Nosten F, McGready R, Asamoa K, et al. (2007) Epidemiology and burden of malaria in pregnancy. Lancet Infect Dis 7: 93‐104.
12. World Health Organization (2006) Guidelines for the treatment of malaria. WHO/HTM/MAL/2006.1108. Available: http://www.who.int/malaria/includes_en/ whomalariapublications19982004.htm. Accessed 10 August 2008.
13. McGready R, Cho T, Keo NK, Thwai KL, Villegas L, et al. (2001) Artemisinin antimalarials in pregnancy: A prospective treatment study of 539 episodes of multidrug‐ resistant Plasmodium falciparum. Clin Infect Dis 33: 2009‐2016.
14. Dellicour S, Hall S, Chandramohan D, Greenwood B (2007) The safety of artemisinins during pregnancy: A pressing question. Malar J 6: 15.
15. Nosten F, McGready R, Alessandro U, Bonell A, Verhoeff F, et al. (2006) Antimalarial Drugs in Pregnancy: A Review. Curr Drug Saf 1: 1‐15.
16. Ward SA, Sevene EJ, Hastings IM, Nosten F, McGready R (2007) Antimalarial drugs and pregnancy: Safety, pharmacokinetics, and pharmacovigilance. Lancet Infect Dis 7: 136‐144.
17. US Food and Drug Administration (2008) List of pregnancy exposure registries. Available: http://www.fda.gov/womens/registries/registries.html. Accessed 10 August 2008.
18. Warkary J (1974) Problems in applying teratologic observations in animals to man. Paediatrics 53: 820.
Chapter 4
74
19. Honein MA, Paulozzi LJ, Cragan JD, Correa A (1999) Evaluation of selected characteristics of pregnancy drug registries. Teratology 60: 356‐364.
20. Simpson JL (1990) Incidence and timing of pregnancy losses: Relevance to evaluating safety of early prenatal diagnosis. Am J Med Genet 35: 165‐173.
21. Andrews E, Stergachis A, Hecht J (1989) Evaluation of alternative methods of assessing pregnancy outcomes using automated indicators of pregnancy J Clin Res Drug Dev 3: 201.
22. Briggs GG (2002) Drug effects on the foetus and breast‐fed infant. Clin Obstet Gynecol 45: 6‐21. 23. Christianson RE, van den Berg BJ, Milkovich L, Oechsli FW (1981) Incidence of congenital anomalies
among white and black live births with long‐term follow‐up. Am J Public Health 71: 1333‐1341. 24. Siffel C, Correa A, Cragan J, Alverson CJ (2004) Prenatal diagnosis, pregnancy terminations and prevalence
of Down syndrome in Atlanta. Birth Defects Res A Clin Mol Teratol 70: 565‐571. 25. Strom BL (2000) Pharmacoepidemiology. Sussex: John Wiley & Sons. 26. Covington DL, Tilson H, Elder J, Doi P (2004) Assessing teratogenicity of antiretroviral drugs: Monitoring
and analysis plan of the Antiretroviral Pregnancy Registry. Pharmacoepidemiol Drug Saf 13: 537‐545. 27. Scheuerle A, Covington D (2004) Clinical review procedures for the Antiretroviral Pregnancy Registry.
Pharmacoepidemiol Drug Saf 13: 529‐536. 28. Joint CIOMS/WHO Working Group (2005) Drug development research and pharmacovigilance in resource‐
poor countries. Available: http://www.cioms.ch/frame_whats_ new.htm. Accessed 10 August 2008. 29. Huff‐Rouselle M, Simooya O, Kabwe V, Hollander I, Handema R, et al. (2007) Pharmacovigilance and new
essential drugs in Africa: Zambia draws lessons from its own experiences and beyond. Glob Pub Health 2: 184‐203.
30. Tilson J, Watts J, Covington D (2006) Effective use of supplemental data in pregnancy exposure registries [presentation]. 22nd International Conference on Pharmacoepidemiology & Risk Management; 24‐27 August 2006; Lisbon, Portugal.
31. Clark R, Kumemura M, Makori N, Nakata Y, Bernard F, et al. (2006) Artesunate: Developmental toxicity in monkeys [abstract]. 46th Annual Meeting of the Teratology Society; 24‐29 June 2006; Tucson, Arizona, United States.
32. Clark RL, White TE, A Clode S, Gaunt I, Winstanley P, et al. (2004) Developmental toxicity of artesunate and an artesunate combination in the rat and rabbit. Birth Defects Res B Dev Reprod Toxicol 71: 380‐394.
33. Longo M, Zanoncelli S, Manera D, Brughera M, Colombo P, et al. (2006) Effects of the antimalarial drug dihydroartemisinin (DHA) on rat embryos in vitro. Reprod Toxicol 21: 83‐93.
34. Longo M, Zanoncelli S, Della Torre P, Rosa F, Giusti A, et al. (2008) Investigations of the effects of the antimalarial drug dihydroartemisinin (DHA) using the Frog Embryo Teratogenesis Assay‐Xenopus (FETAX). Reprod Toxicol 25: 433‐441.
35. World Health Organization (2007) Assessment of the safety of artemisinin compounds in pregnancy: Report of two joint informal consultations convened in 2006. WHO/ CDS/MAL/2003.1094. Available: http:// www.who.int/tdr/publications/publications/ pdf/artemisinin_compounds_pregnancy.pdf. Accessed 10 August 2008.
36. White TEK, Clark RL (2008) Sensitive periods for developmental toxicity of orally administered artesunate in the rat. Birth Defects Res B Dev Reprod Toxicol. E‐pub 9 July 2008.
37. World Health Organization (2004) World health statistics. Basic demographic and socio‐economic indicators: Total fertility rate, female, for latest available data. WHO Global Health Atlas. Available: http://www.who.int/ GlobalAtlas/. Accessed 10 August 2008.
38. Say L, Donner A, Gulmezoglu AM, Taljaard M, Piaggio G (2006) The prevalence of stillbirths: A systematic review. Reprod Health 3: 1.
39. McCombie SC (1996) Treatment seeking for malaria: A review of recent research. Soc Sci Med 43: 933‐945. 40. Hamer DH, Ndhlovu M, Zurovac D, Fox M, Yeboah‐Antwi K, et al. (2007) Improved diagnostic testing and
malaria treatment practices in Zambia. JAMA 297: 2227‐2231. 41. Reyburn H, Mbakilwa H, Mwangi R, Mwerinde O, Olomi R, et al. (2007) Rapid diagnostic tests compared
with malaria microscopy for guiding outpatient treatment of febrile illness in Tanzania: Randomised trial. BMJ 334: 403.
75
Chapter4Appendix&SupportingInformation
Table S1. Number of pregnancies needed in each group to detect various relative risks for
pregnancy outcomes with different prevalence.
Figure S1. Pregnancy Exposure Registry Sample Size calculation according to the exposed to
unexposed ratio for major malformation.
Found at doi:10.1371/journal. pmed.0050187.sd001 (596 KB DOC).
Chapter 4 Appendix
76
Table S1. Number of pregnancies needed in each group to detect various relative risks for pregnancy
outcomes with different prevalence (Ratio of exposed to unexposed of 1:4, Power 80% and one‐
sided α=0.05). These sample size calculations are based on a one‐sided approach because pregnancy
exposure registries are designed to detect safety signals rather than to examine potential protective
effects. Based on the formula for Cohort design described in Strom’s Pharmacoepidemiology[31]:
N=1/[p(1‐R)]2x [Z 1‐α√((1+1/k)U(1‐U))+Z1‐β√(pR(1‐Rp)+(P(1‐P))/k)]2 where p is the incidence of disease
in unexposed; R is the minimum relative risk to detect; k is the ration of unexposed controls to
exposed and U=(Kp+pR)/(k+1). These estimates do not include loss to follow‐up and do not account
for the expected 16‐18% of pregnancies that may not result in a live birth. The latter needs to be
considered for outcomes such as major malformation and specific birth defects.
*This is the number of exposed pregnancies required against a known background rate that was
derived using PASS software for cohort studies [Reference: Hintze J (2006) NCSS, PASS and GESS. In:
NCSS, editor. Kaysville, Utah].
Relative Risk
Exposed to Unexposed Ratio of 1:1
Exposed to Unexposed Ratio of 1:4
Against Background Rate*
Number Exposed
Number Unexposed
Total Number Exposed
Number Unexposed
Total Number Exposed
Miscarriage (p=15%)
1.2 1,897 1,897 3,794 1,168 4,672 5,921 1,098
1.5 335 335 670 202 808 1,011 191
2 95 95 190 56 224 280 54
5 7 7 15 4 16 20 5
10 NA NA NA NA NA NA NA
Stillbirths (p=3%)
1.2 10,991 10,991 21,981 6,747 13,493 20,712 5,490
1.5 1,988 1,988 3,976 1,192 4,767 5,959 955
2 591 591 1,182 343 1,372 1,715 268
5 69 69 139 37 146 183 26
10 22 22 45 11 45 56 8
Major Malformations (p=2%)
1.2 16,674 16,674 33,348 10,233 40,934 51,884 8,236
1.5 3,022 3,022 6,043 1,810 7,241 9,052 1,432
2 901 901 1,802 522 2,090 2,612 402
5 108 108 216 57 227 284 39
10 36 36 73 18 72 90 11
Specific Birth Defect (p=0.1%)
1.2 340,629 340,629 681,259 208,978 835,913
1,059,540 164,712
1.5 61,922 61,922 123,845 37,067 148,270 185,337 28,636
2 18,571 18,571 37,143 10,748 42,992 53,740 8,038
5 2,317 2,317 4,634 1,206 4,824 6,030 777
10 836 836 1,673 409 1,638 2,047 229
Antimalarial pregnancy exposure registries
77
Figure S1. Pregnancy Exposure Registry Sample Size calculation according to the exposed to
unexposed ratio for major malformation (assuming a background rate of 2%; Power 80% and one‐
sided α=0.05).
*This is the number of exposed pregnancies required against a known background rate that was derived using PASS
software for cohort studies [Reference Hintze J (2006) NCSS, PASS and GESS. In: NCSS, editor. Kaysville, Utah].
901 650 565 522 497 479 467 457 450 444 402
901 1300 16962090
24832875
32673659
40514443
1802 19512261
2612
2979
3354
3734
4117
4501
4887
0
1000
2000
3000
4000
5000
6000
7000N
umbe
r of P
regn
anci
es
Ratio of Unexposed-control per Exposed pregnancy
Unexposed Exposed
78
79
Chapter5
ProbabilisticRecordLinkageforMonitoringtheSafetyofArtemisinin‐BasedCombinationTherapyintheFirstTrimesterofPregnancyinSenegal
StephanieDellicour1,2,3,PhilippeBrasseur4,PerThorn5,OumarGaye6,PieroOlliaro7,8,MalikBadiane9,AndyStergachis10,FeikoO.terKuile1
1 Child and Reproductive Health Group, Liverpool School of Tropical Medicine, Liverpool,
United Kingdom, 2 Kenya Medical Research Institute (KEMRI), Center for Global Health
Research (CGHR), Kisumu, Kenya, 3 Graduate School, Academic Medical Centre,
Amsterdam, The Netherlands, 4 Institut de Recherche pour le Développement (IRD),
Dakar, Sénégal, 5 Thorn IT Services Limited, London, United Kingdom, 6 Service de
Parasitologie, Faculté de Médecine, Université Cheikh Anta Diop, Dakar, Senegal 7
UNICEF/UNDP/WB/WHO Special Programme for Research & Training in Tropical Diseases
(TDR), Geneva, Switzerland, 8 Centre for Tropical Medicine and Vaccinology, Nuffield
Department of Medicine, University of Oxford, Churchill Hospital, Oxford, United
Kingdom, 9 District Médical d'Oussouye, Oussouye, Sénégal, 10 Departments of
Epidemiology and Global Health, Global Medicines Program, School of Public Health,
University of Washington, United States of America
Drug Safety 2013, 36:505–513
Copyright© Adis
Chapter 5
80
AbstractBackground There are insufficient data on the safety in early pregnancy of the artemisinins,
a new class of antimalarials. Assessment of drug teratogenicity requires large sample sizes for
an adequate risk‐benefit assessment. There is currently limited pharmacovigilance
infrastructure in malaria‐endemic countries. Monitoring drug safety in early pregnancy is
especially challenging, as it requires early pregnancy detection to assess any potential
increased risk of miscarriage, prospective follow‐up to reduce recall and survival biases, and
accurate data on gestational age assessment. Record linkage approaches for pregnancy
pharmacovigilance using routinely generated health records could be a pragmatic and cost‐
effective approach for pharmacovigilance in early pregnancy, but has not been evaluated in
resource‐poor settings.
Objective Our objective was to assess the feasibility of record linkage using routinely
collected healthcare data as a pragmatic means of monitoring the safety in early pregnancy of
artemisinin‐based combination therapies (ACTs) in Senegal.
Methods Data (2004–2008) from paper‐based registers from outpatient clinics, antenatal care
services (ANC) and the delivery unit from the St Joseph dispensary in Mlomp, south‐western
Senegal, were entered into databases. Record linkage based on a probabilistic matching
approach was used to identify pregnancies exposed to ACTs in the first trimester of pregnancy.
Two record linkage software packages (Link‐Plus and FRIL) were compared and output data were
reviewed independently by two investigators.
Results Information on 685 pregnancies was extracted, 536 of which were from the geographic
catchment area and eligible for record linkage; 94.6 % of them resulted in live births, 2.6 % in
stillbirths and 2.8 % in miscarriages. Major congenital malformations were identified in 1.6 % of
births. Seventy‐three and 75 true matches between pregnancy outcome and the outpatient
treatment registers were identified by two different record linkage software packages,
respectively. Record linkage identified seven exposures to ACTs in the first trimester, all of which
resulted in normal live‐births.
Conclusion Probabilistic record linkage is a potentially cost‐effective method to assess the safety
of antimalarials in early pregnancy in resource‐constrained settings to assess increased risk of
overall birth defects, and stillbirths in settings with good existing health records and well defined
target populations.
Probabilistic Record linkage to assess antimalarials safety in pregnancy
81
BackgroundIn the last decade, Senegal and other malaria‐endemic countries changed their first‐line treatment
policy for malaria to artemisinin‐based combination therapy (ACT). The artemisinin class of
antimalarials all have embryotoxic effects at low dose ranges in all animal species studied [1, 2]
and the World Health Organisation (WHO) does not recommend their use for non‐severe
malaria in the first trimester as there is insufficient information about their safety in humans.
Because of their widespread use, many women in endemic countries risk inadvertent exposure
early in pregnancy when they are either unaware of their pregnancy or do not report being
pregnant. To date, the data from 359 well documented exposures in the first trimester suggests
that the benefits outweigh the potential safety concerns but more data from a wider range of
malaria‐endemic countries are required to provide adequate reassurance [4–7].
Passive mechanisms of spontaneous adverse drug effects reporting are inadequate to detect drug‐
induced fetal risks or lack of such risks [3]. Different prospective study designs, including
pregnancy registers, are used to monitor safety of medication used during pregnancy in the post‐
marketing phase [4, 5]; however, these require considerable resources and a well functioning
health system and records infrastructure, which are often unavailable in resource‐constrained
settings [3, 6].
In the last few years there have been an increasing number of pregnancy postmarketing studies
using record linkage approaches in developed countries [7–10], which may also have application
in resource‐limited countries to evaluate the teratogenicity of a drug. It enables rapid evidence
generation by using existing healthcare data collected prospectively, while eliminating potential
recall bias. Often data on drug exposure, prenatal and pregnancy outcomes are available from
multiple linkable data sources. In industrialised countries, this information can be derived from
medical records and automated databases, including insurance claims [11]. A recent study
showed the feasibility of using record linkage in resource‐constrained settings to assess adverse
reactions to antiretroviral therapy [12], but such approaches have not yet been applied to
assess drug safety in pregnancy in such settings.
Record linkage requires access to long‐term, comprehensive and stable population datasets
with personal identifiers. In situations where unique identifiers are available, a deterministic
record linkage technique, which involves exact matching, can be applied. Probabilistic record
linkage is used in situations where there is no universal unique personal identifier and is based on
the assessment of similarity between pairs of records, allowing for a level of error (such as
typographical or spelling differences) in matching variables [13–15]. The mathematical framework
underlying probabilistic record linkage was first formulated by Newcombe and Kennedy in 1962
[16] and further developed by Fellegi and Sunter [13] in 1969. A linkage probability score is
computed for each pair of records based on the sum of the probability of agreement for each
matching variable. Probability of agreement for each matching variable reflects the probability
that the values match by chance based on the frequency of that value occurring in the
datasets. Thresholds for the combined linkage probability score are set to determine which pairs
are a true match, those that are potential matches and those that are not a match.
We report the results of a study to determine the feasibility of record linkage as a pragmatic
approach to retrospectively examine the potential risk associated with inadvertent exposure to
Chapter 5
82
ACTs in early pregnancy using data generated through routine clinical practice in a rural mission
dispensary in southern Senegal.
Methodology
StudySettingThis study was conducted in 2009 in a mission dispensary based in Mlomp, a rural village of
approximately 8,000 inhabitants in the District of Oussouye, Casamance, south‐ western Senegal.
Malaria is meso‐endemic in this area, occurs year round and peaks during the rainy season (July
to December). A recent study showed that malaria transmission intensity in southern Senegal has
been decreasing significantly in the past 15 years [17].
Mlomp has had a private dispensary operated by French catholic nuns since 1961. The dispensary
offers outpatient services, antenatal care service (ANC) once a week, and has a delivery unit.
Nearly all pregnant women attend ANC, and health facility deliveries have increased from 50 %
in 1961 to 99 % in 1999 [18, 19].
The setting is potentially well suited for record linkage studies. The clinic serves a stable, well
defined population and the dispensary registers for ANC, delivery, child welfare clinic and general
outpatient visits have been meticulously kept since 1993 [20]. Almost all antimalarials used are
provided by the clinic because there is no external source of drugs within the study area; the closest
pharmacy or drug store is in the nearest town, Oussouye, 10 km away, with limited options for
public transport.
DataCollectionThe general outpatient, antenatal, pregnancy complications/miscarriage and delivery registers from
Mlomp dispensary covering the period 2004–2008 were digitalised using a digital camera (Canon
EOS 450D with external flash and tripod). The data from the digital images of the registers were
subsequently entered into Microsoft® Excel spreadsheets. For the purpose of this feasibility study,
only entries for women of childbearing age (15–49 years) with a prescription for an ACT between
January 2004 and December 2007 were extracted from the outpatient register. All treatment
information, including treatment given at the time of antenatal visits, is recorded in the outpatient
register. All recorded pregnancy outcomes between May 2004 and September 2008 were entered
from the pregnancy complication and the delivery registers (Table 1). Information on newborn
abnormalities was recorded in the delivery register based on observation by the midwife attending
the delivery. Data from the outpatient register were double entered and the entries for the delivery
register were compared with data extracted for a previous study [20].
RecordLinkageMethodProbabilistic record linkage based on first names, surname, address and year of birth was used to
link information from the different registers because they did not contain unique patient
identifiers (Table 2). Two stand‐alone software packages were used and compared: (i) Registry
PlusTM Link Plus (Emory University, v.2.0) [21] (henceforth referred to as ‘Link‐Plus’), a royalty‐free
probabilistic record linkage program developed at the Division of Cancer Prevention and Control
of the Centers for Disease Control and Prevention (CDC), USA; and (ii) Fine‐grained Record Linkage
software (FRIL v. 2.1.5), a free open‐ source tool developed by Emory University and the CDC
[22, 23].
Probabilistic R
ecord lin
kage to
assess antim
alarials safety in
pregn
ancy
83
Table 1 Su
mmary o
f data availab
le in each
health
registe
r from th
e Mlomp Disp
ensary in
Senegal
G
eneral outpatient register A
ntenatal register P
regnancy complication register
Delivery register
Identifiers and demographic
• First andlast
names
•First
andlast
names
•First
andlast
names
•First
andlast
names
• A
ge
• Sex • A
ge
• Address
• Date of birth/age
• Marital status
• Date of birth/age
• Address
• A
ddress • M
arital status
• Em
ployment
Clinical
informatio n
• Presenting sym
ptoms
SeeA
NC
row4
• Hospitalisation dates
• Drugs
providedduring
delivery and care provided
• Diagnosis
• Treatm
ent
• Com
ments
• Care and drugs provided
or post-partum to m
other and baby
Obstetric history
None
• No. of previous pregnancies
• No. of previous deliveries
• No. of pregnancies
• No. of live-births
• No. of stillbirths
• No. of m
iscarriage
• No. of children alive
• Date of delivery
• Nam
e of child
• Sex • P
lace of delivery
• Child status
AN
C
None
• Date of A
NC
visit
• LM
P
• Weight, H
b, fundal height, blood pressure for each visit
• Risk factors
• Tetanus vaccination
None
• Date of A
NC
visits • W
eight, Hb, fundal height,
blood pressure for each visit
• Tetanus vaccination
AN
C antenatal care, L
MP
last menstrual period
Chapter 5
84
Following agreement on the matching criteria, two investigators conducted the record linkage
independently. Any discrepancies in results between investigators were compared and discussed
until a consensus was reached.
Table 2 Description of matching variables from the delivery register
Matching Variables Discriminating power/ # possible values
Missing value (with
ANC info)a
Variable limitations
First Name 230 0 • Variation in spelling/ or spelling errors • Truncated/nicknames • Hyphenated /multiple names • Inconsistent and interchangeable use and recording of traditional and
Christian first and second names
Surname 73 0.1% • Variation in spelling/ or spelling errors • Commonly changes for women of child bearing age after
marriage/divorce
Address 29 33% (10%) • Changes over time • Commonly changes for women of child bearing age after
marriage/divorce • Address only consists of neighborhood name which is not very
discriminating
Age or Year of birth 33 10% (4%) • Inaccuracy in age or date of birth is common in the study population a Manual search of the corresponding records in the ANC register enabled completing certain missing values for age and
address.
A variety of matching methods are available for each matching variable to take into account
different spelling, recording or typographical errors. To identify the optimal parameters that
produce the best linkage results, different linkage setups were compared by varying the weights
given to the matching variables. The adequacy of the linkage result was assessed by the
distribution of matches, uncertain and non‐matches according to the linkage score, the number of
true matches identified and the positive predictive value (PPV) for the selected threshold. In Link‐
Plus, the matching score threshold was set to 0 (minimal threshold option) in order to derive
histograms from the linkage output for each matching set up (this was not an option in FRIL).
True matches were defined as records with matching first names, surname, and address, allowing
for some misspelling and year of birth within 5 years of each other. Pairs were assigned as
uncertain matches if first name(s) matched, year of birth was within 10 years of each other but the
address and/or surname did not match. Surname and address can change over time, and this is
especially common among women of childbearing age when they get married. All other pairs of
records were considered as non‐match records. Details of the selected linkage setups are
provided in Table 3. For information on the linkage optimisation procedures, see appendices I
and II. Deduplication of the delivery register was performed using the deduplication function of
Link‐Plus, as some women could have multiple pregnancies within the 4‐year study period, and
to enable one‐to‐many matching.
AnalysisDescriptive statistics were used to determine the proportion of women with inadvertent exposure
to ACTs during the first trimester of pregnancy using the data obtained by linking the records
from the outpatient and the delivery/ pregnancy complication registers. Characteristics of
women included and excluded from the record linkage were compared using the Pearson Chi‐
square statistic. A first trimester exposure was defined as an ACT prescription (either artesunate‐
Probabilistic Record linkage to assess antimalarials safety in pregnancy
85
amodiaquine or artemether‐lumefantrine) provided to any woman in 2–14 weeks (inclusive) of
pregnancy. Data on last menstrual period (LMP) were not collected in either the outpatient or the
delivery registers. Information on the estimated gestational age was only available in the delivery
register and based on the assessment by the midwife at birth. This was categorised as ‘term’ or
‘pre‐term’. For pre‐term deliveries, the gestational age at delivery was specified in months from
LMP in the delivery records, whereas for all ‘term’ deliveries, the specific gestational age was not
recorded other than the notation that they were ‘term’ births. It was therefore set at 40 weeks
from LMP for the purpose of the analysis. The gestational age at the time of exposure was
derived from the date of the ACT prescription in the outpatient register and the estimated date
of conception calculated from the gestational age assessment at birth. Analysis was done in SPSS
version 18.
Table 3 Description of linkage parameters in Link‐Plus and FRIL
Link‐Plus set up FRIL set upb
Linkage Variablesa
M‐ probabiltyc
Matching methodd Weight Matching methodd
First Name 0.97 First Name (J‐W) 32 Edit distance (0.1‐0.4)
Second Name 0.7 Generic String 10 Edit distance (0.1‐0.4)
Surname 0.85 Last Name (J‐W) 25 Edit distance (0.1‐0.4)
Address 0.85 Generic String 23 Edit distance (0.1‐0.4)
Year of birth 0.2 Generic String 10 Numeric distance (+/‐ 5 years) a all variables are standardized b m‐probability determines the reliability of the variable, it ranges from 0 (unreliable) to 1 (very reliable). cMatching Methods descriptions:
J‐W: Jaro‐Winkler Metric is a string comparator which measures the partial agreement between two strings accounting for random insertion, deletions, and transpositions.
Generic String and Edit‐distance: incorporates partial matching to account for typographical errors and calculates the number of operations (insertion, deletion, or substitution of a single character) needed to transform one string into the other. The approve and disapprove levels need to be specified in FRIL for edit distance function.
Numeric distance: allows users to specify a range of values that will have a non‐zero match score. dFRIL doesn’t calculate m‐probability as in Link‐Plus but enables the user to set weights for each matching variables summing to 100. To compute the total score for a pair, the weight of each matching variable is multiplied by the value returned by chosen edit distance function and all these values are added together.
Results
DescriptionBetween May 2004 and September 2008, 685 pregnancy outcomes were captured from the
pregnancy complication and delivery registers. Women who attended their ANCs outside of the
Mlomp catchment area (n = 149) were excluded from the record linkage because no outpatient
treatment records were available from these other clinics.
Their characteristics did not differ significantly (at 5 % significance level) from other women living in
the catchment area, except that they were more likely to be primiparous women and to be slightly
younger. Figure 1 depicts the number of records that contributed to the probabilistic matching. Of
the pregnancies included, 94.6 % resulted in live births, 2.6 % in stillbirths and 2.8 % in miscarriages
(Table 4). Overall, 11 cases of birth defects (1.6 %) were captured in the delivery register (Table 5).
Chapter 5
86
Fig. 1 Flow diagram for inclusion in record linkage, matching result and resulting ACT pregnancy exposure. Note that of the 451 women, 372 had only one pregnancy during the study period (2004–2008), 73 had two pregnancies and six had three pregnancies during the 4‐year period
Table 4 Characteristics of 536 eligible pregnancies from the delivery andpregnancy complication registers (2004–2008) St Joseph Dis‐ pensary Mlomp, Senegal
Table 5 Description of congenital abnormalities identified from the delivery registers (2004–2008) in St Joseph Dispensary Mlomp, Senegal
Characteristic N (%)a
Age in years
Mean (SD) 29.2 (6.6)
Range 15–47
Parity
0 94 (17.5)
1 87 (16.2) 2? 355 (66.2)
Range 0–13
Number reporting previous stillbirth 26 (4.9)
Number reporting previous miscarriage 55 (10.3)
Pregnancy outcomes Live births 507 (94.6)
Stillbirths 14 (2.6)
Miscarriages 15 (2.8)
Preterm at delivery 26 (4.9)
a Unless otherwise indicated
Congenital abnormalitiesa Number
Anophthalmia 1
Ambiguous genitalia 1
Anencephaly 3
Down syndrome 6
Club foot 2
Hydrocephalus 1
Imperforate anus 1
Unspecified malformations 1
aSome cases had more than one abnormality. There were 16abnormalities among 11 births: two cases had both Down syndrome and anencephaly, one case of Down syndrome also had a club foot, one case of anophthalmia also had a club foot and one case with Down syndrome also had hydrocephalus
ProbabilisticMatchingThe optimum set up with highest PPV (86 %) and highest number of matches was accomplished by
including second names. Figure 2 depicts the distribution of match, uncertain and non‐match for
this set up. The relatively high number of false negatives (n = 13) and false positives (n = 8)
suggests clerical review would still be required for pairs with a linkage probability score between 6
and 9.
Probabilistic Record linkage to assess antimalarials safety in pregnancy
87
Fig. 2 Histogram of the optimum set up depicting the distribution of matches, uncertain and non‐matches according to the probability score derived from Link‐Plus
Potential cut off
Seventy‐one and 75 matched pairs between the pregnancy outcome registers and the ACT
treatment data from the outpatient registers were detected using Link‐Plus and FRIL, respectively.
The four additional matches identified through FRIL were not detected through Link‐Plus as it
only allows one‐to‐many matching. After running another round of matching in Link‐Plus using
the non‐match records only and adjusting the matching variables weight, an additional two
matches were identified (total of 73 matched pairs). The other two treatment records were
matched by Link‐Plus with incorrect records from the delivery register and were not compared
again to the correct delivery records. The two cases missed in Link‐Plus were not exposed during
the pregnancy period.
ArtemisininExposureinPregnancyOut of the true matches, 11 of the 536 pregnancies (2 %) had evidence for ACT exposure during
pregnancy, seven during the first trimester (1 %), three of whom were 4–10 weeks (the
projected embryo‐sensitive period from animal models) pregnant (0.6 %) and four in the second or
third trimester. There was no exposure in the month prior to the estimated conception dates.
None of the women exposed during pregnancy had an adverse pregnancy out‐ come, preterm
birth or a baby with malformations.
DiscussionThe findings from this feasibility study suggest that record linkage using routine healthcare data is
feasible in resource‐constrained settings with a relatively well defined catchment population.
Among the 451 women with pregnancy outcomes, 75 records from the delivery register were
matched to an ACT treatment in the outpatient registers, although only 11 were ACT exposures
during the pregnancy period; the other matches represent ACT treatment received before or
after pregnancy. Matching was feasible despite the limited discriminating power resulting from the
lack of variance among two of the five matching variables due to clustering of commonly used
surnames, lack of address details and the lack of precise dates of birth for many individuals.
Chapter 5
88
Furthermore, the interchangeable use of traditional and Christian first names by single individuals
was very common and often one was recorded, but seldom both.
Although there was no other source of medication in Mlomp and the nearby surrounding areas,
the risk of exposure misclassification remains, as women could have obtained drugs from
relatives or friends, could have travelled during their pregnancy and obtained drugs elsewhere or
the drugs could have been prescribed and recorded in the register but never actually taken by the
patient. The limited detail on gestational age (categorised as term or pre‐term with gestational
month) could also cause misclassification for timing of exposure. Furthermore, there could have
been errors or omissions during recording of data in the health registers and at time of data entry.
We attempted to minimise the latter by using double data entry for the general outpatient register
and by comparing the entries for the delivery register with data extracted for a previous study [20].
The prevalence of birth defects was within the expected range for major malformations detectable
at birth by surface exams by non‐specialists without special training for newborn examination.
Although the background prevalence of birth defects in developing countries is unknown,
extrapolation from birth defect prevalence found in industrialised countries suggests that the
defect prevalence detectable at birth would be around 1 % after exclusion of defects of genetic
aetiology and heart defects, which require specialist assessment to be diagnosed. The background
rate of stillbirth (2.3 %) was also within the projected range for Senegal (2.7 %) and sub‐Saharan
Africa (3.2 %) [24]. As expected, the proportion of miscarriages (2.5 %) was much lower than the
risk of 12–15 % typically quoted for miscarriages because most pregnant women do not present for
their first ANC visit before the second trimester. A previous record linkage study from Mlomp
dispensary showed that half of all pregnant women presented for their first ANC visit after 20 weeks
gestation, thus only allowing the prospective recording of miscarriages relatively late in the second
trimester (20–28 weeks) [20].
The two record linkage software programs used had comparable performance. The main advantage
of FRIL was that it included a matching method for numeric variables, which allowed variation in
numeric values (i.e. year of birth could be within 5 years of each other and considered a match).
There is no function for numeric variables in Link‐Plus, and year of birth had to be treated as a
string variable. Many‐to‐many pairwise comparison is possible in FRIL, whereas only one‐to‐many is
enabled in Link‐Plus (many‐to‐many will be available in the next version, 3.0 [25]). This was
highlighted by the two missed true matches identified through FRIL but not Link‐Plus described
above. Limitations for FRIL included the more limited flexibility to export datasets after linkage
(only matched pairs can be exported). Link‐Plus, on the other hand, enables the export of all pairs
reviewed manually (matches, uncertain and non‐matches), which then allows more complete
secondary analysis and graphical depiction of results.
Ruling out the teratogenic risk of a drug requires a very large sample size, as teratogens usually
induce specific patterns of birth defects that occur rarely in the general population. We have
shown previously that 10,748 well characterised exposures and four times as many unexposed
controls would be required to exclude a doubling of risk of a specific birth defect that occurs at a
frequency of 0.1 % [11]. Obtaining reliable data on early pregnancy drug exposures is challenging.
McGready and colleagues recently reported 64 well documented first trimester exposures to
ACTs after reviewing 25 years of data on 48,426 pregnancies [26]. This was enough to exclude
Probabilistic Record linkage to assess antimalarials safety in pregnancy
89
a doubling of risk of miscarriage for first trimester exposures [27]. To obtain a similar level of
reassurance for major malformations, data from several sentinel sites over several years are
required. Collaborating pregnancy registry sites have been set up by WHO [28] and the Malaria
in Pregnancy consortium [29] for this purpose.
Record linkage studies require minimal staff (for data extraction, data entry and analysis) and
resources (i.e. digital camera and a few computers for data entry). In settings with electronic
medical records, only staff for data management and analysis would be required. However,
assessment of the risk of miscarriage and specific birth defects, such as congenital heart defects,
requires dedicated studies. A ‘miscarriage’ endpoint requires an observational cohort study
involving women of childbearing age to capture data as early as possible in their pregnancy, while
assessment of specific rare birth defects would probably require nested case control or case
cohort studies in settings where drug exposure information is captured routinely. Additional data
from sites where record linkage is a possibility would greatly contribute to the achievement of an
adequate sample size to guide policy makers. Requirements for such sites include availability of
reliable and comprehensive medical records for treatment, pregnancy and maternity services, in a
relatively stable population where healthcare is provided in a central location with a high level
of health facility deliveries and limited availability of the drug of interest outside the central health
facility. Settings such as private agricultural estates (e.g. tea, coffee, sugar or cotton plantations),
where employees and their families get healthcare centrally and routine healthcare data are
typically recorded, could be used for similar record linkage studies. Data from health insurance
schemes, where unique identifiers are available, could be used with a combination of deterministic
and probabilistic record linkage [30, 31].
ConclusionOur findings suggest that record linkage to assess drug safety in pregnancy in resource‐
constrained settings is feasible for assessment of stillbirths and major congenital malformations
detectable by surface examination. Tapping into readily available data sources of sites adequate
for record linkage would greatly contribute to the high numbers needed to provide adequate
reassurance for ACT use in the first trimester of pregnancy.
Acknowledgments
This work was made possible by the dedication of the healthcare personnel of St Joseph Dispensary
in Mlomp and the help of Sister Marie Joelle.
Author’s contributions SD, FtK, OG and PO contributed to the concept of the project. SD and FtK developed the protocol with contributions from PO, PB and AS. SD and PT conducted the data collection and record linkage analyses. PB supervised the data collection. SD and FtK wrote the first draft of the manuscript; all authors reviewed and revised the final version.
References1. WHO. Assessment of the safety of artemisinin compounds in pregnancy. Report of two informal
consultations convened by WHO in 2002. Geneva: WHO; 2003.
2. WHO, TDR. Assessment of the safety of artemisinin compounds in pregnancy: report of two joint informal
consultations convened in 2006 by The Special Programme for Research and Training in Tropical Diseases
(TDR) sponsored by UNICEF/UNDP/World Bank/WHO. Geneva: WHO; 2006.
3. Kennedy DL, Uhl K, Kweder SL. Pregnancy exposure registries. Drug Saf. 2004;27(4):215–28.
Chapter 5
90
4. Ka llen BA. Methodological issues in the epidemiological study of the teratogenicity of drugs. Congenit
Anom (Kyoto). 2005; 45(2):44–51.
5. Hernandez‐Diaz S, Hernan MA, Meyer K, Werler MM, Mitchell AA. Case‐crossover and case‐time‐control
designs in birth defects epidemiology. Am J Epidemiol. 2003;158(4):385–91.
6. Adam MP, Polifka JE, Friedman JM. Evolving knowledge of the teratogenicity of medications in human
pregnancy. Am J Med Genet C Semin Med Genet. 2011;157(3):175–82.
7. Andrews EB, Tennis P. Promise and pitfalls of administrative data in evaluating pregnancy outcomes.
Pharmacoepidemiol Drug Saf. 2007;16(11):1181–3.
8. Charlton RA, Weil JG, Cunnington MC, Ray S, de Vries CS. Comparing the General Practice Research
Database and the UK Epilepsy and Pregnancy Register as tools for postmarketing teratogen surveillance:
anticonvulsants and the risk of major congenital malformations. Drug Saf. 2011;34(2):157–71.
9. Colvin L, Slack‐Smith L, Stanley FJ, Bower C. Pharmacovigi‐ lance in pregnancy using population‐based
linked datasets. Pharmacoepidemiol Drug Saf. 2009;18(3):211–25.
10. Irvine L, Flynn RW, Libby G, Crombie IK, Evans JM. Drugs dispensed in primary care during pregnancy:
a record‐linkage analysis in Tayside, Scotland. Drug Saf. 2010;33(7):593–604.
11. Dellicour S, Ter Kuile FO, Stergachis A. Pregnancy exposure registries for assessing antimalarial drug
safety in pregnancy in malaria‐endemic countries. PLoS Med. 2008;5(9):e187.
12. Corbell C, Katjitae I, Mengistu A, Kalemeera F, Sagwa E, Mabirizi D, et al. Records linkage of electronic
databases for the assessment of adverse effects of antiretroviral therapy in sub‐ Saharan Africa.
Pharmacoepidemiol Drug Saf. 2012;21(4):407–14.
13. Fellegi IP, Sunter AB. A theory for record linkage. J Am Stat Assoc. 1969;64(328):1183–210.
14. Jaro MA. Probabilistic linkage of large public health data files. Stat Med. 1995;14(5–7):491–8.
15. Meray N, Reitsma JB, Ravelli AC, Bonsel GJ. Probabilistic record linkage is a valid and transparent tool to
combine dat‐ abases without a patient identification number. J Clin Epidemiol. 2007;60(9):883–91.
16. Newcombe H, Kennedy J. Record linkage: making maximum use of the discriminating power of identifying
information. Commun ACM. 1962;5(11):563–6.
17. Brasseur P, Badiane M, Cisse M, Agnamey P, Vaillant MT, Ol‐ liaro PL. Changing patterns of malaria
during 1996–2010 in an area of moderate transmission in southern Senegal. Malar J. 2011; 10:203.
18. Enel C, Pison G, Lefebvre M. De l’accouchement traditionnel a l’accouchement moderne au Senegal.
Cahiers Sante. 1993;3: 441–6.
19. Pison G, Kodio B, Guyavarch E, Etard J‐F. La mortalite ma‐ ternelle en milieu rural au Senegal.
Population. 2000;55(6): 1003–18.
20. Olliaro PL, Delenne H, Cisse M, Badiane M, Olliaro A, Vaillant M, et al. Implementation of intermittent
preventive treatment in pregnancy with sulphadoxine/pyrimethamine (IPTp‐SP) at a district health
centre in rural Senegal. Malar J. 2008;7:234.
21. US Centers for Disease Control and Prevention. Registry PlusTM
22. Link Plus technical information and installation. Atlanta (GA): CDC; 2006 [online].
http://www.cdc.gov/cancer/npcr/tools/registry plus/lp_tech_info.htm. Accessed 24 Apr 2013.
23. Jurczyk P. Fine‐grained record integration and linkage tool. Open source record linkage software. 2.1.4 edn.
Atlanta (GA), 2009.
24. Jurczyk P, Lu JJ, Xiong L, Cragan JD, Correa A. Fine‐grained record integration and linkage tool. Birth
Defects Res A Clin Mol Teratol. 2008;82(11):822–9.
25. Stanton C, Lawn JE, Rahman H, Wilczynska‐Ketende K, Hill K. Stillbirth rates: delivering estimates in 190
countries. Lancet. 2006;367(9521):1487–94.
26. NPCR. Link Plus Development Priority List (updated January 26, 2011). 2011.
http://www.cdc.gov/cancer/npcr/tools/registryplus/ lp_features.htm. Accessed 7 Feb 2012.
27. McGready R, Lee SJ, Wiladphaingern J, Ashley EA, Rijken MJ, Boel M, et al. Adverse effects of falciparum
and vivax malaria and the safety of antimalarial treatment in early pregnancy: a population based
study. Lancet Infect Dis. 2012;12(5):388–96.
Probabilistic Record linkage to assess antimalarials safety in pregnancy
91
28. Desai M, Dellicour S. Effects of malaria and its treatment in early pregnancy. Lancet Infect Dis.
2012;12(5):359–60.
29. Gomes M. New global pregnancy register established. 2009 [online].
http://www.who.int/maternal_child_adolescent/
news_events/news/2009/pregnancy_register/en/index.html. Acce‐ ssed 15 May 2012.
30. MiPc. Malaria in Pregnancy Consortium. 2012. http://www.mip‐ consortium.org/. Accessed 2012.
31. Romitti PA, Watanabe‐Galloway S, Budelier WT, Lynch CF, Puzhankara S, Wong‐Gibbons D, et al.
Identification of Iowa live births in the Agricultural Health Study. Arch Environ Occup Health.
2010;65(3):154–62.
32. Roos LL Jr, Wajda A, Nicol JP. The art and science of record linkage: methods that work with few
identifiers. Comput Biol Med. 1986;16(1):45–57.
92
Chapter5Appendix&SupportingInformation
Text S1. Description of the steps used for data preparation before record linkage.
Text S2. Description of record linkage optimisation using Link‐Plus.
Probabilistic Record linkage to assess antimalarials safety in pregnancy
93
TextS1.Descriptionofthestepsusedfordatapreparationbeforerecordlinkage.
Semi‐structuring:
Age was transformed to year of birth as the data covered a 4 year period and some records only included year of birth.
First names with more than one name or hyphenated were parsed into first name, second name, and third name. The order of names was standardized so that Christian name was set as first name and Senegalese names as second/third names. In the study population it is not uncommon for people to have multiple first names (one Christian/European name and a Senegalese name) and the order by which they were used vary by occasion or the preference of the person recording the information.
Standardisation:
Spelling of common names was standardised (e.g. Lucy and Lucie). Standardisation of names was particularly important as the available phonetic algorithms commonly used for record linkage (e.g. Soundex and NYSIIS) are based on English names and would therefore have limited application for French and Senegalese name with non‐English pronunciation.
Abbreviation and nicknames were replaced by full name were possible (e.g. Fatou and Fatoumata or Cons and Constance).
Missingvalues:
Missing values can considerably affect the matching success. Significant missing values were found for age/date of birth (33%) and address (10%) in the delivery registry (see table 2). As data on ANC visits is available in the delivery register (including date of each visit), records with missing information were manually crossed‐check with the ANC registers and missing data extracted were available.
To avoid exposure misclassification, women who attended ANC outside of the catchment areas were excluded (n=149), as these women could have sought treatment and care for malaria elsewhere as well.
Standardfileformat:
After performing the transformations outlined above the dataset files were saved to standard file format (CSV) for import into the record linkage software.
Deduplication
Deduplication of the delivery register was performed as some women could have multiple pregnancies within the 4 years covered and to enable one‐to‐many matching. This was performed using the deduplication function of Link‐Plus. Following identification of duplicates, the files were transformed in order to have one record per individual in SPSS (v18).
Chapter 5 Appendix
94
TextS2‐DescriptionofrecordlinkageoptimisationusingLink‐Plus.With the linking variables fixed, different linkage set ups with varying matching methods and weights for each matching variables were compared to identify the optimal parameters. The linkage variables, matching methods and associated linkage weights are presented in Table i.
Link‐Plus
For optimisation purposes the matching score threshold was set to 0 (minimal cut‐off allowed) in order to produce histograms for each matching set up. Histograms were derived from the distribution of matches, uncertain and non‐matches according to the matching scores for the different set ups. Pairs were assigned as uncertain matches if first name(s) matched, year of birth was within 10 years of each other but the address and/or surname did not match. Surname and address can change over time, the former in case of marriage (which is likely to occur during child bearing years) and address if the person moves home after marriage or for other reasons. Youden indexes were derived for each option by adding the sensitivity and specificity and subtracting 1, to provide an indication of suitability of the proposed cut off values.
Option A: matching variable weights were estimated by the expectation‐maximization (EM) algorithm through maximum likelihood estimation based on the current data. This resulted in 50 true matches, 127 non‐matches and 96 uncertain matches. The histogram (figure i) below shows that the distribution of matches is spread and overlaps with non‐matches between scores 3.2 to 10.7. There is no clear cut off point for assigning true match & non‐match. Setting a cutoff point at 7 (as per the other set ups, Youden index 0.69), the PPV was estimated at 78%.
Figure iv. Option A histogram‐ linkage weights based on expectation‐maximization algorithm
Option B: The matching approach was adjusted by increasing the weight assigned to first name, decreasing the weight assigned to surname, address and the year of birth. This resulted in 67 true matches, 197 non‐matches and 142 uncertain matches with PPV of 82% (Youden index 0.81). This option enables a better discrimination between matches and non‐matches, with potential cut‐off point around a score of 7 (see figure i). The high number of false negatives (n=13) and the high number of false positive (n=8) suggest clerical review would be required between score 5.2 and 8.
0
5
10
15
20
25
30
0
0.3
0.8
1.3
1.9
3.4
3.8
4.2
4.7
5.1
5.4
5.7 6
6.3
6.8
7.1
8.1
8.5
9.4
10.1
10.6
11.1
11.8
12.6
14.6
Number of Pairs
Linkage Probability Score
Match
Non‐Match
Uncertain
Probabilistic Record linkage to assess antimalarials safety in pregnancy
95
Figure v. Option B histogram‐ adjusted linkage weights and normalised linkage variables.
Option C: To assess the benefit of using normalised names and address, the same matching Option as B was run using raw data for first name, surname and address. This resulted in much lower number of true matches (n=36), 180 non‐matches and 150 uncertain matches for clerical review. This Option has the lowest PPV (71%, Youden Index 0.82) and the highest number of uncertain matches (41% of pairs reviewed). Figure iii below shows that discriminating power is lost, with a much wider spread of matches and overlapping non‐matches.
Figure vi. Option C histogram‐ adjusted linkage weights and raw linkage variables.
0
5
10
15
20
25
30
0
0.5
1.7
2.1
2.6
2.9
3.2
3.6
4.2
4.6 5
5.3
5.6 6
6.3
6.7
7.1
7.5
7.8
8.3
8.8
9.1
9.4
9.9
10.2
10.9
12.2
13.6
Number of Pairs
Linkage Probability Score
Match
Non‐Match
Uncertain
Potential cut off
0
5
10
15
20
25
30
0
0.3
0.8
1.3
1.9
3.4
3.8
4.2
4.7
5.1
5.4
5.7 6
6.3
6.8
7.1
8.1
8.5
9.4
10.1
10.6
11.1
11.8
12.6
14.6
Number of Pairs
Linkage Probability Score
Match
Non‐Match
Uncertain
Potential cut off
Chapter 5 Appendix
96
Option D: As the use of multiple/hyphenated names is common in the study population but their use is irregular, the inclusion of second name as a linkage variable was assessed. This resulted in 71 true matches, 219 non‐matches and 93 uncertain matches for clerical review with a PPV of 86% (Youden Index 0.78). Figure iv below shows improved discriminating power.
Figure vii. Option D histogram‐ adjusted linkage weights and normalised linkage variables including middle name.
Table i below summaries the outputs of the different linkage Options. The optimal option should have the least overlap for true and false matches to minimise the number of pairs which have to be reviewed manually. In this case set up D provided the highest number of matches and best positive predictive value (PPV) although set up C has the highest sensitivity.
Table i. Description of linkage parameters and matching result for each set up in Link‐Plus.
Linkage Variables Weight Option A Option B Option C Option D
First Name 0.96 (normalised) 0.97 (normalised) 0.97 (raw) 0.97 (normalised)
Second Name NA NA NA 0.7(normalised)
Surname 0.97 (normalised) 0.85 (normalised) 0.85 (raw) 0.85 (normalised)
Address 0.95 (normalised) 0.85 (normalised) 0.85 (raw) 0.85 (normalised)
Year of birth 0.95 0.2 0.2 0.2
Linkage Output (Score range)
Pairs for Review 269 366 366 383
Match 50 (score 3.2‐14.9) 67 (score 2.4‐13.6) 36 (score 1.7‐12.8) 71 (score 6.2‐14.7)
Uncertain 92 (score 0.1‐14.6) 142 (score 1.7‐10.5) 150 (score 1.4‐10.4) 93 (score 1.1‐11.4 )
Non‐match 127 (score 0‐10.7) 197 (score 0.3‐8.0) 180 (score 0.3‐8.0) 212 (score 0.2‐10.1)
Linkage quality Threshold=7 (YI 0.69) Threshold=7 (YI 0.71) Threshold=7(YI 0.82) Threshold=7 (YI 0.78)
Positive Predictive value 78% 82% 71% 86%
Sensitivity 78% 88% 89% 79%
Specificity 91% 93% 93% 96%
0
5
10
15
20
25
30
0.2
0.9
1.7
3.1
3.5
3.8
4.2
4.5
4.9
5.3
5.8
6.2
6.5
6.8
7.2
7.6
7.9
8.3
8.6
9.4
9.8
10.2
10.6 11
11.4 12
13.1 14
(bla…
Number of Pairs
Linkage Probability Score
Match
Non‐Match
Uncertain
Potential cut off
97
Chapter6
ExploringRiskPerceptionandAttitudestoMiscarriageandCongenitalAnomalyinRuralWesternKenya
StephanieDellicour1,2,MeghnaDesai1,3,LindaMason2,BeatriceOdidi1,GeorgeAol4,PenelopeA.Phillips‐Howard2,KaylaF.Laserson3,5,FeikoO.terKuile2
1 Malaria Branch, Kenya Medical Research Institute (KEMRI)/ Centers for Disease Control
and Prevention (CDC) Research and Public Health Collaboration, Kisumu, Nyanza Province,
Kenya, 2 Department of Clinical Sciences, Liverpool School of Tropical Medicine, Liverpool,
Merseyside, United Kingdom, 3 Center for Global Health, Centers for Disease Control and
Prevention, Atlanta, Georgia, United States of America, 4 International Emerging Infection
Branch, Kenya Medical Research Institute/ Centers for Disease Control and Prevention
Research and Public Health Collaboration, Kisumu, Nyanza Province, Kenya Research and
Public Health Collaboration, Kisumu, Nyanza Province, Kenya, 5 Health and Demographic
Surveillance System Branch, Kenya Medical Research Institute/ Centers for Disease
Control and Prevention Research and Public Health Collaboration, Kisumu, Nyanza
Province, Kenya
PLoS One 2013, 8:11
Copyright© This is an open‐access article distributed under the terms of the Creative
Commons Attribution License, which permits unrestricted use, distribution, and
reproduction in any medium, provided the original author and source are credited
Chapter 6
98
Abstract
BackgroundUnderstanding the socio‐cultural context and perceptions of adverse pregnancy outcomes is
important for informing the best approaches for public health programs. This article describes the
perceptions, beliefs and health‐seeking behaviours of women from rural western Kenya regarding
congenital anomalies and miscarriages.
MethodsTen focus group discussions (FGDs) were undertaken in a rural district in western Kenya in
September 2010. The FGDs included separate groups consisting of adult women of childbearing age,
adolescent girls, recently pregnant women, traditional birth attendants and mothers of children with
a birth defect. Participants were selected purposively. A deductive thematic framework approach
using the questions from the FGD guides was used to analyse the transcripts.
ResultsThere was substantial overlap between perceived causes of miscarriages and congenital anomalies
and these were broadly categorized into two groups: biomedical and cultural. The biomedical causes
included medications, illnesses, physical and emotional stresses, as well as hereditary causes.
Cultural beliefs mostly related to the breaking of a taboo or not following cultural norms. Mothers
were often stigmatised and blamed following miscarriage, or the birth of a child with a congenital
anomaly. Often, women did not seek care following miscarriage unless there was a complication.
Most reported that children with a congenital anomaly were neglected either because of lack of
knowledge of where care could be sought or because these children brought shame to the family
and were hidden from society.
ConclusionThe local explanatory model of miscarriage and congenital anomalies covered many perceived
causes within biomedical and cultural beliefs. Some of these fuelled stigmatisation and blame of the
mother. Understanding of these beliefs, improving access to information about the possible causes
of adverse outcomes, and greater collaboration between traditional healers and healthcare
providers may help to reduce stigma and increase access to formal healthcare providers.
Community perceptions of miscarriage and congenital anomalies
99
IntroductionIn Kenya, as in most countries of sub‐Saharan Africa, there is a lack of information on the risk of
adverse pregnancy outcomes. Reviews using regional estimates project the risk of miscarriages and
stillbirths in 2007 to be 12.2% and 3.3% per pregnancy, respectively [1,2]. Appropriate and prompt
management of adverse pregnancy outcomes can reduce maternal mortality. For example, the two‐
fold higher risk of maternal death following miscarriage compared to those who had a live‐birth
[3,4] mainly reflects poor management of retained products and subsequent infection.
Globally, it is estimated that every year over three million children die due to congenital
anomalies (defined as structural or functional anomalies which occur during the embryo‐fetal
development and are present at the time of birth). In addition, another three million babies born
with a congenital anomaly and do not access care at birth may be disabled for life [5]. The risk of
major congenital anomalies in industrialised countries is around 3‐5% [6,7]. It is unclear if this is
similar in low income countries or if it is higher due to a variety of factors such as different
health‐seeking behaviours, poor nutrition, a higher prevalence of infectious diseases, weak health
systems and a higher availability of a wide range of prescription‐only drugs over‐the‐counter
[5,8,9]. This availability of drugs, and the high prevalence of infectious disease, means that
pregnant women are more likely to be exposed to teratogenic drugs or drugs with limited
information on their safety for pregnant patients. Further, misuse of drugs for induced abortion
also has political ramifications, particularly in countries with strict abortion laws and limited
access to contraceptive and family planning services. Due to limited information and knowledge,
infants born with a congenital anomaly might miss the opportunity to have access to appropriate
care in time and suffer the social and economic burden of lifelong disabilities. Receipt of
appropriate care significantly impacts the prognosis of newborns with a congenital malformation;
for children born with club foot, for example, this can be the difference between a life with limited
mobility (and often loss of social and economic opportunities) and that of normal active life
with painless functional feet.
Understanding the local perception of health, stigma, and disclosure is essential to improve
access to health services and for patient uptake of new programmes [10,11]. However, women’s
perceptions, clients and service providers understanding and attitudes towards adverse
pregnancy outcomes, and how these affect health‐seeking behaviour, have received scant
research attention. This paper describes the results from a qualitative study that explored the
perceived causes of miscarriages and congenital anomalies, stigma associated with these adverse
outcomes, and issues around disclosure and health‐seeking behaviour.
MethodsThis study was part of a formative research program carried out to inform the best approach to
setting up a prospective pharmacovigilance pregnancy cohort to monitor the use of medications
during pregnancy, the safety of these drugs in pregnancy, and their impact on birth outcomes.
There was no teratogenic drug monitoring program at the time we conducted focus group
discussions in the study area. Topics around the socio‐cultural context of pregnancy and adverse
pregnancy outcomes were explored.
Chapter 6
100
StudysiteThe study took place in Rarieda District, Nyanza Province, in western Kenya (within Siaya County
since 2013). This area has been under continuous surveillance as part of the Kenya Medical
Research Institute and the US Centers for Disease Control and Prevention (KEMRI/CDC) Health
and Demographic Surveillance System (HDSS) since 2001 [12]. Subsistence farming is the main
occupation in this area and more than 95% of the population are from the Luo ethnic group [13].
The prevalence of malaria (KEMRI/CDC, unpublished data), HIV [14] and TB [15] are high. Malaria
transmission is intense and holo‐endemic with peaks after the two rainy seasons usually in June‐
July and November‐December. HIV prevalence in Nyanza province is 13.9% (16.0% among women
and 11.4% among men) compared to a national prevalence of 6.3% [14]. HIV prevalence among
pregnant women is around 27% [16]. The total fertility rate in this area calculated for 2009 was 4.3
live births per woman [17]. Only 17% of females have had a secondary school education, while
19% have neither primary nor secondary school education [17].
ParticipantsFemales, aged 15 years and older, purposively sampled by village based KEMRI/CDC fieldworkers
(village reporters; VRs) were invited to be part of the focus group discussions (FGDs). VRs are
community members involved with sensitisation activities for new initiatives or research projects
in their villages. Prior to participant selection, VRs were trained on the aims of the formative
research and the characteristics of the participants required for each FGD by the study moderator
(B.O.), a Kenyan social scientist and the note‐taker. The different FGD groups consisted of adult
women of childbearing age (WOCBAs; 18‐49 years), adolescent girls (15‐18 years), recently or
currently pregnant women, traditional birth attendants (TBAs; locally called Nyamrerwas) and one
group of mothers of children born with a congenital anomaly. Each group consisted of no more
than one woman from each village. The separate groups were selected to have fairly homogenous
demographic characteristics and avoided dominance from older women, enabling freer
discussions. A total of 10 focus groups (n=90 participants) were conducted between September
1st and September 22nd, 2010.
StudyproceduresThe FGDs took place in central locations such as schools, or churches, that were easily accessible
to all participants, and where privacy could be maintained. Information on the pharmacovigilance
study and the formative research was discussed at community meetings with village chiefs and
counsellors as well as with a community advisory board (CAB, comprising stakeholders and
members of the community who provide advice to researchers on proposed studies). Individual
participants gave consent after the information sheet had been verbally explained and discussed.
FGD guides were semi‐ structured, open‐ended and probing. The discussion guides were initially
written in English and then translated to Dholuo, the local dialect, by the Kenyan social scientist
and moderator. The FGD guides covered topics relating to pregnancy recognition, disclosure,
health‐seeking behaviour and pregnancy related behaviour change as well as use of medication
and herbal remedies during pregnancy, practices around delivery, perception of adverse
outcomes and unwanted pregnancies. Not all topics were covered by all groups due to time
limitations. The FGDs ranged from one to two hours. The moderator and note‐taker, both
females from the study area, had many years of fieldwork experience, including FGDs, and were
trained on the study protocol and tools before the first FGD. All FGDs were recorded using
audiotapes. Audio files were transcribed verbatim and translated by the moderator and two
Community perceptions of miscarriage and congenital anomalies
101
independent transcribers; this activity was guided by the notes taken during the FGDs. Each
transcript was reviewed by the moderator to ensure consistency and accuracy was maintained.
DataanalysisA deductive thematic framework approach was used to provide a detailed account of a group
of themes using the questions from the FGD guides on perception, attitudes, and risk factors
associated with adverse pregnancy outcomes within the data. Each transcript was uploaded to
QSR Nvivo 9 software (QSR International Pty Ltd, Melbourne, Australia). The code structure was
developed according to the topic guide and the codes were applied to each transcript in a
systematic fashion across the entire dataset by collating data relevant to each code. Additional
codes were identified inductively from the transcripts. The coding was done independently by
S.D. and L.M. and comparisons made; any areas of disagreement were discussed and resolved on
revisiting the data. Thematic maps were generated to visualise the main themes and sub‐ themes
and how they connected. Tree maps were generated using Nvivo to compare coding references
between the different sub‐groups. Table S1 provides a list of the main themes and sub‐themes.
EthicsThe protocol and consent procedures were reviewed and approved by 1) the Kenya Medical
Research Institute (KEMRI, Nairobi, Kenya) National Ethics Review Committee; 2) the US Centers
for Disease Control and Prevention (CDC, Atlanta, GA, USA) IRB and 3)the Liverpool School of
Tropical Medicine (LSTM, UK) Research Ethics Committee. Informed consent was obtained verbally
in the local dialect (Dholuo) and tape recorded. This informed consent procedure was approved
by all 3 ethics committees. This is common procedure for FGDs as written consent can sometime
make participant wary or guarded, particularly in a setting were illiteracy is common and would
require finding an appropriate witness.
ResultsTable 1 provides a summary of the participants’ characteristics. The findings are presented as
three main themes relating to the specific research questions: 1) Perceived causes of miscarriages
and congenital anomalies, 2) stigma and the community perception of miscarriages and congenital
anomalies, and 3) health‐seeking behaviours for miscarriages and congenital anomalies. There was
no major difference observed in the beliefs and themes that emerged between the different
groups (i.e. WOCBAs compared with adolescents or TBAs or mothers of children born with a
congenital anomaly) and the findings are consequently presented jointly for all groups.
No definition of congenital anomaly was provided to the participants but a wide range of
anomalies, both structural and functional, were cited when participants were asked about the type
of congenital anomalies they had observed in the community. The most frequently mentioned
anomaly was deformity of the hand, followed by mental retardation, and paralysis. Deafness, cleft‐
lip and eye anomalies were also mentioned by different groups. Explicit anomalies were not
specified during the discussion on causes, stigma and health seeking behaviours; therefore we
refer to congenital anomalies in general in the text.
Chapter 6
102
Table 1. Summary of Focus Group Discussions Participant Characteristics.
FGD# Group N Average Age in years
Average Gravidity (range)
Marital status Education Single Married Widowed None Primary Secondary
FGD1 WOCBA 10 38 7 (3‐10) 0 9 (90%) 1 (10%) 0 7 (70%) 3 (30%)
FGD2 WOCBA 9 31 5 (1‐9) 0 9(100%) 0 0 8 (89%) 1 (11%)
FGD3 Recently Pregnant
8 32 6 (2‐10) 0 7 (88%) 1 (13%) 0 8 (100%) 0
FGD4 TBA/
Nyamrerwas 10 54 7 (4‐12) 0 4 (40%) 6 (60%) 3 (30%) 2 (20%) 5 (50%)
FGD5 Adolescent 9 16 0 9 (100%) 0 0 0 9 (100%) 0
FGD6 Adolescent 9 17 0 9 (100%) 0 0 0 9 (100%) 0
FGD7 Recently Pregnant
9 28 4 (2‐7) 0 9 (100%) 0 0 8 (89%) 1 (11%)
FGD8 TBA/
Nyamrerwas 8 41 6 (2‐10) 0 8 (100%) 0 1 (13%) 7 (88%) 0
FGD9 WOCBA 9 31 5 (1‐10) 1 (11%) 6 (67%) 2 (22%) 1 (11%) 6 (67%) 2 (22%)
FGD10
Mothers of children with congenital anomalies
9 31 5 (1‐10) 2 (22%) 6 (67%) 1 (11%) 1 (11%) 8 (89%) 0
Overall 90 32 5 (0‐12) 21 (23%) 58 (64%) 11 (12%) 6 (7%) 72 (80%) 12 (13%)
Acronyms: WOCBA‐ women of childbearing age; TBA‐ traditional birth attendant
PerceptionofcausesofmiscarriagesandcongenitalanomaliesMany different possible causes of miscarriages and congenital anomalies were reported with a
significant overlap between these two outcomes. We divided these into two broad groups: those
with a biomedical basis and those with a cultural basis.
1: Biomedical Causes. The role and potential dangers of biomedical causes were thematically
sub‐grouped into medications, illness, physical and emotional stress, and hereditary causes.
Most women reported that certain drugs could lead to miscarriages or congenital anomalies, but
only if they were not prescribed, or if the dose taken was above that prescribed by a healthcare
worker. There was an overarching theme of trust by the women that any treatment given or
prescribed by clinicians would be safe. One participant mentioned that a pregnant woman should
only take medicines if the illness is confirmed. This infers a natural understanding of the concept
of balancing the risks and benefits of treatment (i.e. it is only worth taking the risk of consuming
medicines if one is truly ill) by this participant. When asked about reservations over the ingestion
of drugs in pregnancy, two separate groups brought up the issue of nausea and heightened
sensitivity to smell (i.e. relating this to difficultly in taking drugs orally during pregnancy). Thus, it
was suggested that pregnant women should be given injections rather than tablets. This implies
that reluctance to take medication in pregnancy is associated with pregnancy related nausea,
rather than a fear of an adverse effect on the pregnancy or the fetus. Only one participant
mentioned fear of side‐effects (itching and drowsiness for the pregnant women) as a reason for
not taking the drugs given to her during antenatal care.
“I’m requesting that if one is pregnant you should avoid the drugs, it should just be
injection given when one is sick. They don’t take drugs. The drugs that I was given in 1997
when I gave birth just expired the other day. So it should be the injection which you come
and get.” (FGD2, P9)
Community perceptions of miscarriage and congenital anomalies
103
“When some women go to the clinic and are given medicine because when you go to the
clinic you are given medicine for blood, you are given medicine that would give energy…when
they are given such medicine, they do not take saying that it smells bad.” (FGD1, P6)
“A pregnant woman should not just take medicine if it is not prescribed by the doctor.”
(FGD3, P1)
Family planning drugs were considered to be distinct types of drugs, separate from medications
used for illness, and all groups believed that these could lead to congenital anomalies.
“You find that when you give birth to a child with a defect then people would say it is
because of the family planning that has caused it.” (FGD10, P2)
Other drugs also deemed to be potentially harmful for the pregnancy were antimalarials,
including chloroquine (which is no longer recommended for treatment due to malaria parasite
resistance ), quinine (which is considered safe for use in all trimesters of pregnancy and used for
severe and complicated malaria which in itself can cause miscarriage), and artemether‐
lumefantrine (which is the first line treatment for malaria but not recommended for use in the first
trimester of pregnancy due to its unknown safety). Chloroquine and quinine were drugs reported
to be used by pregnant women within the community to induce abortions for unwanted
pregnancies. Anti‐retroviral drugs were mentioned by one participant as having the potential to
cause congenital anomalies. One participant noted antibiotics could hinder bone development of
the fetus.
“[M: Which other way would you use to do abortion?]In some cases I hear people overdose.
[What overdose?] For the tablets, there are some drugs if they take like the malariaquine
[chloroquine] if you overdose then you will abort.” (FGD7, p5)
When asked about the potential risk period for taking medication during pregnancy only one
participant mentioned that drugs should be avoided during the first three months of pregnancy.
The majority mentioned drugs should be avoided closer to the time of delivery.
Disease or infections were mentioned by all groups as a potential cause of miscarriage. Most
participants simply cited ‘disease’ without being specific. However, malaria was mentioned by a
few participants, and one person cited HIV. Other illnesses, particularly sexually transmitted
infections (gonorrhoea was specifically mentioned), and measles, were commonly cited causes of
congenital anomalies.
“Like gonorrhoea maybe you were sick and you don’t seek treatment, so if you gave birth to
a child he/she must be disabled in some way.” (FGD9, P1)
Participants considered that adverse pregnancy outcomes were the consequence of pregnant
women not attending antenatal care services, or not completing their vaccinations (supposedly
tetanus vaccine, which is provided routinely through ANC). This could be interpreted as a
perception that pregnant women are vulnerable to adverse effects from an illness, which
would have been prevented, had the woman attended ANC services.
Chapter 6
104
Carrying out strenuous work while pregnant was mentioned by all groups as a potential cause of
miscarriage. A few respondents also thought physical trauma (such as trauma near or around the
uterus) could lead to miscarriage. Women also considered that adverse effects were more likely
among women who had conceived many times, suggesting women understood the physical toll
of many pregnancies and short birth intervals between pregnancies. Similarly, the position of
the fetus in the uterus was reported by one TBA as a potential cause of deformity. Emotional distress
such as being shocked by bad news or arguing with a husband/partner was considered to be
dangerous for the pregnancy, and could lead to pregnancy loss. Being raped was also mentioned as a
cause of miscarriage.
“Miscarriage can come when you fight with someone and you are hit in the wrong place.”
(FGD10, P9)
“One can miscarry if you are shocked by bad news.” (FGD10, P4)
Hereditary transmission (referred to as “hereditary” or “inherited trait”) was a prominent sub‐
theme for congenital anomalies. Many women reported that the mother of a child with a
congenital anomaly is often thought to have conceived with a man other than the husband, since
the husband did not appear to have, or physically display, the same congenital anomaly. This
highlights an underlying theme of mistrust and blaming of mothers for any adverse pregnancy
outcome (explored more below under “cultural beliefs”). Incest was also cited as a cause for
congenital anomalies and miscarriages. Incest was reported as a reason to kill a baby born
with a congenital anomaly by one group.
“Sometimes they would say you conceived with a relative, so most people normally kill them
[child with congenital anomaly].” (FGD6, P7)
2: Cultural Beliefs. There were many beliefs around the causes of miscarriage and congenital
anomalies including extra‐marital sex, not respecting traditional ways, and being cursed or
possessed.
A strong theme for causing both miscarriage and congenital anomalies was infidelity, where the
woman conceived outside of wedlock. This includes being raped, although this could also be
associated with the physical strain and emotional distress caused, if the husband cheated on the
wife, and as a consequence to incest.
“I have heard one but I’m not sure about it, that if you spend [a night] with a man other than
your husband when pregnant then you can abort.” (FGD9, P9)
“For a baby born with a defect, people would think how possible it is especially if the mother
has no defect. Someone would think how she can give birth to a child with a defect like
having only one eye. She would think that perhaps this might have come as a result of going
out with other men.” (FGD8, P6)
Women in different groups reported that if their husband either had an extra marital affair or had
slept with an inherited wife, then the current wife would be more likely to miscarry or bear a child
with a congenital anomaly.
Community perceptions of miscarriage and congenital anomalies
105
Not conforming to traditional rules was mentioned as a cause of miscarriage and congenital
anomalies. Mostly this related to not performing traditional rituals surrounding marriage,
becoming pregnant before the husband paid the bride‐price, or building a new house while the wife
is pregnant. For the latter it was not clear whether this was a consequence of physical
exhaustion of the pregnant woman, or the breaking of a specific traditional taboo. Other beliefs
that could all lead to adverse pregnancy outcomes included the breaking of the taboo around
sharing the cooking place with a grandmother, taking on the responsibilities of elders before
reaching this senior position, planting when not the owner of the field, or eating meat from a
pregnant cow. Many mentioned that adverse outcomes occurred because the woman or her
family broke a taboo that was prohibited by the clan and had been cursed. Being possessed by
demons was considered another possible cause of a miscarriage or a congenital anomaly.
“If she did something prohibited in her clan and then she gives birth to a deformed child then
the community would say it came as a result of committing a taboo.” (FGD4, P4)
“A curse can fall upon you depending on how you live in the home or according to some
cultural practices that you are supposed to do.” (FGD9, P5)
The above quotation indicates a very strong relationship between the breaking of taboos, against
cultural norms, and subsequent adverse pregnancy outcomes. These taboos appear to be deeply
rooted within the cultural belief systems, and across all age groups. Although the community
under study are largely Christian, and very religious, only a few suggested that adverse pregnancy
outcomes happen without explanation or have a spiritual/religious explanation as “God’s plan.”
“There is malformation that comes as a result of God’s creation; you can give birth to a child
without knowing that it will have a malformation. So that is the will of God.” (FGD8, P5)
StigmaassociatedwithmiscarriagesandcongenitalanomaliesA number of the causes given were associated with blame, either on the mother or her family,
such as a woman’s infidelity to her husband, or breaking a traditional taboo. This could lead to
distrust between husband, mother‐in‐law and the wife/ mother. Women reported that mothers
are thus stigmatised rather than supported after going through the distress of having an adverse
pregnancy outcome. Some reported that women who suffered from a miscarriage were
ostracised and not allowed to come out of their home until they had been cleansed by a spiritual
healer. Other pregnant women were also not able to come into contact with them as this might be
passed on and cause them also to miscarry.
“In our place, I normally see when one has miscarriage they are not allowed to go out. They
say that you stay in the house until maybe the church members come and clean you that is
when you can move out. So the reason why they don’t want people to go out and I don’t
know what would happen if you went out.” (FGD7, p8)
Neglect of children born with abnormalities was reported by a number of groups. Children with any
congenital anomaly/ disability were seen as not “useful” to the family and an extra burden, as
Chapter 6
106
well as stigmatized. This stigma might imply some taboo had been broken. Many respondents said
such children are often hidden in homes or even killed. Only three women mentioned that they
would love and provide the extra‐care the child needed.
“You would not love such kind of a child [with congenital anomaly]; I would not love the child
because I would feel that he will not be of any help to me. And then taking care of him would
be hard.” (FGD9, P5)
“I would love the child [born with a birth defect/disability] because he/she is mine; I’m the
one who gave birth to her/him.” (FGD9, P1)
“Sometimes the grandmother speaks a lot and she might say that the child is not of her
blood.” (FGD7, P7)
Health‐seekingbehaviourParticipants reported that women seek care either from a health facility or from a traditional
healer such as a TBA, herbalist, spiritual healer or “Jarwecho” (a special healer that deals with
skeleton issues and bone fracture). Women also may seek advice from family members such as a
grandmother, mother‐in‐law and/or the husband. Women reported that they wanted to seek care
from health facilities so that they could be treated in the event of complications (i.e. when there
is excessive bleeding), and also so that they could better understand the potential causes of the
adverse outcome. Many women reported they would go to the TBA to get herbs or go to see a
spiritual healer who would “pray for them” or cleanse them of the curse. In the case of a
miscarriage, it seems women would only seek care if there were perceived complications. Most
women would otherwise keep the miscarriage a secret. Many women reported that they, and
other women, would not seek care if their baby was born with a congenital anomaly.
M: “When they miscarry, do they seek for any care?”
P7: “They don’t seek for any care when they miscarry.”
P1: “Sometimes they go to the hospital.”
P7: “Some people would not like others to know if they have miscarried.” FGD8
“From what I hear, there is nothing one can do about it [child born with malformation] and
no one can ever give you a piece of advice on how they can be helped.” (FGD10, P7)
The decision about where to seek care was discussed by one participant, who noted that it
depended on the cost of care and resources available to the family.
DiscussionUsing FGD methodology, we explored the community explanatory models for adverse pregnancy
outcomes and how these explanatory models might influence health‐seeking behaviours. Although
various biologically plausible causes of miscarriage and congenital anomalies were known, cultural
beliefs seem to play a central role in the community perception of these adverse pregnancy
outcomes. Such beliefs can result in stigma and influence health‐seeking behaviours, as reported in
previous studies on relatively minor but disfiguring congenital anomalies [18]. Stigma seems to be
strongly related to the belief that the woman either cheated on her husband or she or her family
Community perceptions of miscarriage and congenital anomalies
107
had broken some traditional taboo. Some of the cultural beliefs around miscarriage and
congenital anomalies lead to stigmatisation, with the mother largely held responsible for the cause
of the malformation; this may have a negative impact on health‐seeking behaviours and
disclosure of such pregnancy outcomes. Women who experienced adverse pregnancy outcomes
are thus often additionally burdened by negative attitudes and stigmatisation from the
community. In particular, mothers of children with congenital anomalies were often blamed for
cheating on their husbands or being cursed because they have apparently transgressed some
cultural taboos. Similar studies from rural India and Nigeria found that parents of disabled children
were socially marginalized because of widely held beliefs that they had broken a taboo [19], and
their children were neglected [20]. Improving access to information about the possible causes of
such adverse outcomes may help to reduce the stigma and shame that these women undergo and
increase access to formal healthcare providers.
These data suggest that mothers of children with congenital anomalies or disabilities do not know
where they can seek help or whether any help is available. This, together with social stigma, may
contribute to the neglected care of these children in this setting. Other factors such as
perception of treatment effectiveness, satisfaction with health care services, and external
barriers (e.g. financial constraints, accessibility of health services) also play an important role in
driving health‐ seeking behaviours. We note that limited disclosure of adverse pregnancy outcomes
could also impact on the accuracy of public health surveillance programmes and could lead to an
under‐estimation of the problem.
Fertility and successful childbearing remain very important factors for status and recognition of a
woman’s worth in society, in many countries, including those in sub‐Saharan Africa. Childbearing
confers social status and strengthens conjugal relationships, and children are seen as contributors
to the daily labour. The presence of children ensures the rights of property and maintains the
family lineage, as described more fully by Dyer [21]. Unsuccessful childbearing and adverse
pregnancy outcomes could thus be a potential cause or driver of gender‐based violence. Gender‐
based violence is widespread in Kenya, and particularly high in Nyanza Province where 60% of
women report having ever experienced emotional, physical, or sexual violence by their husbands
[14]. Although this was not reported in our FGDs, we noted that females in the focus groups
considered violence and rape to be causes of miscarriage. We also recognise that in some
countries, with strict abortion laws and poor access to contraception, and family planning,
women may be driven to take medicine perceived locally to induce abortion, but which may be
potentially teratogenic. This requires further study.
It is not clear how the overlap between biomedical knowledge and cultural beliefs is
incorporated into society’s perception, whether these beliefs are mutually exclusive, or blend
according to different societal pressures. Under the biomedical paradigm, illness and disease are
attributable to biologically plausible causes based on scientific theories such as factors with
biochemical effects. If understanding of ill health is based on a biomedical paradigm, the individual
is more likely to seek medical care from modern rather than traditional sources [22]. On the other
hand, cultural beliefs are based on supernatural phenomenon, traditional values or religious
beliefs. Within this explanatory model individuals would seek help or care from either a
spiritual or traditional healer [22]. However, adverse health outcomes are not always categorised
as one or another. For example Hausmann‐Muela et al [23], reported how in Tanzania there is
Chapter 6
108
a widespread belief that witchcraft can affect biomedical treatment of malaria and impede
detection of malaria parasites. Addressing such beliefs, and acknowledging the role of traditional
healers and birth attendants, should be part of public health and research programs. Creating
collaborative links between traditional and modern medicine, such as by empowering TBAs to
refer their clients to health facilities, is critical to increase access to care in developing countries
[24].
There was an underlying theme of trust among this rural population in western medicine and
healthcare providers. Medications were deemed safe when prescribed and when taken at the
correct dosage as prescribed, and adverse outcomes were considered to be due to a lack of
medical follow up for women who didn’t attend ANC. Similar to other studies, our research
findings suggest that women in rural African settings might suffer from a “white‐coat complex”
where they avoid asking questions to healthcare workers, and assume any prescribed drugs are
safe [25,26]. At the same time, however, pregnant women do not always take medicines given
at the health facilities, not because they fear the potential teratogenic risk, but because pregnancy
predisposes them to nausea. This has important implications for any intervention targeting
pregnant women, and consideration should be given to this to ensure compliance, including the
use of directly observed therapy where feasible. Furthermore, in a context where healthcare
clients do not ask questions and rely on the providers’ judgement, it is particularly important that
healthcare providers who see pregnant (or potentially pregnant) women should be familiar with
all drugs contraindicated during pregnancy and potential teratogens. Whilst women were aware of
the potential risk of using medication during pregnancy, most did not know the potential risk in
the first trimester. The few specific drugs that they did mention (such as quinine and chloroquine)
are safe for use in all trimesters of pregnancy, exemplifying this population’s lack of
knowledge. Others mentioned that antimalarials should be avoided towards the end of
pregnancy in the third trimester. This highlights the need for more information, education and
communication on the risk of medication used during the first trimester of pregnancy and the
importance for women to explicitly disclose their potential pregnancy to healthcare providers.
The widespread belief that modern family planning methods increases the risk of having a child
with congenital anomalies also urgently needs to be addressed, particularly within the context of
a country with a high fertility rate since the 1990s [27]. Family planning interventions that
promote healthy spacing of pregnancy are vital for development and it will be important to ensure
accurate information is available to offset myths which could negatively impact uptake of family
planning programmes. Only about 25% of women in the study area report using a family
planning method (unpublished data), compared to 73% nationally [14].
This study has some possible limitations. Despite efforts to provide a non‐threatening
environment, respondents may have withheld information and some members might have been
affected by a social desirability effect. FGD participants could have been inclined to give what
they thought would be acceptable answers, and may have feared telling the truth, particularly
around local myths and taboos. This is exemplified by the fact that most of the reported themes
under cultural beliefs were reported in the third person (i.e. “it is believed that” rather than “I
believe”). The FGDs were conducted as formative research for setting up a pregnancy
pharmacovigilance study, which may have influenced the moderator and the participant to focus
on issues around drug safety. Although one of the coders (S.D.) is the study coordinator for the
pharmacovigilance study and has a specific interest in perception of drug safety in pregnancy, the
Community perceptions of miscarriage and congenital anomalies
109
second coder was not associated with the pharmacovigilance study. Many topics were included in
the FGDs which limited the depth and details of the information collected. Moreover, the findings
reflect the circumstances affecting a particular population, i.e., women in one rural area in
western Kenya, and might not be generalisable. Although validation of findings representing the
views elicited through 10 FGDs through data triangulation was not possible, investigator
triangulation (where two investigators independently coded the data and compared notes) was
used to enhance credibility of the emerging sub‐themes. This study did not include the perspective
of men or healthcare providers who often play an important role influencing healthcare seeking
behaviours. Future studies including such groups would contribute additional understanding to the
barriers to healthcare seeking for adverse pregnancy outcomes.
ConclusionTo our knowledge, this is the first qualitative study undertaken in Kenya exploring the
perceptions, attitudes and health‐seeking behaviours on adverse pregnancy outcomes. Better
understanding of these concepts can inform strategies to improve women’s health‐seeking
behaviour. Development of appropriate information, education and communication and outreach
materials should inform women of true causes of adverse outcomes. This will also help to reduce
the burden of guilt they feel, the stigmatisation from the community, and provide guidance on
caring for women who have adverse pregnancy outcomes and have children with disabilities. To
support this, there is a need for greater integration and collaboration between traditional healers
and modern medical practitioners, and to ensure culturally acceptable management to better
protect the unborn child and expectant mothers.
AcknowledgementsWe thank all the study participants, the late Beatrice Odidi (moderator), Jane Oiro (note‐taker)
and the village reporters for their diligent efforts in the field. We thank Florence Were and Walter
Mchembere for reviewing the FGD guides. This manuscript is approved by the KEMRI Director.
KEMRI/CDC is a member of The INDEPTH Network. The findings and conclusions in this paper are
those of the authors and do not necessarily represent the views of the Centers for Disease Control
and Prevention.
AuthorContributionsConceived and designed the experiments: SD. Performed the experiments: BO. Analyzed the data:
SD LM. Wrote the manuscript: SD MD LM BO GA PAP KFL FOtK.
References1. Dellicour S, Tatem AJ, Guerra CA, Snow RW, ter Kuile FO (2010) Quantifying the number of
pregnancies at risk of malaria in 2007: a demographic study. PLOS Med 7: e1000221. PubMed: 20126256.
2. Stanton C, Lawn JE, Rahman H, Wilczynska‐Ketende K, Hill K (2006) Stillbirth rates: delivering estimates in 190 countries. Lancet 367: 1487‐1494. doi:10.1016/S0140‐6736(06)68586‐3. PubMed: 16679161.
3. Rahman M, DaVanzo J, Razzaque A (2010) The role of pregnancy outcomes in the maternal mortality rates of two areas in Matlab, Bangladesh. Int Perspect Sex Reprod Health 36: 170‐177. doi: 10.1363/3617010.
4. Reardon DC, Ney PG, Scheuren F, Cougle J, Coleman PK et al. (2002) Deaths associated with pregnancy outcome: a record linkage study of low income women. South Med J 95: 834‐841. doi: 10.1097/00007611‐200208000‐00010. PubMed: 12190217.
Chapter 6
110
5. Howson C, Christianson A, Modell B (2008) Controlling Birth Defects: Reducing the Hidden Toll of Dying and Disabled Children in Lower‐ Income Countries. Washington, DC.
6. Baird PA (1999) Prenatal screening and the reduction of birth defects in populations. Community Genet 2: 9‐17. doi:10.1159/000016177. PubMed: 15178956.
7. Edmonds LD, James LM (1993) Temporal trends in the birth prevalence of selected congenital malformations in the Birth Defects Monitoring Program/Commission on Professional and Hospital Activities, 1979‐1989. Teratology 48: 647‐649. doi:10.1002/tera. 1420480606. PubMed: 8115973.
8. Bale JR, Stoll BJ, Lucas AO (2003) Reducing Birth Defects: Meeting the Challenge in the Developing World. National Academy of Sciences.
9. Penchaszadeh VB (2002) Preventing congenital anomalies in developing countries. Community Genet 5: 61‐69. doi: 10.1159/000064632. PubMed: 14960901.
10. Landrine H, Klonoff EA (1994) Cultural diversity in causal attributions for illness: the role of the supernatural. J Behav Med 17: 181‐193. doi: 10.1007/BF01858104. PubMed: 8035451.
11. Filippi V, Marshall T, Bulut A, Graham W, Yolsal N (1997) Asking questions about women's reproductive health: validity and reliability of survey findings from Istanbul. Trop Med Int Health 2: 47‐56. doi: 10.1046/j.1365‐3156.1997.d01‐126.x. PubMed: 9018302.
12. Odhiambo FO, Laserson KF, Sewe M, Hamel MJ, Feikin DR et al. (2012) Profile: the KEMRI/CDC Health and Demographic Surveillance System‐‐Western Kenya. Int J Epidemiol 41: 977‐987. doi:10.1093/ije/ dys108. PubMed: 22933646.
13. Adazu K, Lindblade KA, Rosen DH, Odhiambo F, Ofware P et al. (2005) Health and demographic surveillance in rural western Kenya: a platform for evaluating interventions to reduce morbidity and mortality from infectious diseases. Am J Trop Med Hyg 73: 1151‐1158. PubMed: 16354829.
14. Kenya National Bureau of Statistics (KNBS) (2010) Macro I. Kenya Demographic and Health Survey 2008‐09. Calverton, Maryland.
15. van't Hoog AH, Laserson KF, Githui WA, Meme HK, Agaya JA et al. (2011) High prevalence of pulmonary tuberculosis and inadequate case finding in rural western Kenya. Am J Respir Crit Care Med 183: 1245‐1253. doi:10.1164/rccm.201008‐1269OC. PubMed: 21239690.
16. Amornkul PN, Vandenhoudt H, Nasokho P, Odhiambo F, Mwaengo D et al. (2009) HIV prevalence and associated risk factors among individuals aged 13‐34 years in Rural Western Kenya. PLOS ONE 4: e6470. doi:10.1371/journal.pone.0006470. PubMed: 19649242.
17. KEMRI/CDC (2009) KEMRI/CDC Health and Demographic Surveillance System Report for 2009. Kisumu: KEMRI/CDC.
18. Olasoji HO, Ugboko VI, Arotiba GT (2007) Cultural and religious components in Nigerian parents' perceptions of the aetiology of cleft lip and palate: implications for treatment and rehabilitation. Br J Oral Maxillofac Surg 45: 302‐305. doi:10.1016/j.bjoms.2006.09.005. PubMed: 17056161.
19. Minhas RS, Hillard R (2007) Maternal Care Practices and Perceptions of Birth Defects in Central India. Univ Tor Med J 84.
20. Okunade AO (1981) Attitude of Yoruba of Western Nigeria to handicap in children. Child Care Health Dev 7: 187‐194. doi:10.1111/j. 1365‐2214.1981.tb00836.x. PubMed: 6456848.
21. Dyer SJ (2007) The value of children in African countries: insights from studies on infertility. J Psychosom Obstet Gynaecol 28: 69‐77. doi: 10.1080/01674820701409959. PubMed: 17538814.
22. Kgwatalala G (2009) Health Seeking Behaviour among the People of the Africa Gospel Church in Francistown. UNISA.
23. Hausmann‐Muela S, Muela Ribera J, Tanner M (1998) Fake malaria and hidden parasites—the ambiguity of malaria. Anthropol Med 5.
24. Leininger MM (1991) Culture Care theory and uses in nursing administration. NLN Publ: 373‐390. PubMed: 1795996.
25. Stokes E, Dumbaya I, Owens S, Brabin L (2008) The right to remain silent: a qualitative study of the medical and social ramifications of pregnancy disclosure for Gambian women. BJOG 115: 1641‐1647; discussion: 10.1111/j.1471‐0528.2008.01950.x. PubMed: 19035940.
26. Brabin L, Stokes E, Dumbaya I, Owens S (2009) Rural Gambian women's reliance on health workers to deliver sulphadoxine – pyrimethamine as recommended intermittent preventive treatment for malaria in pregnancy. Malar J 8.
27. UN (2010) World Population Prospects: The 2010 Revision. Population Division of the Department of Economic and Social Affairs of the United Nations Secretariat.
Community perceptions of miscarriage and congenital anomalies
111
Chapter6Appendix&SupportingInformation
TableS1.Mainthemesandsub‐themesusedinthethematicanalysisforexploringriskperceptionandattitudestomiscarriageandcongenitalanomaly.
Themes Sub‐themes
Causes Biomedical
Illnesses
Medications
Stress
Hereditary
Cultural
Not conforming to traditional norms
Infidelity
Curse
None
Stigma Stigmatisation
Hide child with abnormality
Isolate woman who had a miscarriage
Blame/Gossip
No disclosure
No stigma
Accept child with a congenital malformation
Freely disclose adverse pregnancy outcome
Health seeking behaviours Health facility
Traditional
TBA
Spiritual healer
Herbal remedies
Family
Husband
Parents in law
Grand‐mother
None
112
113
Chapter7
Assessment of Knowledge and Adherence to the NationalGuidelines forMalariaCaseManagement inPregnancyamongHealthcare Providers and Drug Outlet Dispensers in rural,westernKenya
ChristinaRiley1,StephanieDellicour3,4,PeterOuma2,UrbanusKioko5,FeikoO.terKuile2,AhmeddinOmar5,SimonKariuki3,4,AnnM.Buff6,7,MeghnaDesai3,6,JulieGutman6
1 Rollins School of Public Health, Emory University, Atlanta, USA, 2 Liverpool School of Tropical Medicine, Liverpool, United Kingdom, 3 Kenya Medical Research Institute and Centres for Disease Control and Prevention (KEMRI/CDC)Research and Public Health Collaboration, Kisumu, Kenya, 4 KEMRI, Centre for Global Health Research, Kisumu, Kenya, 5 Malaria Control Unit, Ministry of Health, Nairobi, Kenya, 6 Malaria Branch, Division of Parasitic Diseases and Malaria, Center for Global Health, US Centers for Disease Control and Prevention, Atlanta, USA and Kenya, 7 US President’s Malaria Initiative, Nairobi, Kenya
To be submitted
Chapter 7
114
ABSTRACTAlthough prompt and effective treatment is a cornerstone of malaria control, information on
healthcare provider adherence to malaria treatment guidelines in pregnancy is lacking. Incorrect or
sub‐optimal treatment can cause adverse consequences to the mother and fetus.
We conducted a cross‐sectional study from September to November 2013, in all health facilities and
randomly selected drug outlets in the HDSS catchment area in Siaya County western Kenya, to assess
healthcare provider and drug dispenser adherence to and knowledge of national guidelines for
treatment of uncomplicated malaria in pregnancy. In health facilities, exit interviews of women of
childbearing age, inclusive of pregnant women, who had been assessed for fever were used to
assess adherence to malaria treatment guidelines. Simulated clients posing as 1st trimester pregnant
women or as relatives of women in 3rd trimester collected information from drug outlets.
Information on treatment was recorded from prescriptions or after reviewing medications in
patient’s possession. Standardized questionnaires were used to assess healthcare providers’ and
drug outlet dispensers’ knowledge of treatment guidelines.
Fifty‐six percent of health facility providers, versus none of the drug outlet dispensers, knew the
appropriate treatment for 1st trimester patients, while 39% and 87%, respectively, knew the
appropriate treatment for 2nd/3rd trimester. Prescription of the correct drug for pregnancy trimester
at the correct dosage was observed in 66% of cases in health facilities and 40% in drug outlets.
Prescribing was more often correct in 2nd/3rd trimester than in 1st (65% vs. 32%, p=0.004, and 38% vs.
0%, p<0.001, at health facilities and drug outlets, respectively). Sulfadoxine‐pyrimethamine, which is
no longer recommended for treatment of acute malaria, was prescribed in 3% of cases in health
facilities and 18% of simulations in drug outlets (p<0.001). Exposure to artemether‐lumefantrine in
1st trimester, which is contraindicated due to its unknown safety, occurred in 16% and 51% of cases
in health facilities and drug outlets, respectively (p=0.04); none were a result of quinine stock‐out.
This study highlights knowledge inadequacies and incorrect prescribing practices in the treatment of
malaria in pregnancy. These should be addressed through comprehensive trainings and adequate
supervision by the Kenya Ministry of Health to improve the quality of patient care and maximize
therapeutic outcomes.
Knowledge and adherence to malaria in pregnancy treatment guidelines
115
INTRODUCTIONIt is estimated that around 125 million pregnancies occur in areas at risk of P. falciparum and/or P.
vivax infections every year; an estimated 1.3 million of these are in Kenya [1]. Malaria in pregnancy
(MiP) can have devastating consequences for the woman and her unborn baby. Adverse effects of
MiP include maternal anemia, fetal loss, intrauterine growth retardation, premature delivery and
low birth weight (LBW); LBW associated with MiP results in an estimated 100,000 deaths each year
in Africa alone [2]. In order to prevent the adverse consequences associated with MiP, the World
Health Organization (WHO) and the Kenyan Ministry of Health recommend that pregnant women
living in malaria endemic area use long‐lasting insecticidal nets (LLINs), intermittent preventive
treatment in pregnancy (IPTp) with sulfadoxine‐pyrimethamine (SP), and receive prompt and
effective diagnosis and treatment of malaria infections with a safe drug.
In Kenya, following WHO recommendations, artemether‐lumefantrine (AL) is the 1st‐line treatment
for uncomplicated P. falciparum malaria in the general population and women in 2nd/3rd trimester of
pregnancy; due to insufficient safety data [3,4,5], this is not recommended in 1st trimester, and oral
quinine should be used instead [6,7]. In practice, this means that all women of childbearing age
(WOCBA) must be assessed for pregnancy inclusive of the trimester of pregnancy. In addition,
Kenyan guidelines, updated in 2010, stipulate that ACTs should only be provided for malaria cases
confirmed by diagnostic test. Antimalarial treatment on the basis of clinical suspicion of malaria
should only be considered in situations where a parasitological diagnosis is not accessible [6].
There is limited data on healthcare provider adherence to diagnostic and treatment guidelines for
MiP. A review of studies of antimalarial use in the general population found that only 51% of cases
were treated with recommended antimalarials during 2004‐2006 [8]. A 2008 study in Kenyan health
facilities reported limited health worker compliance with the recommended treatment guidelines in
patients over 5 years of age [9]. Although 99% of patients with a positive test received an
antimalarial, only 80% received AL, the recommended first line therapy, despite the fact that the
study was restricted to health facilities, which had both malaria diagnostics and AL available on the
survey day. A more recent 2010 study in Kenya found improved adherence, with 90% of test positive
patients receiving the recommended first‐line therapy [10]. Very few studies have looked at
adherence to treatment guidelines in pregnancy. In Uganda, 70% of women in the 1st trimester of
pregnancy received a drug that is contraindicated [11]; in Tanzania, 43% of drug dispensers in
registered pharmacies offered AL regardless of the pregnant client’s gestation [12]. In addition, the
few studies on provider knowledge and prescribing and dispensing practice of antimalarials for
WOCBA show limited evidence of knowledge regarding the potential for teratogenicity with ACTs
[12,13,14,15]; in one study only 20% of drug dispensers in private pharmacies knew that AL was
contraindicated in 1st trimester. Given that WOCBA represent about 25% of the total population, up
to 14% of whom could be pregnant at any time and one‐third of them in the 1st trimester, it is crucial
that providers recognize regimens that have the potential for teratogenicity and assess WOCBA for
pregnancy status and gestational age.
Understanding provider prescribing behaviour in pregnant patients can play a key role in improving
the prescribing, administration, and use of antimalarials while minimizing potential harmful
exposures. This cross‐sectional study assessed healthcare provider and drug dispenser prescribing
behaviors and knowledge of malaria treatment guidelines for pregnant clients in a malaria endemic
region of western Kenya.
Chapter 7
116
METHODSPrescribing practice was observed by a) use of simulated client approach within randomly sampled
drug outlets, b) exit interviews with WOCBA (18‐49 years) and pregnant clients being treated for
febrile illness at all health facilities (HF) within the study area, and c) provider surveys using
structured questionnaires conducted for healthcare providers and drug dispensers to assess
knowledge of malaria treatment guidelines and self‐reported prescribing behavior for case
management of MiP. The latter surveys were administered following completion of the provider
practice component so as to avoid any influence in provider behavior.
StudySite&Sampling
This study was carried out from September to November 2013, in and around the KEMRI and CDC
Health and Demographic Surveillance System (HDSS) catchment area in Siaya County, Nyanza
Province in western Kenya. The HDSS collects birth, death, and migration information quarterly from
a large, rural area of approximately 700 km2 with 220,000 inhabitants, 95% of whom are ethnically
Luo [16]. Malaria transmission in western Kenya is perennial and holo‐endemic with peaks following
the two rainy seasons, from March through May and October through December). In the study area,
approximately 20% of pregnant women coming for the first antenatal clinic visit are parasitemic,
70% are anemic [17], 26% are HIV positive, and 18% of women delivering in Siaya District Hospital
had placental malaria [18].
HealthFacilitySelection
All facilities in the HDSS study area and within a 5 km buffer zone were deemed eligible if they were
operational, stocked antimalarials, and were visited by WOCBA for treatment of potential febrile
illness. Nine facilities were excluded due to ongoing studies that could have influenced study results.
Fifty‐two health facilities, including hospitals, health centers, and dispensaries, were eligible for the
study; 50 consented to participate.
DrugOutletsSelection
Prior to the start of data collection, a census was conducted of all registered and unregistered
entities selling antimalarial drugs within the HDSS border (Kioko et al., unpublished). Forty‐one drug
outlets were randomly selected from the 179 (99%) outlets that consented to participate at the time
of census; an additional 18 were randomly selected for a reserve list in case of closure or stock‐out
at the time of the simulated client component. Assuming that approximately 45% of providers have
adequate knowledge and prescribing practice [12], sampling 40 drug outlets out of 200 allowed an
estimation of the proportion of providers with adequate knowledge with 14% precision at 80%
power.
DataCollection
Training
Fieldworkers underwent two weeks of training, including interviewing techniques, data recording,
and piloting of survey tools. A subset of four fieldworkers (two women and two men) was trained on
the methodology behind the simulated client approach; and the standardized, pre‐determined
scenarios were piloted for an additional week in outlets outside the study area prior to
implementation.
Knowledge and adherence to malaria in pregnancy treatment guidelines
117
ExitInterviewsinHealthFacilities
Patients were approached for eligibility assessment after completion of provider consultation, from
either outpatient department (OPD) or antenatal care clinic (ANC), and receipt of all prescribed
medications. Eligible patients included any WOCBA, either pregnant or non‐pregnant, that presented
with febrile illness and consented to participate in the study. Fieldworkers tried to interview at least
one of each of the following categories of patients per facility: 1) WOCBA who could potentially be
pregnant, 2) women in early pregnancy (1st trimester, defined as up to 14 weeks inclusive), and 3)
women in late pregnancy (2nd/3rd trimester, defined as 15 weeks gestation or greater). Pregnancy
status was based on patient report; gestational age and trimester were later confirmed by
calculation from patient reported date of last menstrual period (LMP).
After obtaining informed consent, exit interviews were conducted following a standard format. In
cases where an antimalarial contraindicated for pregnancy had been prescribed, the patient was
informed of the national treatment guidelines for MiP. The field supervisor (a Kenyan clinician) and
study coordinator were immediately informed and the recommended treatment was given to the
patient with appropriate dosage instructions and information.
SimulatedClientsinDrugOutlets
The simulated client (also known as mystery clients or shoppers [19,20]) approach was used to
assess prescribing practice within drug outlets. Female fieldworkers presented themselves as either
WOCBA or in early pregnancy, and male fieldworkers presented as the husband of a WOCBA or
woman in 3rd trimester of pregnancy. All simulated clients exhibited care‐seeking behavior for
uncomplicated malaria by presenting with general malaise, and, if prompted, complained of fever,
headache, chills, joint or muscle pain and nausea. The simulated clients were trained not to disclose
pregnancy status unless it was asked about by the dispenser. If dispensers failed to assess pregnancy
status, following receipt of a prescription the simulated clients would then disclose pregnancy status
(using local language to convey either early or late pregnancy depending on the pre‐set scenario)
and note any changes in the prescribed treatment or advice given. Simulated clients were able to
purchase medications up to an allotted 250 KSh (3.00 USD), but were instructed not to take
pregnancy or malaria diagnostic tests or treatment, if offered. The study coordinator and/or field
supervisor were in the vicinity at the time of simulation in case the simulated client was uncovered.
Immediately following completion of the scenario, the checklist for simulated client interaction was
completed under guidance of the study coordinator and/or field supervisor.
ProviderSurveysinHealthFacilities&DrugOutlets
Following the completion of exit interviews or client simulations at each facility or outlet, a separate
fieldworker administered a structured questionnaire to the provider to assess knowledge and self‐
reported prescribing practice, including: training, knowledge of symptoms, diagnosis, availability of
the most recent treatment guidelines, and treatment and preventive regimens for different
scenarios including pregnant women and the general population.
DataManagement&Analysis
Information for the drug outlet census and mapping and the provider survey components was
collected via personal digital assistant (PDA), and data for the simulated client and exit interview
components were collected via scannable form.
Chapter 7
118
All datasets were cleaned and analyzed using SAS 9.3 (SAS Institute, Cary, NC, USA). Descriptive
statistics were performed on all data to identify the extent of adherence to and knowledge of
Kenyan National Treatment Guidelines as they pertain to MiP across the provider study population.
Exit interview and simulated client data were analyzed to describe provider prescribing and
dispensing behavior in reference to pregnancy status, malaria diagnosis prior to treatment, correct
treatment and dosage, and provision of appropriate information as pertaining to treatment advice.
Provider survey data were also analyzed across these categories pertaining to the malaria treatment
guidelines. Variables within these categories were coded to give a threshold for dichotomous correct
vs. incorrect practice or knowledge for each category.
Correct practice and adequate knowledge definitions (Table 1) were based on the 2010 Kenyan
National Malaria Treatment Guidelines (MTGs) [7]; where these were insufficient, the 2010 WHO
Malaria Treatment Guidelines [6] were used. For exit interviews and client simulations, treatment
was considered correct if either first‐ or second‐line treatment was prescribed.
Table 1. Definitions of Correct Practice & Adequate Knowledge
Correct Malaria Diagnosis
Utilization of microscopy or RDT
Clinical diagnosis when diagnostic test unavailable
Correct Pregnancy Assessment
Inquired about pregnancy and/ or offered pregnancy test
Inquiry on LMP or gestational age
Correct Treatment & Dosage
Acceptable Knowledge Answers Acceptable Prescriptions in Practice
Non‐pregnant Non‐pregnant
1st‐line: Artemether‐lumefantrine (4x2x3) Artemether‐lumefantrine (4x2x3)
2nd‐line: DHA‐piperaquine (3x1x3 or 4x1x3) DHA‐piperaquine (3x1x3 or 4x1x3)
Quinine (2x3x7)
1st Trimester 1st Trimester
Quinine (2x3x7) Quinine (2x3x7)
2nd/3rd Trimester 2nd/3rd Trimester
Quinine (2x3x7) Quinine (2x3x7)
Artemether‐lumefantrine (4x2x3) Artemether‐lumefantrine (4x2x3)
DHA‐piperaquine (3x1x3 or 4x1x3)
Treatment regimens:
Artemether‐lumefantrine tablets (20/120 mg): 4 tablets, 2 times daily for 3 days (4x2x3)
DHA‐piperaquine tablets (40/320 mg): 3 or 4 tablets, once daily for 3 days (3x1x3) (4x1x3)
Quinine: 2 tablets of 300 mg, 3 times daily for 7 days (2x3x7)
Acronyms: RDT, rapid diagnostic test; LMP, date of last menstruation period; DHA, dihydroartemisnin
Chi square test or Fisher exact test were used to assess statistical significance of comparisons
between categorical variables; p≤0.05 indicates statistical significance. The total proportion of
providers (clustering on facility) who met adequate pregnancy assessment standards, adequate
malaria diagnostic standards, and correctly prescribed drug and dosage was calculated; these
measures were used to define overall correct prescribing practice. Logistic regressions were
performed at the individual provider level (clustering on provider) to identify potential predictors of
knowledge and practice level. Bivariate analyses of all provider characteristics were regressed upon
Knowledge and adherence to malaria in pregnancy treatment guidelines
119
binary ‘Adequate Knowledge Score;’ variables identified as significant predictors of adequate
provider knowledge (p<0.2) were included in the multivariate model and further analyzed for
interaction to determine which provider characteristics were the best predictors of provider
knowledge. Bivariate and multivariate analyses of provider characteristics and correct provider
practice were done similarly.
Ethics
This study was approved by the ethical and institutional review boards of the U.S. Centers for
Disease Control and Prevention (CDC), the Kenya Medical Research Institute (KEMRI), Liverpool
School of Tropical Medicine (LSTM), and Emory University prior to the start of study. In addition,
permission to operate within the study area was obtained via meetings with the Siaya County
Director of Health and the District Health Management Teams for Bondo, Gem, Rarieda and Siaya
districts. The community advisory boards for each district were notified of the study presence within
the area. Signed letters of approval were obtained from the Siaya County Director of Health and
each District Health Management Team. Verbal consent was obtained from health facility in‐charges
prior to any interview activities at the respective facility; informed consent regarding future
potential participation in a study for MiP treatment assessment was obtained from the dispenser
during the drug‐outlet mapping and census. Written, informed consent was obtained from all
providers and patients interviewed during the study period in the participants’ preferred language of
Dholuo, Kiswahili, or English. Study participants were free to withdraw from the study at any time.
RESULTS
PrescribingPracticeandDispensingBehaviours:ExitInterviewsinHealthFacilities
A total of 210 patients were interviewed across 51 health facilities: 108 non‐pregnant women, 19
women in the 1st trimester of pregnancy, 77 women in 2nd or 3rd trimester of pregnancy, and 6
women who were unsure of their pregnancy status; one interview with a woman in her 2nd trimester
of pregnancy with severe malaria was excluded from the analysis. The average age of the patients
was 26 years; 26% of respondents had completed secondary school (Tables 2a & 2b).
Table 2a. Exit interview: health facility characteristics
Total Facilities Total Interviews
N % N %
District 51 209
Bondo 6 11.8 18 8.6
Gem 20 39.2 89 42.6
Rarieda 9 17.6 28 13.4
Siaya 16 31.4 74 35.4
Facility Type
Hospital 4 7.8 18 8.6
Health Center 19 37.3 83 39.7
Dispensary 28 54.9 108 51.7
Facility Managing Authority
Government 44 86.3 188 90.0
Mission 2 3.9 4 1.9
Private 5 9.8 17 8.1
Chapter 7
120
Table 2b. Health facility exit interview: respondent characteristics
Overall Unsure Non‐Pregnant 1st trimester** 2nd/3rd trimester
N % N % N % N % N %
Education Level 209 6 108 19 76
No Education 16 7.7 1 16.7 13 12.0 1 5.3 1 1.3
Primary 138 66.0 5 83.3 68 63.0 13 68.4 52 68.4
Secondary 38 18.2 0 0 20 18.5 2 10.5 16 20.8
Higher than secondary
17 8.1 0 0 7 6.5 3 15.8 7 9.2
Age mean (range) & Standard Deviation
26.3 (17‐48)
7.2 26.8
(18‐45) 9.7
28.4 (18‐48)
7.8 24.5
(18‐35) 5.0
23.9 (17‐40)
5.6
Symptoms Reported to provider*
Fever 138 66.0 4 66.7 76 70.4 13 68.4 45 59.2
Headache 183 87.6 6 100.0 98 90.7 16 84.2 63 82.9
Pain 104 49.8 4 66.7 53 49.1 7 36.8 24 31.6
Nausea 72 34.4 2 33.3 25 23.1 7 36.8 31 40.8
Malaise 80 38.3 2 33.3 39 36.1 5 26.3 34 44.7
Chills 17 8.1 0 0.0 9 8.3 2 10.5 6 7.9
Stomach Pain 23 11.0 1 16.7 11 10.2 3 15.8 8 10.5
Cough 18 8.6 1 16.7 6 5.6 3 15.8 8 10.5
Dizziness 3 1.4 0 0.0 0 0.0 0 0.0 3 3.9
Diarrhea 2 1.0 0 0.0 0 0.0 1 5.3 1 1.3
Gravidity
0 33 21.2 1 33.3 11 17.5 3 15.8 18 25.4
1 31 19.9 0.0 11 17.5 7 36.8 14 18.3
2 33 21.2 0.0 12 19.0 3 15.8 18 25.4
3‐4 32 20.5 1 33.3 12 19.0 5 26.3 13 19.7
5‐6 23 14.7 1 33.3 13 20.6 1 5.3 8 11.3
7+ 4 2.6 0 0.0 4 6.3 0 0.0 0 0.0
Missing 53 3 45 0 5
*2 reported no symptoms to provider **Patients with gestational age of up to 14 weeks, 6 days were included in 1st trimester given that treatment guidelines use 'quickening' as a treatment indicator
MalariaDiagnosisinHealthFacilities
Of the 209 women interviewed, 160 (77%) women were tested for malaria (RDT or microscopy). Of
those women who were not tested, 28 women were appropriately clinically diagnosed in facilities
that did not have the capacity to perform malaria diagnostics. Taking this into account, 90% were
properly assessed for malaria according to the facility diagnostic capability (Table 3).
Knowledge and adherence to malaria in pregnancy treatment guidelines
121
Table 3 Malaria diagnostics practice in health facilities as observed through exit interviews across facility type
Overall Hospital Health Center Dispensary
N % N % N % N %
Diagnostic Test Performed* 209 18 83 108
Yes 160 77.0 16 88.9 75 90.4 70 63.9
No 48 23.0 2 11.1 8 9.6 38 35.2
Don't Know 1 0.5 0 0.0 0 0.0 1 0.9
Malaria Test Results 160 16 75 69
Positive 151 94.4 14 87.5 70 93.3 68 97.1
Negative 3 1.9 2 12.5 1 1.3 0 0.0
Don't Know 6 3.8 0 0.0 4 5.3 2 2.9
Test Location 160 16 75 69
OPD 28 17.5 0 0.0 3 4.0 25 36.2
Laboratory 132 81.9 16 100.0 72 96.0 44 62.3
Pharmacy 1 0.6 0 0.0 0 0.0 1 1.4
No/Unknown Diagnostic Test 49 23.3 2 11.1 18 21.7 57 52.3
Correct Clinical Diagnosis** 28 57.1 NA NA 4 50.0 24 61.5
Incorrect Clinical Diagnosis*** 21 42.9 2 100.0 4 50.0 15 38.5
Correct Malaria Diagnosis 188 90.0 16 88.9 79 95.2 93 81.6
*Tested via Rapid Diagnostic Test (RDT) or microscopy **Correct Clinical Diagnosis indicates women presenting with fever, multiple symptoms, and/or were pregnant with symptom(s) at facilities without laboratory or RDT diagnostic capacity ***Incorrect Diagnosis indicates patients treated for malaria without diagnostic testing at facilities where it was available)
or without clinical presentation if at a facility with no diagnostic capacity
PregnancyAssessmentinHealthFacilities
Overall, 92 (44%) of 209 patients were asked about their potential pregnancy status; inquiry was
more common among pregnant patients (90% and 61% for 1st and 2nd/3rd trimesters, respectively)
than among non‐pregnant patients (26%, Table 4). Only 43% of women were asked about their LMP;
this occurred with much greater frequency in pregnant (70%) versus non‐pregnant patients (24%).
Only 20 (10%) women were offered a pregnancy test; 80% of these women were pregnant (9 in the
1st, 5 in the 2nd and 2 in the 3rd trimester). Three of the women who were unsure of their pregnancy
status and 23 women who reported to us that they were not pregnant had LMPs of greater than 6
weeks, thus should have been tested for pregnancy. Overall, 52% of patients were correctly assessed
for pregnancy status (Table 4); however this was significantly higher in pregnant women versus self‐
reported non‐pregnant women (84% vs. 24%, p<0.0001). There was also a statistically significant
difference across facility types (hospital 78%, health center 53% and dispensaries 45%; p=0.03).
Chapter 7
122
Table 4. Pregnancy assessment practice in Health Facilities as observed through exit interviews stratified across pregnancy status.
Overall Unknown Non‐Pregnant 1st Trimester
2nd/3rd Trimester
N % N % N % N % N %
All Patients 209 6 108 19 76
Pregnancy Status Inquiry 92 44.0 1 16.7 28 25.9 17 89.5 46 60.5
Pregnancy Test Offered 20 9.6 0 0 4 3.7 9 47.4 7 9.2
LMP Inquiry 89 42.6 1 16.7 26 24.1 16 84.2 46 60.5
Pregnancy Duration/Timing 67 70.5 NA NA 14 73.7 53 69.7
Additional Confirmation* 41 43.2 NA NA 5 26.3 36 47.4
Correct pregnancy assessment 108 51.7 1 16.7 26 24.1 18 94.7 63 82.9
*Additional confirmation included palpation in 1st trimester, and palpation or observation in 2nd/3rd trimester cases.
TreatmentPrescribedinHealthFacilities
An antimalarial medication was prescribed to 205 (98%) of the 209 women; the most frequently
prescribed was AL (84%), followed by quinine (14%), and sulfadoxine‐pyrimethamine (SP) (3%).
Overall, 66% of providers prescribed the correct treatment and dosage to the patient across all
pregnancy scenarios (Table 5). AL was incorrectly prescribed in 1st trimester to 3/19 (16%) women.
The majority of prescriptions for AL and SP were for the correct dosage (73% and 71%, respectively);
in contrast, the correct dose of quinine was prescribed only 31% of the time. The correct drug and
dosage was prescribed more frequently to non‐pregnant patients (70%) and those in the 2nd/3rd
trimester of pregnancy (66%) than to those in 1st trimester (32%, p=0.001). Hospital‐based providers
(78%) were the most likely to provide correct treatment to non‐pregnant patients (versus Health
Centre‐72%, Dispensary‐ 68%). However, they were least likely to provide correct treatment to
pregnant patients regardless of trimester (hospitals 33%, health centres 62% and dispensaries 58%).
The first dose of antimalarial was directly observed in 25% of cases. Very few patients (7%) were
informed of potential side effects.
Knowledge and adherence to malaria in pregnancy treatment guidelines
123
Table 5. Malaria treatment practice in health facilities as observed through exit interviews stratified across pregnancy status.
Overall Unsure* Non‐Pregnant 1st Trimester 2nd/3rd Trimester
N % N % N % N % N %
Prescribed Antimalarials 205 6 107 18 74
Antimalarial prescribed Correct Dosage (tabs x doses x days)**
Artemether‐lumefantrine 173 83.9 6 102 94.4 3 15.8 61 82.4
(4x2x3) 125 72.7 2 33.3 74 72.5 3 100.0 46 75.4
DHA‐piperaquine 2 1.0 0 0.0 2 1.9 0 0.0 0 0.0
(3x1x3) 1 50.0 1 50.0
Quinine 29 14.1 0 0.0 4 3.7 13 68.4 12 16.2
(2x3x7) 8 27.6 0 0.0 6 46.2 2 16.7
(150mgxkg) 1 3.4 1 25.0 0 0.0 0 0.0
Sulfadoxine‐pyrimethamine 7 3.4 0 0.0 1 0.9 2 10.5 4 5.4
(3x1x1) 5 71.4 0 0.0 2 100.0 3 75.0
Artemether Injection 1 0.5 0 0.0 1 0.9 0 0.0 0 0.0
(60mg) 1 100.0 1 100.0
Correct Drug 195 98.0 NA 108 100.0 13 68.4 75 97.4
Correct Drug & Dosage 131 65.8 NA 76 70.4 6 31.6 50 66.2
Concomitant Medications
Analgesic 148 72.2 4 66.7 77 72.0 15 83.3 52 70.3
Antibiotic 75 36.6 1 16.7 37 34.6 6 33.3 31 41.9
Treatment Advice
Reason for Prescription 58 28.3 1 16.7 32 29.9 4 22.2 21 28.4
Side Effects 14 6.8 0 0.0 5 4.7 4 22.2 5 6.8
Any other advice 46 22.3 2 33.3 20 18.7 5 27.8 19 25.7
Any treatment Advice 83 40.5 2 33.3 41 38.3 10 55.6 30 40.5
*Unsure pregnancy status refers to women who self‐reported as not knowing if they were pregnant or not, correct prescribing practice was not assessable for those; these women are not included in the ‘Correct Drug’ or ‘Drug & Dosage’ denominator . **Percentage for correct dosage is based on the numbers receiving the specific antimalarial. ***Any other advice includes patient‐reported advice by the provider including emphasis of complete medication regimen, eating prior to taking medication, sleeping under ITNs, etc.
PredictorsofCorrectPracticeinHealthFacilities
Significant predictors of correct prescribing and diagnostic practice in health facilities were
respondent cadre, dispensing medication and working in the OPD (Table 6). Health facility‐based
pharmacists were more likely to correctly prescribe and diagnose patients then their clinical
counterparts [RNs, COs, MDs] (OR=1.8 [95% CI 1.4‐2.2]). Likewise, providers that reported
dispensing medicines were more adherent to correct practice than their counterparts who did not
dispense medicines (OR=1.2 [95% CI 1.1‐1.4]). Providers that reported sometimes working in the
OPD were less likely to correctly diagnose and treat patients than those who reported always
working in the OPD (OR=0.8 [95% CI 0.7‐0.9]). Age and gender were controlled for in the model;
however they had no significant effect on correct prescribing practice. Neither malaria diagnostic
training nor MiP training within the last five years were statistically significant predictors of correct
provider practice in health facilities.
Chapter 7
124
Table 6 Predictors of correct prescribing and diagnostic practice in health facilities
Provider Characteristic N % Crude OR
95% CI P
value Adjusted*
OR 95% CI P value
Respondent Cadre 184
Nurse (ref) 109 59.2 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐
Clinical Officer/M.D. 52 28.3 1.1 (0.9, 1.3) 0.44 1.1 (0.9, 1.3) 0.43
Pharmacist 5 2.7 1.6 (1.1, 2.3) 0.01 1.8 (1.4, 2.2) <0.001
Other 18 9.8 0.7 (0.7, 0.8) <0.001 0.8 (0.7, 0.9) <0.001
Dispenses Medicine
No (ref) 19 10.3 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐
Yes 165 89.7 1.2 (1.0, 1.3) 0.04 1.2 (1.1, 1.4) <0.001
Works in Outpatient Department
Always (ref) 95 51.6 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐
Sometimes 68 37.0 0.9 (0.7, 1.0) 0.09 0.8 (0.7, 0.9) <0.001
Never 21 11.4 0.7 (0.6, 0.8) <0.001 0.9 (0.7, 1.1) 0.18
Gender
Male (ref) 100 54.3 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐
Female 84 45.7 0.9 (0.8, 1.1) 0.23 1.1 (0.9, 1.3) 0.57
Age 1.0 (1.0, 1.0) 0.31 1.0 (1.0, 1.0) 0.42
Malaria Diagnostic training within 5 years
No (ref) 51 27.7 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐
Yes 133 72.3 0.9 (0.8, 1.1) 0.46 ‐‐ ‐‐ ‐‐
MiP training within 5 years
No (ref) 118 64.1 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐
Yes 66 35.9 0.9 (0.8, 1.1) 0.50 ‐‐ ‐‐ ‐‐
* Adjusted for respondent cadre, dispensing medicine, working in OPD, gender and age.
PrescribingPracticeandDispensingBehaviours:SimulatedClientsinDrugOutlets
Simulations were completed at 41 drug outlets; two facilities were homesteads (individuals selling
antimalarials from their residence) and were included as informal drug outlets. There were 81
simulated client‐drug dispenser interactions with 154 total scenarios simulated Dispensers were a
mean of 33 years old, and 59% were female. The majority had completed only primary (18%) or
secondary school (41%). Between 36 and 40 simulations per scenario were completed (Table 7).
Knowledge and adherence to malaria in pregnancy treatment guidelines
125
Table 7 Drug outlet & simulation characteristics
Drug Outlets Overall Registered Pharmacy
Informal Drug Outlet*
General Shop
N %
[SD] N
%
N % N %
District/Sub‐county 41 9 15 17
Gem 13 31.7 3 33.3 6 40.0 4 23.5
Rarieda 10 24.4 2 22.2 5 33.3 3 17.6
Siaya 18 43.9 4 44.4 4 26.7 10 58.8
Provider Gender
Male 17 41.5 5 55.6 5 33.3 7 41.2
Female 24 58.5 4 44.4 10 66.7 10 58.8
Education Level**
Primary School 7 17.9 0 0.0 0 0.0 7 41.2
Secondary School 16 41.0 5 55.6 6 40.0 5 29.4
Higher Education 6 15.4 2 22.2 2 13.3 2 11.8
Clinical Officer/MD 1 2.6 0 0.0 1 6.7 0 0.0
Registered Midwife/Nurse 2 5.1 1 11.1 1 6.7 0 0.0
Enrolled Midwife/Nurse 1 2.6 0 0.0 1 6.7 0 0.0
Pharmacist 3 7.7 1 11.1 1 6.7 1 5.9
Other technical 3 7.7 0 0.0 3 20.0 0 0.0
Age mean (range) 32.5
(19‐60) [9.4]
28.0 (19‐46)
[8.5] 32.3
(20‐50) [8.2]
35.5 (21‐60)
[10.5]
Simulation Characteristics 154 35 56 63
WOCBA 39 25.3 9 25.7 15 26.8 15 23.8
1st Trimester Pregnancy 39 25.3 9 25.7 14 25.0 16 25.4
Relative of WOCBA 36 23.4 8 22.9 12 21.4 16 25.4
Relative of 3rd Trimester 40 26.0 9 25.7 15 26.8 16 25.4
*There were 2 homesteads visited; these were included as informal drug outlets. **Education level was obtained from matched provider surveys (39 of 41 dispensers were interviewed) and this is missing for 2 providers in the general shop
MalariaDiagnosisinDrugOutlets
Dispensers assessed for malaria in 37% of all interactions (Table 8). Assessment for malaria was
more common when the simulator posed as the patient as opposed to the patient’s husband. RDTs
were offered in 10% of interactions where the client was present; RDTs were offered in both of the
homestead client‐present scenarios. Simulators posing as husbands were asked if their febrile wife
had been previously given a diagnostic test in 7% of interactions. 36% of dispensers asked about
symptoms, with just less than half of these inquiring about specific symptoms. No drug outlet
dispenser offered to take the simulated client’s temperature. A prescription was requested by the
dispenser in only 5% of interactions.
Chapter 7
126
Table 8 Malaria diagnostics practice in drug outlets as observed through simulated clients across pregnancy status
Overall WOCBA / 1st Trimester Relative of WOCBA/
3rd Trimester N % N % N %
Symptoms 81 40 41
Any Inquiry 29 35.8 17 42.5 12 29.3
Specific 14 17.3 12 30.0 2 4.9
Fever 8 9.9 7 17.5 1 2.4
Chills 3 3.7 3 7.5 0 0.0
Headache 11 13.6 9 22.5 2 4.9
Nausea 6 7.4 4 10.0 2 4.9
Pain 4 4.9 3 7.5 1 2.4
Prescription 4 4.9 1 2.5 3 7.3
Temperature 0 0.0 0 0.0 NA
Diagnostic Test or Test Inquiry 7 8.6 4 10.0 3 7.3
Any Malaria Diagnostic 30 37.0 18 45.0 12 29.3
PregnancyAssessmentinDrugOutlets
There were only six unprompted pregnancy inquiries across 81 total interactions (7%) and a
pregnancy test was offered in two of these cases. Gestation was inquired in four of six interactions.
Dispensers were informed of positive pregnancy status in 73 interactions where there was no initial
inquiry on the part of the dispenser; in 58% of these interactions the dispenser followed up with
gestational age or LMP inquiry. Inquiry about gestational age was highest in registered pharmacies
(77%) and informal drug shops (74%), with general shops significantly lower (33%, p=0.003); this did
not differ between interactions where the client was the patient versus the patient relative (table 9).
Knowledge and adherence to malaria in pregnancy treatment guidelines
127
Table 9 Pregnancy assessment practice in drug outlets as observed through simulated clients across pregnancy status
Overall 1st Trimester 3rd Trimester
N % N % N %
Pregnancy Inquiry 81 40 41
Unprompted Pregnancy Inquiry 6 7.4 2 5 4 9.8
Confirmation
Timing 4 66.7 2 100.0 2 50.0
Last menstruation period 1 25.0 1 50.0 0 0.0
Gestation 3 75.0 1 50.0 2 100.0
Pregnancy Test Offered 2 33.3 2 100.0 NA
None 2 33.3 0 0.0 2 50.0
Informed Provider of Pregnancy Status 73 90.1 37 92.5 36 87.8
Confirmation
Timing 42 57.5 22 59.5 20 55.6
Last menstruation period 2 4.8 2 9.1 0 0.0
Gestation 40 95.2 20 90.9 20 100
Pregnancy Test Offered 0 0.0 0 0.0 0 0.0
None 31 42.5 15 40.5 16 44.4
Correct Pregnancy Assessment* 48 59.3 23 57.5 25 61
*Correct Pregnancy Assessment indicates that the provider confirmed pregnancy status via LMP, gestational inquiry, or pregnancy test
TreatmentDispensedinDrugOutlets
Correct prescribing and diagnostic practice was only observed in 3% of the 154 interactions in drug
outlets, with no significant difference across outlet types (Table 11). There were highly significant
differences in correct treatment and dosage across pregnancy status, with 63% of non‐pregnant,
38% of 3rd trimester, and 0% of 1st trimester client simulations receiving the appropriate treatment
(p<0.0001). Antimalarials were dispensed in 83% of all interactions; AL was most commonly
dispensed (76%), followed by SP (22%). Quinine was not dispensed in the drug outlets. Dispensers
were 7.6 times more likely to prescribe SP for treatment of acute malaria to pregnant versus non‐
pregnant women (p<0.0001). About half (51%) of the 39 1st trimester clients were prescribed AL; all
but 1 of the remaining clients were prescribed SP. AL was initially prescribed to over 90% of
simulated client patients; in 27% of cases treatment was changed from AL to SP after finding out the
patient was pregnant, regardless of trimester, and in another 17% AL was withdrawn and the patient
was referred to a health facility.
Chapter 7
128
Table 10. Correct treatment and dosage characteristics by pregnancy status in drug outlets
Overall WOCBA 1st Trimester 3rd Trimester
N % N % N % N %
154 75 39 40
Dispensed Antimalarials 127 82.5 66 88.0 32 82.1 29 72.5
Antimalarial dispensed Correct Dosage (tabs x doses x days)*
Artemether lumefantrine 96 75.6 60 90.9 20 62.5 16 55.2
(4x2x3) 78 61.4 47 78.3 16 80.0 15 51.7
Artesunate amodiaquine 1 0.8 1 1.5 0 0.0 0 0.0
(4x1x3) 0 0.0 0 0.0
Amodiaquine 2 1.6 1 1.5 1 3.1 0 0.0
(3x1x1) 1 0.8 0 0.0 1 100.0
Sulfadoxine Pyrimethamine 28 22.0 4 6.1 11 34.4 13 44.8
(3x1x1) 22 17.3 4 100.0 7 63.6 11 84.6
Correct Drug 77 60.6 61 92.4 0 0.0 16 55.2
Correct Drug & Dosage 62 48.8 47 71.2 0 0.0 15 51.7
Concomitant Medications
Analgesic 102 80.3 53 80.3 28 87.5 21 72.4
Multivitamin 2 1.6 0 0.0 0 0.0 2 6.9
Treatment Advice
Dosage directions 110 86.6 56 84.8 26 81.3 28 96.6
Visual instructions 72 56.7 35 53.0 20 62.5 17 58.6
Emphasize need to finish dose 11 8.7 4 6.1 3 9.4 4 13.8
Side effects 2 1.6 1 1.5 0 0.0 1 3.4
Advice if symptoms persist 6 4.7 3 4.5 3 9.4 0 0.0
Any treatment advice 110 86.6 56 84.8 26 81.3 28 96.6
*Percentage for correct dosage is based on the numbers receiving the specific antimalarial
Of 27 clients not given an antimalarial, 17 (63%) were referred to the hospital, and 8 (30%) clients
did not receive a medication due to antimalarial stock out. Other reasons for not receiving treatment
included refusal to treat without a prescription, diagnostic test, or clinical evaluation.
Prior to dispensing, only 16% of dispensers questioned the simulated client if any previous treatment
had been given for the current illness and only 5% asked about potential allergies. Dosage directions
were given to 87% of simulated clients.
Table 11. Antimalarial dispensing practice in drug outlets as observed through simulated clients across pregnancy status.
Overall Registered Pharmacy Informal Drug Shop General Shop
N % N % N % N %
154 35 48 71
Malaria Diagnostics 55 35.7 13 37.1 21 43.8 21 29.6
Pregnancy Assessment 47 30.5 14 40.0 18 37.5 15 21.1
Treatment & Dosage 62 40.3 18 51.4 25 52.1 19 26.8
Non‐pregnant N=75 47 62.7 14 82.4 19 82.6 14 40.0
1st Trimester N=39 0 0.0 0 0.0 0 0.0 0 0.0
3rd Trimester N=40 15 37.5 4 44.4 6 46.2 5 27.8
Correct Practice 5 3.2 1 2.9 1 2.1 3 4.2
Knowledge and adherence to malaria in pregnancy treatment guidelines
129
PredictorsofCorrectDispensingPracticeinDrugOutlets
Prior MiP training was the only predictor of correct practice in drug outlets in the adjusted model
(Table 12). Dispensers who had been trained on MiP in the last 5 years were more likely to have
correct prescribing and diagnostic practice than those who had not been trained (p=0.05). Age,
gender, and prescribing status were controlled for in the model; however they had no significant
effect on correct prescribing practice.
Table 12. Predictors of correct dispensing practice in drug outlets
Provider Characteristic N % Crude OR
95% CI P
value Adjusted OR*
95% CI P
value
Respondent Cadre 113
Pharmacist (ref) 45 39.8 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐
Clinical Officer/M.D./Nurse 7 6.2 1.2 (0.9, 1.4) 0.23 1.0 (0.9, 1.2) 0.56
Shopkeeper/Other 61 54.0 1.0 (1.0, 1.1) 0.14 1.0 (1.0, 1.1) 0.28
Prescribes Medicine
No (ref) 51 45.1 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐
Yes 62 54.9 1.1 (1.0, 1.1) 0.06 1.0 (1.0, 1.1) 0.15
Malaria diagnostic training within 5 years
No (ref) 94 83.2 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐
Yes 19 16.8 1.1 (1.0, 1.2) 0.13 ‐‐ ‐‐ ‐‐
MiP Training within 5 years
No (ref) 99 87.6 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐
Yes 14 12.4 1.1 (1.0, 1.3) 0.07 1.1 (1.0, 1.3) 0.05
Sources of Information (ref='No')
CME as a source of info 29 25.7 1.1 (1.0, 1.2) 0.04 ‐‐ ‐‐ ‐‐
Gender
Male (ref) 42 37.2 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐
Female 71 62.8 1.0 (0.9, 1.0) 0.34 1.0 (0.9, 1.0) 0.17
Age 1.0 (1.0, 1.0) 0.31 1.0 (1.0, 1.0) 0.15
* Adjusted for respondent cadre, prescribing medicine, MiP training, gender and age. Malaria diagnostic training was not included in the adjusted model due to collinearity with MiP training. Continuing Medical Education (CME) was not included in adjusted model as it was not statistically significant.
Chapter 7
130
KnowledgeoftheNationalMalariaTreatmentGuidelinesamongProviders
CharacteristicsofRespondents
We surveyed 114 providers across 88 locations; 75 in HF and 39 in drug outlets. 44% of respondents
were nursing staff, 15% were COs/MDs, 18% pharmacists, and 13% were shopkeepers. 68% of
providers stated that they both prescribed and dispensed medication (Table 13a & 13b).
Table 13a. Facility characteristics from the provider survey on national malaria treatment guidelines
Total Facilities Total Providers Surveyed
N % N %
District 88 114
Bondo 6 6.8 9 7.9
Gem 32 36.4 42 36.8
Rarieda 17 19.3 19 16.7
Siaya 33 37.5 44 38.6
Facility Type
Hospital 4 4.5 6 5.3
Health Center 19 21.6 28 24.6
Dispensary 26 29.5 41 36.0
Total Health Facilities 49 55.7 75 65.8
Drug Outlet Type N % N %
Registered Pharmacy 9 10.2 9 7.9
Informal Drug Shop 13 14.8 13 11.4
General Shop 15 17.0 15 13.2
Homestead 2 2.3 2 1.8
Total Drug Outlets 39 44.3 39 34.2
Facility Managing Authority
Government 42 47.7 67 58.8
Mission 2 2.3 3 2.6
Private 44 50.0 44 38.6
Knowledge and adherence to malaria in pregnancy treatment guidelines
131
Table 13b. Provider characteristics from the provider survey on national malaria treatment guidelines
Overall Health facilities Drug Outlets
N % N % N %
114 75 39
Sex
Male 54 47.4 39 52.0 15 38.5
Female 60 52.6 36 48.0 24 61.5
Respondent Cadre 0.0
Registered Nurse 33 28.9 32 42.7 1 2.6
Enrolled Nurse 16 14.0 16 21.3 0 0.0
Clinical Officer/MD 18 15.8 17 22.7 1 2.6
Pharmacist 20 17.5 5 6.7 15 38.5
Shopkeeper 15 13.2 0 0.0 15 38.5
CHW/VR/other* 12 10.5 5 6.7 7 17.9
Professional Qualification 0.0
Primary School 9 7.9 2 2.7 7 17.9
Secondary School 23 20.2 7 9.3 16 41.0
Higher Education 19 16.7 13 17.3 6 15.4
Clinical Officer/MD 14 12.3 13 17.3 1 2.6
Registered Midwife/Nurse 22 19.3 20 26.7 2 5.1
Enrolled Midwife/Nurse 11 9.6 10 13.3 1 2.6
Pharmacist 4 3.5 1 1.3 3 7.7
Other technical 12 10.5 9 12.0 3 7.7
* Other included clerk, economist, statistical clerk, and support staff
NationalMalariaTreatmentGuidelinesAwareness
75% of all providers said they were aware of the National Malaria Treatment Guidelines (MTGs);
67% had read the MTGs, 55% were in possession of them, and 58% were aware of the government
initiative to disseminate them (Table 14). However, HF providers were much more likely to respond
in the affirmative to all MTG‐related questions compared to drug outlet providers. The majority of
all providers (54%) had attended a malaria management workshop (94% within the past 5 years),
however only 31% of all providers had attended a workshop specific to MiP. Over 86% of those that
reported attending any workshop within the past 5 years were HF providers.
Chapter 7
132
Table 14. Malaria treatment guideline awareness, comparing health facilities vs. drug outlets
Overall Health Facilities Drug Outlets P‐value*
N % N % N %
MTGs 114 75 39
Awareness of Government Initiative 66 57.9 62 82.7 4 10.3 p<0.001
Read the MTGs 76 66.7 67 89.3 9 23.1 p<0.001
In Possession 63 55.3 60 80.0 3 7.7 p<0.001
Additional Materials 73 64 71 94.7 2 5.1 p<0.001
Awareness of MTGs 85 74.6 74 98.7 11 28.2 p<0.001
Additional Sources of Information
Training/CME 64 56.1 52 69.3 12 30.8 p<0.001
DHMT/health facility memos 46 40.4 38 50.7 8 20.5 p<0.001
Colleagues 40 35.1 28 37.3 12 30.8 p=0.25
Media 57 50 33 44.0 24 61.5 p=0.04
Medical Journals 19 16.7 19 25.3 0 0.0 p<0.001
Medical Representatives 18 15.8 11 14.7 7 17.9 p=0.33
Other** 9 7.9 3 4.0 7 17.9 p=0.01
Training Workshops
Malaria Training 62 54.4 51 68.0 11 28.2 p<0.001
within past 5 years 58 93.5 50 98.0 8 72.7 p=0.01
training prior to 2008 4 6.5 1 2.0 3 27.3
MIP Training 35 30.7 30 40.0 5 12.8 p=0.001
within past 5 years 32 91.4 28 93.3 4 80.0 p=0.21
training prior to 2008 3 8.6 2 6.7 1 20.0
*P‐values from Chi‐square test **Other includes community meetings (Barazas), CDC staff, NGO staff, & Village Reporters Acronyms: MTGs, malaria treatment guidelines; CME, continuing medical education; DHMT, district health medical team; MiP, malaria in pregnancy
MalariainPregnancyConsequenceandDiagnosisKnowledge
98% of all providers surveyed knew that MiP can cause adverse effects; 90% were able to cite at
least one adverse effect. 90% of all providers suspected malaria in cases of fever; other clinical
symptoms cited included headache (84%), vomiting (82%), body ache (67%), and chills (65%).
Providers in health facilities had statistically significant greater knowledge of both MiP consequences
(maternal anemia, maternal death, LBW, and neonatal death) and clinical symptoms (fever,
headache, body ache, chills, vomiting) compared to drug outlet providers. 84% of providers in health
facilities reported utilizing laboratory diagnosis (81% RDT, 70% microscopy); however, of those that
did not use lab diagnostics, 25% reported always treating clinically (versus regularly, sometimes, and
never). In drug outlets, where diagnostics are not widely available, 33% reported utilizing RDTs or
microscopy while 33% reported that they always treat clinically. Correct malaria diagnostic usage
was reported more often by HF providers (97%) versus drug outlets (85%) (p=0.018). Of those
providers that reported always (59%) or sometimes performing a pregnancy assessment, 79%
reported asking for LMP and 48% reported offering a pregnancy test. Knowledge of correct
pregnancy assessment was higher among health facilities providers (93%) versus their drug outlet‐
based counterparts (49%, p<0.001) (Table 15).
Knowledge and adherence to malaria in pregnancy treatment guidelines
133
MalariaTreatment&TreatmentContraindicationKnowledge
Thirty‐eight (51%) HF providers knew the correct 1st‐line treatment and dosage for all scenarios
(non‐pregnant, 1st trimester, and 2nd/3rd trimester women) compared to none of the 39 drug
dispensers. Correct knowledge specifically for 1st trimester patients was given by 56% of HF
providers and none of drug dispensers (p<0.0001). Correct treatment knowledge for 2nd/3rd
trimester patients was reported in 87% of HF and 39% of drug outlets (p<0.0001). Overall provider
knowledge was considerably higher for 1st‐line treatment versus 2nd‐line treatment (p<0.0001, Table
15).
Table 15. Adequate provider knowledge of malaria in pregnancy based on national malaria treatment guidelines comparing health facilities to drug outlets
Overall Health Facilities Drug Outlets P‐value*
N % N % N %
114 75 39
Consequences of MiP 112 98.2 74 98.7 38 97.4 p=0.34
Awareness of MTGs 85 74.6 74 98.7 11 28.2 <0.0001
Malaria Diagnostics 107 93.9 73 97.3 33 84.6 p=0.01
Pregnancy Assessment 89 78.1 70 93.3 19 48.7 p<0.0001
Treatment & Dosage 38 33.3 38 50.7 0 0.0 p<0.0001
NP‐1st Line 97 85.1 72 96.0 25 64.1 p<0.0001
NP‐2nd Line 43 37.7 37 49.3 6 15.4 p=0.0002
1st Tri‐ 1st Line 42 36.8 42 56.0 0 0.0 p<0.0001
2nd/3rd Tri‐ 1st Line 80 70.2 65 86.7 15 38.5 p<0.0001
Severe Malaria 72 63.2 63 84.0 9 23.1 p<0.0001
Adequate Knowledge 34 29.8 34 45.3 0 0.0 p<0.0001
*P‐values from Chi‐square test and Fisher Exact used for strata with <5 observations. Acronyms: MTG, malaria treatment guidelines, MiP, malaria in pregnancy, NP, non‐pregnant; Tri, trimester of pregnancy
SP was incorrectly cited as 1st and 2nd line treatment for adults by 4% of providers, in 1st trimester
pregnant patients by 19% of providers, in 2nd/3rd trimester by 7% of providers, and for treatment of
severe malaria in pregnancy by 5% of providers. Two‐thirds cited IPTp with SP as a preventive
measure for MiP; however, only 54% knew that SP could only be used as preventive therapy and not
as treatment. Additionally, 6% of all providers thought AL could be used as preventive therapy and
another 14% were not able to cite a drug for preventive treatment. An ACT was incorrectly cited as
the appropriate treatment in 1st trimester pregnancies by 5% of HF providers and 18% of drug outlet
providers.
The majority of providers stated the reason behind their chosen antimalarial for 1st and 2nd‐line
treatment was either due to the observed effectiveness of the drug in their practice (45% and 36%
,respectively) or due to national guidelines (35% and 30%, respectively). However, of the providers
that cited the national guidelines as the reason, 83% and 74% had chosen an antimalarial not
recommended by the national MTGs for 1st‐line and 2nd‐line, respectively. 71% of HF providers and
28% of drug outlet dispensers cited 1st trimester as a contraindication for AL treatment (p<0.0001).
37% of HF providers cited ‘allergy’ as a contraindication for both 1st and 2nd line treatments versus
13% (p=0.003) and 3% (p<0.0001) of drug outlet providers, respectively.
Chapter 7
134
Correct treatment knowledge for severe malaria in pregnancy was given by 84% of HF providers and
23% of drug outlet providers (p<0.0001). Overall adequate knowledge of MiP (inclusive of
diagnostics, pregnancy assessment, and treatment knowledge) was reported by 34 providers (30%),
all of which were HF based.
MalariainPregnancyComprehensiveCareKnowledge
85% of HF providers reported giving any type of comprehensive care practices to pregnant patients
with malaria versus 26% of drug outlet providers (p=0.8). Fetal monitoring (60% HF, 17% drug outlet)
and anemia treatment (61% HF, 10% drug outlet) were the most commonly cited practices, followed
by hypoglycaemia prevention (43% HF, 0% drug outlet) and guidance on antipyretic usage (32% HF,
3% drug outlet). 87% of all providers reported that they give pregnant patients instructions on
treatment; 56% reported informing a pregnant patient of potential side effects, 52% reported telling
the patient to return if symptoms continued, and 7% reported informing the patient of danger signs
to look out for. A greater proportion of HF providers than drug outlet‐based providers reported
giving such information to a pregnant patient (p<0.001), with the exception of danger signs (Table
16).
Table 16. Comprehensive care practices provided during pregnancy, comparing health facilities vs. drug outlets
Overall Health Facilities Drug Outlets P‐value*
N % N % N %
114 75 39
Care Practices in Pregnancy
Prevent Hypoglycemia 32 28.1 32 42.7 0 0.0 p<0.001
Fetal Monitoring 52 45.6 45 60.0 7 17.9 p<0.001
Anemia Treatment 50 43.9 46 61.3 4 10.3 p<0.001
Antipyretics 25 21.9 24 32.0 1 2.6 p<0.001
None 8 7.0 0 0.0 8 20.5 p<0.001
Other* 55 48.2 25 33.3 30 76.9 p<0.001
Information provided with Treatment
Instructions 99 86.8 71 94.7 28 71.8 p<0.001
Side Effects 64 56.1 57 76.0 7 17.9 p<0.001
Return if Symptoms Continue 59 51.8 47 62.7 12 30.8 p<0.001
Danger Signs** 8 7.0 5 6.7 3 7.7 p=0.41
Other 13 11.4 3 4.0 10 25.6 p<0.001
Any Information Given 104 91.2 74 98.7 30 76.9 p<0.001
*P‐values from Chi‐square test and Fisher Exact used for strata with <5 observations **Included nutritious diet, ITNs, IPTp, and medication compliance. *** Danger signs included death, convulsions, dizziness, spotting, & fetal movement
PredictorsofKnowledgeMalariainPregnancyTreatmentGuidelines
Important predictors of adequate knowledge of MiP diagnostic and prescribing practice for
providers in health facilities included facility type, respondent cadre, dispensing medication, gender,
and having received trainings on malaria diagnostics and MiP (Table 17). When compared to nurses,
pharmacists had higher odds to have adequate knowledge (OR= 1.6 95%CI [1.0‐2.6]) and providers
that identified as non‐clinically trained staff were less likely to possess adequate knowledge (OR=0.6
Knowledge and adherence to malaria in pregnancy treatment guidelines
135
95% CI [0.5‐0.8]). Female providers had higher odds of adequate knowledge criteria than male
providers (OR=1.9 95% CI [1.5‐2.4]). Providers that had specifically attended workshops on malaria in
pregnancy were more likely to possess adequate knowledge than those who had not (OR=1.4 95% CI
[1.1‐1.8]).
Table 17. Health facility provider predictors of adequate knowledge of malaria in pregnancy treatment guidelines
Provider Characteristic N % Crude OR
95% CI P value Adjusted*
OR 95% CI P value
Facility Type 75
Health Center (ref) 28 37.3 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐
Hospital 6 8.0 0.6 (0.5, 0.9) 0.00 0.6 (0.4, 1.2) 0.15
Dispensary 41 54.7 0.8 (0.6, 1.0) 0.06 0.9 (0.7, 1.1) 0.21
Respondent Cadre
Nurse (ref) 48 64.0 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐
Clinical Officer/M.D. 17 22.7 0.9 (0.7, 1.2) 0.60 1.2 (0.9, 1.6) 0.14
Pharmacist 5 6.7 0.9 (0.6, 1.5) 0.69 1.6 (1.0, 2.6) 0.05
Other** 5 6.7 0.6 (0.5, 0.7) <0.001 0.6 (0.5, 0.8) <0.001
Dispenses Medicine
No (ref) 9 12.0 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐
Yes 66 88.0 1.3 (1.0, 1.8) 0.10 1.2 (0.9, 1.7) 0.15
Malaria diagnosis Training within 5 years
No (ref) 25 33.3 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐
Yes 50 66.7 1.3 (1.1, 1.7) 0.01 1.4 (1.0, 1.9) 0.06
MiP Training within 5 years
No (ref) 47 62.7 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐
Yes 28 37.3 1.2 (1.0, 1.5) 0.09 1.4 (1.1, 1.8) <0.01
Sources of Information
CME as a source of info 52 69.3 1.3 (1.1, 1.7) 0.01 ‐‐ ‐‐ ‐‐
Gender
Male (ref) 39 52.0 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐
Female 36 48.0 1.2 (1.0, 1.5) 0.10 1.9 (1.5, 2.4) <0.001
Age 1.0 (1.0, 1.0) 0.60 1.0 (1.0, 1.0) 0.32
*Adjusted model included facility type, respondent cadre, dispensing medicine, Malaria diagnostic training and MiP training, gender, and age. CME dropped from multivariate model due to non‐significance. **Other includes community health workers, Village Reporters, or other technical staff at the health facility.
Correct knowledge of treatment and dosage in 1st trimester patients was removed from the
adequate knowledge definition for logistic regression with drug outlet providers because none of the
providers interviewed were able to provide the correct response. Facility type was a predictor of
adequate knowledge among drug outlet providers (Table 18). As compared to providers working in a
general shop, providers in registered pharmacies were more likely to know correct pregnancy
assessment, malaria diagnostic information and correct treatment and dosage in non‐pregnant and
2nd/3rd trimester patients, (OR= 1.1, 95%CI [1.1‐1.9]). Age, gender and respondent cadre were
controlled for, though they were not found to be significant predictors of knowledge.
Chapter 7
136
Table 18. Drug outlet dispenser predictors of adequate knowledge of malaria in pregnancy treatment guidelines
Provider Characteristic N % Crude OR
95% CI P value Adjusted
OR 95% CI P value
Facility Type 39
General Shop (ref) 15 38.5 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐
Registered Pharmacy 9 23.1 1.2 (1.0, 1.9) 0.03 1.1 (1.1, 1.9) <0.01
Informal Drug Outlet 15 38.5 1.2 (1.0, 1.5) 0.05 1.1 (0.9, 1.5) 0.25
Respondent Cadre
Pharmacist (ref) 15 38.5 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐
Clinical Officer/M.D./Nurse 2 5.1 1.5 (0.7, 2.8) 0.42 1.4 (0.6, 2.5) 0.52
Shopkeeper/Other** 22 56.4 1.1 (0.7, 1.1) 0.36 1.1 (0.7, 1.2) 0.38
Malaria diagnosis Training within 5 years
No (ref) 31 79.5 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐
Yes 8 20.5 1.2 (0.8, 1.6) 0.46 ‐‐ ‐‐ ‐‐
MiP Training within 5 years
No (ref) 35 89.7 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐
Yes 4 10.3 1.3 (0.7, 1.7) 0.63 ‐‐ ‐‐ ‐‐
Gender
Male (ref) 15 38.5 ‐‐ ‐‐ ‐‐ ‐‐ ‐‐ ‐‐
Female 24 61.5 1.1 (0.7, 1.1) 0.15 1.1 (0.8, 1.1) 0.39
Age 1.0 (1.0, 1.0) 0.24 1.0 (1.0, 1.0) 0.02
*Adjusted model included facility type, respondent cadre, gender and age. **Other included community health workers and Village Reporters
DISCUSSION
This study found that correct malaria case management in pregnant women with symptoms of
uncomplicated malaria in Siaya County is low overall, particularly in the 1st trimester. Only 43% of
WOCBA who were not visibly pregnant were assessed for a potential pregnancy in health facilities
and only 7% of the female simulated clients in drug outlets were assessed without being prompted.
Only 32% of women diagnosed with malaria in their 1st trimester attending health facilities and none
of the 1st trimester simulated clients attending drug outlets were correctly offered quinine as per
national guidelines.
The failure of providers to assess for possible pregnancy in a large proportion of women is
problematic and most often results in inadvertently exposing early pregnancies to ACTs. In addition,
this represents a missed opportunity to refer women for early antenatal care in an area where most
women initiate ANC late in pregnancy [21]. In our study, five of the women who were not assessed
for pregnancy by providers had an LMP consistent with a 1st trimester pregnancy, and all of these
women received AL. Likewise, all six women who reported being unsure of their pregnancy status
were prescribed AL; all these women warranted further assessment before prescribing AL.
The most common mistakes in prescribing practices were giving AL in early pregnancy, SP as
treatment, and the incorrect dosage of quinine. Prescribing of AL in known early pregnancy was very
common, consistent with our observation in the cohort studies of women of child bearing age
(Dellicour chapter 8). This is likely reflective of a lack of knowledge by healthcare providers regarding
the potential teratogenicity; when surveyed later, only 56% of providers reported they were aware
of contraindications to using ACTs in the 1st trimester. Overall, contraindicated regimens were
prescribed in 65% of 1st trimester pregnancies (there was no stock‐out of quinine reported); this is
Knowledge and adherence to malaria in pregnancy treatment guidelines
137
similar to the 70% self‐reported exposure rate in early pregnancy Uganda in 2011 [11]. Correct
prescribing of AL in 61% of women who were not pregnant or in late pregnancy is similar to rates
observed previously [9,10]; however, only 75% of women received the correct dosage for AL in our
study.
The prescribing pattern in drug outlets uncovers a more troubling and complex knowledge issue.
Dispensers in drug outlets initially prescribed AL to over 90% of simulated client patients. Upon
being informed of pregnancy status, treatment was changed from AL to SP in 27% of cases due to
finding out the patient was pregnant, regardless of trimester. There was also a tendency among
dispensers to withdraw AL and refer the woman to a health facility upon learning of pregnancy
status. This occurred in only 13% of early pregnancy cases and occurred incorrectly in 21% of third‐
trimester pregnancy cases. In the provider survey, only 28% of drug outlet dispensers were aware of
contraindications to AL in 1st trimester.
The prescribing of SP in place of AL indicates that some dispensers still incorrectly believe that SP
can be used as treatment, which was supported by the provider survey in which 49% of drug outlets
providers and 33% of HF providers incorrectly reported that SP could be used for both treatment
and preventive purposes. This is alarming given that Kenya changed the national malaria treatment
policy to recommend artemisinin‐based combination therapy (ACT) as the first‐line treatment for
uncomplicated malaria in 2004, and by 2006, the new ACT guidelines were implemented
countrywide [3, 9]. Overall, SP was prescribed as treatment in 11% of all cases; 37% to women in
their 1st trimester. Using SP as treatment is likely to result in a high risk of treatment failure in this
area with high levels of SP resistance with potential for immediate consequences that affect both
the mother and fetus (Desai et al, personal communications & [22]).
Over 70% of all women that were prescribed quinine were given an insufficient supply (often 3‐5
days rather than the recommended 7) and incorrect instructions (incorrect number of tablets or
times per day), resulting in prescriptions ranging from 10‐70% of the full dose, increasing the risk of
treatment failure and malaria relapse, and the development of drug resistance [23,24]. This is
particularly troubling given that quinine is currently the only safe and effective treatment available
to women in early pregnancy. Poor knowledge of correct quinine dosage versus that of other
commonly prescribed antimalarials such as AL was likewise observed in the provider survey.
Although almost all HF providers reported a high rate of awareness of the national MTGs versus
only slightly more than a quarter of drug outlet dispensers, provider knowledge in both settings was
poor and were reflective of the low levels of correct case management observed in practice. The
greatest knowledge deficiencies were observed in pregnancy assessment, correct treatment and
drug regimen. Although a number of providers were aware of contraindications in 1st trimester,
knowledge of the correct treatment for 1st trimester patients was low for both HFs and especially
drug outlets, where not a single provider cited quinine as the correct drug of choice, consistent with
the observed practices.
The only consistently significant indicators for correct practice and/or knowledge were the
provider’s professional cadre in health facilities. MiP training in the last 5 years was only a predictor
of correct knowledge for health facility providers and correct practice in drug outlet dispensers. It is
likely that the number of dispensers in our study was underpowered for the purposes of MiP training
assessment. A push for dissemination of the MTGs in drug outlets may result in improved MiP
practice, however targeted trainings for MiP and overall capacity building via increased contact
Chapter 7
138
between non‐clinically trained dispensers and qualified healthcare professionals is likely to have
more of an impact in drug outlets; similar documented efforts have been shown to be effective in
the region [25,26]
Limitations&Challenges
The relatively short time‐frame of the overall study limited the number of exit interviews completed
at each facility. In particular, the identification of febrile patients in 1st trimester for interview was
challenging, possibly due to shortcomings in early pregnancy detection. Gestational age assessment
was based on reported LMP which could have lead to pregnancy trimesters misclassification for late
1st trimester pregnancies. Unless the provider had an alternative approach to assess gestation (such
as fundal height) it is unlikely that assessment of correct practice would have been affected.
It is likely that correct diagnostic practice was overestimated as the diagnostic capacity of health
facilities or drug outlets was not collected at the time of exit interviews nor simulated client
interactions. It was assumed that drug outlets, health centers and dispensaries didn’t have access to
diagnostic tests if none of the participants at these facilities received a diagnostic test and clinical
diagnosis in these facilities was considered correct.
Exit interviews and provider surveys were susceptible to courtesy bias. Information obtained from
patients via exit interviews may also be biased due to patient recall/information loss, although this
was minimized by conducting the interview immediately upon completion of the consultation. In
addition, errors may have been introduced if the patient did not understand the information given
or procedures done when in the presence of the provider.
CONCLUSION
Observed practice in health facility and drug outlet settings indicates a very low uptake of standard
clinical practice to assess all women of childbearing age for pregnancy and poor prescribing
practices, particularly for the case‐management of 1st trimester uncomplicated malaria. Timely,
significant efforts to increase pregnancy assessment and correct prescribing practices are required
to improve the safe and effective case management of malaria in pregnancy.
Disclaimer
The findings and conclusions presented in this manuscript are those of the authors and do not
necessarily reflect the official position of the U.S. President’s Malaria Initiative, United States Agency
for International Development, or U.S. Centers for Disease Control and Prevention.
AcknowledgementsWe are grateful to the communities of Asembo, Gem and Karemo for their participation in and
support of the HDSS. We also thank numerous field, clinical, data and administrative staff, without
whom, this work would not have been possible. We thank INDEPTH for their ongoing collaboration
to strengthen and support health and demographic surveillance systems; the KEMRI/CDC Research
and Public Health Collaboration is a member of the INDEPTH Network. This paper is published with
the permission of the Director, KEMRI.
Knowledge and adherence to malaria in pregnancy treatment guidelines
139
Financialsupport This publication was made possible through support provided by the United States President’s
Malaria Initiative, U.S. Agency for International Development and U.S. Centers for Disease Control
and Prevention (CDC), under the terms of an Interagency Agreement with CDC and through a
Cooperative Agreement between the CDC and the Kenya Medical Research Institute (KEMRI). The
sponsor of the study had no role in study design, data collection, data analysis, data interpretation,
or writing of the report. The corresponding author had full access to all of the data in the study and
had final responsibility for the decision to submit for publication.
PotentialconflictsofinterestAll authors: No reported conflicts.
References1. Dellicour S, Tatem AJ, Guerra CA, Snow RW, ter Kuile FO (2010) Quantifying the number of pregnancies at
risk of malaria in 2007: a demographic study. PLoS Med 7: e1000221. 2. Desai M, ter Kuile FO, Nosten F, McGready R, Asamoa K, et al. (2007) Epidemiology and burden of malaria in
pregnancy. Lancet Infect Dis 7: 93‐104. 3. Ward SA, Sevene EJ, Hastings IM, Nosten F, McGready R (2007) Antimalarial drugs and pregnancy: safety,
pharmacokinetics, and pharmacovigilance. Lancet Infect Dis 7: 136‐144. 4. WHO (2003) Assessment of the safety of artemisinin compounds in pregnancy. Report of two informal
consultations convened by WHO in 2002. Geneva: WHO. 5. WHO (2007) Assessment of the safety of artemisinin compounds in pregnancy: report of two joint informal
consultations convened in 2006. Geneva: WHO. 6. WHO (2010) Guidelines for the treatment of malaria. Geneva: World Health Organisation. 7. DOMC (2010) Division of Malaria Control National guidelines for the diagnosis treatment and prevention of
malaria in Kenya.: Ministry of Public Health and Sanitation, Nairobi. 8. WHO (2009) Medicines use in primary care in developing and transitional countries: Fact Book summarizing
results from studies reported between 1990 and 2006 Geneva: World Health Organization. 9. Zurovac D, Njogu J, Akhwale W, Hamer DH, Larson BA, et al. (2008) Effects of revised diagnostic
recommendations on malaria treatment practices across age groups in Kenya. Trop Med Int Health 13: 784‐787.
10. Nyandigisi A, Memusi D, Mbithi A, Ang'wa N, Shieshia M, et al. (2011) Malaria case‐management following change of policy to universal parasitological diagnosis and targeted artemisinin‐based combination therapy in Kenya. PLoS One 6: e24781.
11. Sangare LR, Weiss NS, Brentlinger PE, Richardson BA, Staedke SG, et al. (2011) Patterns of anti‐malarial drug treatment among pregnant women in Uganda. Malar J 10: 152.
12. Kamuhabwa A, Jalal R (2011) Drug use in pregnancy: Knowledge of drug dispensers and pregnant women in Dar es Salaam, Tanzania. Indian Journal of Pharmacology 43: 345‐349.
13. Aviv RI, Chubb K, Lindow SW (1993) The prevalence of maternal medication ingestion in the antenatal period. S Afr Med J 83: 657‐660.
14. Kebede B, Gedif T, Getachew A (2009) Assessment of drug use among pregnant women in Addis Ababa, Ethiopia. Pharmacoepidemiol Drug Saf 18: 462‐468.
15. Rohra DK, Das N, Azam SI, Solangi NA, Memon Z, et al. (2008) Drug‐prescribing patterns during pregnancy in the tertiary care hospitals of Pakistan: a cross sectional study. BMC Pregnancy Childbirth 8: 24.
16. Odhiambo FO, Laserson KF, Sewe M, Hamel MJ, Feikin DR, et al. (2012) Profile: the KEMRI/CDC Health and Demographic Surveillance System‐‐Western Kenya. Int J Epidemiol 41: 977‐987.
17. Ouma P, van Eijk AM, Hamel MJ, Parise M, Ayisi JG, et al. (2007) Malaria and anaemia among pregnant women at first antenatal clinic visit in Kisumu, western Kenya. Trop Med Int Health 12: 1515‐1523.
18. Perrault SD, Hajek J, Zhong K, Owino SO, Sichangi M, et al. (2009) Human immunodeficiency virus co‐infection increases placental parasite density and transplacental malaria transmission in Western Kenya. Am J Trop Med Hyg 80: 119‐125.
19. Huntington D, Schuler SR (1993) The simulated client method: evaluating client‐provider interactions in family planning clinics. Stud Fam Plann 24: 187‐193.
Chapter 7
140
20. Madden JM, Quick JD, Ross‐Degnan D, Kafle KK (1997) Undercover careseekers: simulated clients in the study of health provider behavior in developing countries. Soc Sci Med 45: 1465‐1482.
21. Kenya National Bureau of Statistics (KNBS) and ICF Macro (2011) Kenya Demographic and Health Survey 2008‐09. Calverton, Maryland, USA. http://www.measuredhs.com/pubs/pdf/FR229/FR229.pdf.
22. McCollum AM, Schneider KA, Griffing SM, Zhou Z, Kariuki S, et al. (2012) Differences in selective pressure on dhps and dhfr drug resistant mutations in western Kenya. Malar J 11: 77.
23. Cheruiyot J, Ingasia LA, Omondi AA, Juma DW, Opot BH, et al. (2014) Polymorphisms in Pfmdr1, Pfcrt, and Pfnhe1 Genes Are Associated with Reduced In Vitro Activities of Quinine in Plasmodium falciparum Isolates from Western Kenya. Antimicrob Agents Chemother 58: 3737‐3743.
24. Tarning J, Kloprogge F, Dhorda M, Jullien V, Nosten F, et al. (2013) Pharmacokinetic properties of artemether, dihydroartemisinin, lumefantrine, and quinine in pregnant women with uncomplicated plasmodium falciparum malaria in Uganda. Antimicrob Agents Chemother 57: 5096‐5103.
25. Brantuo MN, Cristofalo E, Mehes MM, Ameh J, Brako NO, et al. (2014) Evidence‐based training and mentorship combined with enhanced outcomes surveillance to address the leading causes of neonatal mortality at the district hospital level in Ghana. Trop Med Int Health 19: 417‐426.
26. Hanson C, Waiswa P, Marchant T, Marx M, Manzi F, et al. (2014) Expanded Quality Management Using Information Power (EQUIP): protocol for a quasi‐experimental study to improve maternal and newborn health in Tanzania and Uganda. Implement Sci 9: 41.
141
Chapter8
Riskof inadvertentexposure toartemisininderivatives in thefirst trimester of pregnancy and its association withmiscarriage:aprospectivestudyinWesternKenya
StephanieDellicour1,2,MeghnaDesai1,3,GeorgeAol1,MartinaOneko1,PeterOuma1,GodfreyBigogo1,DeronBurton3,RobertBreiman3,MaryHamel3,LarrySlutsker3,DannyFeikin3,SimonKariuki1,FrankOdhiambo1,JayeshPandit4,KaylaF.Laserson3,5,N.Yan1,2, GregCalip,AndyStergachis5,FeikoO.terKuile1,2
1 Kenya Medical Research Institute and Centres for Disease Control and Prevention
(KEMRI/CDC)Research and Public Health Collaboration, Kisumu, Kenya, 2 Liverpool School
of Tropical Medicine, Liverpool, United Kingdom,3 Centres for Disease Control and
Prevention, Atlanta GA, United States of America, 4 Kenya Pharmacy and Poisons Board
(KPPB), Kenya, 5 Departments of Epidemiology and Global Health, University of
Washington, United States of America
To be submitted
Chapter 8
142
AbstractBackground: Artemisinins are widely deployed as artemisinin‐based combination therapy (ACT),
including as first‐line therapy for malaria in the 2nd and 3rd trimester of pregnancy. However, they are
not recommended for uncomplicated malaria during the 1st trimester because they are embryo‐toxic
in animal models and safety data from human early pregnancies are scarce. More safety data are
potentially accessible as inadvertent exposure in early pregnancy to ACTs is common, but methods
to capture this in resource poor settings have not been developed.
Methods: We conducted a prospective cohort study of women of childbearing age to document the
relationship between ACT exposure during the 1st trimester and the risk of miscarriage in a
population under health and demographic surveillance in western Kenya. Community‐based
surveillance was used to identify early pregnancies. To identify pregnancies exposed to ACTs and
other antimalarials in the 1st trimester, record linkage was used to combine study‐specific pregnancy
follow‐up questionnaire data with health surveillance data from out‐patient records and weekly or
twice‐monthly household‐level surveillance to ascertain recent history of symptoms and drug intake.
Cox proportional hazard models with left truncation were used accounting for potential survival bias
and controlling for confounders.
Results: In 2011‐2013, among 5,536 women of childbearing age, 1,453 pregnancies were detected;
1,126 of which were included in the analysis. The cumulative probability of miscarriage by 28 weeks
gestation was 19%. 296 pregnancies (26%) had data indicative of exposure to ACTs in the 1st
trimester, 12 to quinine (1%) and an additional 22 (2%) to both. There were 75 (7%) confirmed
exposures in the 1st trimester, defined as matching exposure data in 2 of 3 source databases, and 42
(4%) confirmed ACT exposures in the suggested artemisinin embryo‐sensitive period (6‐12 weeks
post last menstruation). Pregnancies with confirmed ACT exposures in the 1st trimester, overall, or
during the embryo‐sensitive period where not at a significant increased risk of miscarriage relative to
women not exposed to antimalarials [Hazard ratio (HR)=1.60 95% CI (0.70‐3.68) and HR=0.85 95% CI
(0.22‐3.33) respectively]. Similar findings were observed for the small numbers exposed to quinine
alone compared to pregnancies unexposed to antimalarials: 1st trimester HR=1.53 95%CI (0.19‐
12.40); or 6‐12 weeks gestation HR=2.02 95%CI (0.24‐17.10).
Conclusion: ACT exposure in early pregnancy was more common than quinine exposure. No
significant association between confirmed early artemisinin exposure and miscarriage was found.
The limited sample size rules out a 3.7 fold or greater increased risk of miscarriage associated with
artemisinin exposure in the 1st trimester. Further confirmatory studies will be needed. Targeting
malaria prevention strategies for women in the early stages of pregnancy should be a priority.
Association between artemisinin exposure in the first trimester and miscarriage
143
BackgroundThe artemisinin‐based combination therapy (ACT) antimalarials have been adopted as first‐line
treatment for P. falciparum in almost all endemic countries, providing life‐saving benefits to
children, adults and pregnant women globally.[1] However, their safety when used in early
pregnancy is uncertain. Artemisinins are embryotoxic in several animal species, including primate
models. Teratogenic effects observed in mice and rabbits included death of the fetus, malformations
of the heart, great vessels and limb defects. Primate models exposed to prolonged courses of ACTs
had high rates of fetal loss.[2] Animal models suggested that artemisinin embryotoxicity targets
primitive erythroblasts which are the primary form of red blood cells in circulation between weeks 4‐
10 post‐conception in humans. Therefore the suggested artemisinin embryo‐sensitive period in
humans, if any, is thought to occur at 6‐12 weeks gestation post‐last menstrual period
(LMP).[3,4,5,6]
There are limited data available to assess whether ACTs are similarly embryotoxic or teratogenic in
humans; fewer than 700 exposures in the 1st trimester have been well‐
documented.[7,8,9,10,11,12,13] After reviewing all existing evidence in 2003 and then in 2006, the
WHO recommended that artemisinins could be used during the second or third trimesters of
pregnancy and that, due to lack of sufficient safety data, treatment in the 1st trimester was not
recommended unless the life of the mother is at risk, or oral quinine is not available.[4,14] However,
as women may not be aware of their pregnancy or do not report an early pregnancy and because
clinicians often do no assess for pregnancy in women of childbearing age (WOCBA), the risk of
exposure to drugs not recommended in pregnancy, including to potential teratogens is high. We
have previously reported that in malaria endemic countries in Africa, up to 2 million pregnancies
could be inadvertently exposed to ACTs annually during a suggested embryo‐sensitive
time‐window.[15]
Malaria can have severe consequences to the health of the pregnant woman and her unborn baby
including maternal anaemia, fetal loss, preterm birth, low birth weight and perinatal mortality, and
in some cases maternal death. The impact of malaria infection in early pregnancy had been
identified as a major knowledge gap for estimating the burden of malaria in pregnancy.[16] Recent
studies provided insight into the potential adverse consequences of malaria infections early in
pregnancy with major impact on birth weight and maternal anaemia [17,18], and modelling by
Walker et al, suggest that as many as two thirds of placental malaria infections can be prevented by
avoiding exposure during the first 12 weeks of pregnancy.[19] Placental infections are thought to
play an important role in the adverse effect of malaria in pregnancy due to restricted nutrition flow
through the placenta.[20] McGready et al reported the findings from a retrospective analysis from
25 years of data from their research clinics on the Thai‐Burmese border and showed that malaria
infection in the 1st trimester of pregnancy (both symptomatic and asymptomatic) was a significant
risk factor for miscarriage.[11,21] They found no association between miscarriage and 1st trimester
ACT exposure. However more data from a wider range of malaria‐endemic countries are required to
provide an increased level of reassurance. We report the findings from a prospective cohort study of
WOCBA designed to examine whether ACT exposure in the 1st trimester is associated with
miscarriage.
Chapter 8
144
Methods
OverviewofstudydesignThis was a cohort study conducted among women aged 15‐49 years residing in the KEMRI/CDC
health and demographic surveillance system area (KEMRI/CDC HDSS) in a highly malarious area in
western Kenya. The study site also included facility‐ based and household level surveillance of recent
history of illness, symptom and drug intake [See Appendix for more details]. Pregnancies were
identified through health facility and community‐based strategies (described below), and followed
prospectively (i.e. before the pregnancy outcome was known) to document birth outcome. As this
was an observational study, the study staffs were not involved in treatment of study participants.
Participants were informed and reminded throughout the study that the purpose of the study was to
monitor safety of antimalarial medicine in early pregnancy and that ACTs are not recommended in
the 1st trimester. Participants received treatment through the usual channels including health
facilities and drug outlets.
RecruitmentofwomenofchildbearingageandpregnancydetectionBetween February 15, 2011 and February 15, 2013, WOCBA participating in an ongoing population‐
based infectious disease surveillance project (PBIDS) in rural western Kenya [22,23] were invited to
participate in the “Evaluation of Medications used in Early Pregnancy” (EMEP) cohort study. The
PBIDS catchment population included 33 villages within 5 km of Lwak Hospital, the designated
referral health facility in the area of the KEMRI/CDC HDSS located in Asembo, Bondo district.[24,25]
All PBIDS participants were visited weekly (from January 5, 2010 to May 26, 2011) and then twice
monthly (May 27, 2011 onwards) at home to collect information on any symptoms, where care was
sought and whether any medications were taken. PBIDS participants received free care at the
referral facility for all potential infectious illnesses and detailed information on diagnosis and
treatment was recorded. EMEP study staff visited all homes in the PBIDS and enrolled consenting
WOCBA, who met eligibility criteria for EMEP. WOCBA were eligible for EMEP if they were between
15 and 49 years of age and were active participants of PBIDS. EMEP exclusion criteria were the
following: refusal to participate or to be followed up at the end of pregnancy; and any condition
(physical, social or mental) that would interfere with the ability to give informed consent or to
provide an accurate medical history. WOCBA who consented to participate were asked if they could
be pregnant and offered a pregnancy test at the time of enrolment and again approximately every 3
months thereafter. Any participant with a detected pregnancy was referred to the antenatal clinic at
Lwak Hospital where trained EMEP study nurses would confirm the pregnancy and offer free
antenatal care (ANC). Additionally, all pregnant patients presenting at Lwak Hospital were assessed
for EMEP study eligibility by the study nurse and enrolled if all criteria were met.
GestationalageassessmentGestational age was assessed using multiple methods, including ultrasound scans at the 1st antenatal
visit at Lwak ANC (for participant presenting before 24 weeks); reported 1st day of LMP; reported
gestational age at the time of pregnancy loss; Ballard scoring for live‐births captured within 7 days of
delivery [26]; and fundal height measurements recorded at antenatal checkups. Not all methods
were available for all pregnancies since some were not seen at ANC (no fundal height or ultrasound
measurement available) or were seen at ANC but beyond 23 weeks. Ballard score was only available
for live‐births seen within 7 days of delivery. Furthermore some participants couldn’t recall their
LMP or in some instances hadn’t resumed their menses since their previous pregnancy. Therefore,
Association between artemisinin exposure in the first trimester and miscarriage
145
for this analysis, gestational age was determined using the most accurate measurement available for
each participant. Methods in order of decreasing accuracy were: ultrasound scan taken before 24
weeks gestation, Ballard estimates, LMP or reported gestation at time of pregnancy loss and lastly
gestational age derived from fundal height assessment [see Appendix].
PregnancyoutcomePregnancy outcomes were assessed using a combination of health facility‐based and home‐based
follow‐ups. The latter is particularly important for miscarriages, because the vast majority do not
occur at health facilities. Village‐based staff received monthly lists of participants with pending
estimated delivery dates in their respective catchment area. Study nurses were notified of
pregnancy outcomes by village‐based staff and follow ups were done either at home or at the health
facility. A detailed structured questionnaire about the delivery and outcome was administered in
addition to questions regarding any illnesses and medication used during pregnancy. Pregnancy
outcomes captured included pregnancy losses (miscarriages, induced abortions and stillbirths), live‐
births and major congenital malformations detectable at births by surface examination. The analysis
presented in this manuscript focuses on miscarriages defined as spontaneous pregnancy loss at or
before 28 completed weeks gestation (2‐28 weeks inclusive), which is considered the gestational age
of viability in resource constrained settings.[27] Risk of other adverse pregnancy outcomes
(stillbirths and major congenital malformations) associated with ACT exposures in early pregnancy
will be reported elsewhere.
AntimalarialdrugexposureascertainmentDrug exposure data were captured using 3 approaches: a) interviews with pregnant women visiting
the antenatal clinic in Lwak Hospital and at the time of pregnancy outcome follow up (from here on
referred to as EMEP data); b) record linkage of data on drugs prescribed to WOCBA at the outpatient
department in Lwak Hospital (from here on referred to as Lwak OPD data) and c) weekly to twice
monthly home visits by fieldworkers as part of PBIDS (Table 1).
Table 10. Description of drug information sources used to determine antimalarial and malaria exposure status
Approach Format Drug Information available
EMEP self‐report Retrospective self‐report of illness and medication used since the beginning of the pregnancy collected at every ANC visit and at pregnancy outcome follow up visit. A general open question about any drug use as well as a prompted question for specific antimalarials were included as using medication/indication‐specific questions have been shown to improve accuracy.[28,29,30] Visual aids depicting photographs of all antimalarial drugs found in the study area were used to facilitate recognition of drug names. Calendar marking public holidays and school closures was also used to enhance recall of dates.
Drug name
Drug start date Duration Number of tablets per day
Indication and indication diagnosis
Drug source
Lwak OPD records Prospective documentation by health facility clinician of diagnosis and treatment prescribed at out‐patient department (OPD) whenever a PBIDS participant sought care at Lwak Hospital for an infectious syndrome
Date of visit Diagnosis Prescribed treatment
PBIDS weekly & twice‐monthly home visits
Self‐report of symptoms, health‐seeking behaviour and medication. This information was collected continuously on a weekly (from January 5, 2010 to May 26, 2011) and then twice‐monthly basis (May 27, 2011 onwards). The same visual aids as described above were used for recall of drug intake.
Date of visit Symptoms in previous week/2 weeks
Treatment taken for the symptoms including drug name
If and where care was sought
Chapter 8
146
MalariainfectionInformation on malaria was captured through interviews by study nurses or through OPD records for
those who sought care from Lwak Hospital. Reliable information on malaria diagnosis was only
available from Lwak OPD which has the laboratory capacity to perform malaria slide microscopy
through PBIDS.[23]EMEP nurses asked about recent care‐seeking and diagnosis for each reported
illness including malaria episodes but reliable information on laboratory confirmed malaria was
mostly not available.
OtherCovariatesObstetric history and ANC laboratory information collected routinely at antenatal booking
(haemoglobin level, HIV and syphilis testing, and malaria microscopy) were abstracted by study
nurses from the ANC records at Lwak hospital or antenatal cards.
Dataanalysis
Miscarriagerateandsurvivalprobabilities Analyses were performed using Stata v12.1 (StataCorp LP, College Station, Texas). Survival analysis
with left truncation was used to estimate the miscarriage rate by gestational week to account for
delayed pregnancy detection and the range in gestational ages at the time of pregnancy detection.
Left truncation was used to account for survival bias as the average gestational age that pregnancies
were detected was only around 13 weeks and only pregnancies that survive the early weeks of
gestation (the highest risk of miscarriage) were followed prospectively.[31,32] The Kaplan–Meier
product‐limit method was used to calculate the cumulative probability of survival and cumulative
probability of miscarriage. Standard methods were used to calculate probability of miscarriage by
gestational week.[33] In brief, the miscarriage rate during the specific week of gestation was
converted to probability using the formula: (Miscarriage Rate)/(1+ (Miscarriage Rate x 0.5)). The
remaining risk of miscarriage by gestational week was calculated by subtracting the probability of
surviving the remaining weeks from 1. The probability of fetal survival during the remaining weeks is
the product of the probability of survival for week x and the probability of survival for week x+1.
ExposuredefinitionAnimal reprotoxicology studies suggest that the embryo‐sensitive period for artemisinins in humans
would be 6‐12 weeks post‐LMP (4‐10 weeks post conception).[6] A trend of increase in risk during
this artemisinin specific embryo‐sensitive period would corroborate the biological mechanism
observed in animal models and suggest a causal association with ACT exposures. Therefore it is
important to consider and compare the effect of these two exposure groups. The analysis focused
on two exposure definitions: 1) antimalarial drug reported/prescribed in the 1st trimester between
gestational weeks 2 and 13 inclusive (first trimester) and 2) antimalarial drug reported/prescribed
between weeks 6 and 12 post‐LMP (artemisinin embryo‐sensitive period as suggested by animal
reprotoxicology studies [6]). In the absence of a gold‐standard, we used agreement between at least
2 of the 3 data sources to define confirmed exposure. Probable exposures were defined as
antimalarial reported/prescribed according to at least 1 of the 3 data sources in the two periods
defined above. Probable exposures were considered in sensitivity analysis. Unconfirmed exposures
were defined as exposures identified by only 1 of the data sources.
Association between artemisinin exposure in the first trimester and miscarriage
147
CoxregressionmodelCox proportional hazard regression models with left truncation were fitted to estimate the effect on
miscarriage of ACT exposure during the 1st trimester and during the artemisinin embryo‐sensitive
period. Since women were exposed to ACTs at different gestational ages in the 1st trimester, ACT
exposure was treated as a time‐dependent variable. For example, if a woman took an ACT during
gestation week 9, she was considered unexposed during the 1st weeks and then exposed from week
9 onwards.[34] The reference group consisted of pregnancy‐weeks without exposure to
antimalarials or malaria during the 1st trimester according to any of the 3 exposure data sources. The
proportional hazard (PH) assumption was assessed for each covariate using a graphical approach
(the log‐log Kaplan‐Meier survival plots assessing if the curves for different strata are parallel) and
based on Schoenfeld residuals for the global test, and Schoenfeld scaled residuals for separate tests
with each covariate. The cluster function for robust standard errors in Stata was used to account for
the fact that some participants had multiple pregnancies during the study period. Pregnancies that
ended in induced abortions (9), withdrawal (4) or loss to follow up before 28 weeks (20) were
censored at the time of that event, or at the time of their last follow‐up visit in case of loss to follow‐
up.
We considered the following potential confounding factors based on known risk factors for
miscarriage as well as common socioeconomic variables: maternal age at the time of pregnancy
detection [35,36]; gravidity [37]; previous pregnancy loss [38]; socioeconomic characteristics
(education, marital status and occupation [39,40]). Although HIV status [41,42]; syphilis [43] and
anaemia (defined as Hb <11g/dl) could be potential confounders, we could not include these factors
due to missing data. Missing data were particularly frequent for miscarriage cases due to the fact
that most of these pregnancies didn’t persist long enough to attend ANC (mean gestational age for
1st ANC visit 20.8 weeks and mean gestational age for pregnancy loss 14.4 weeks). None reported to
smoke, while alcohol consumption (n=7), diabetes (n=2) and epilepsy (n=5) were infrequent and
these factors were not included in further analysis. Malaria infection was not considered in the
multivariate model since it is strongly correlated with the antimalarial drug exposure of interest. To
determine which variables remained in the final model, assessment of confounding was based on
the impact a variable had on the hazard ratio, followed by the consideration of its precision. If the
HR changed by >=10% the variable was retained in the model. Variables identified as non‐
confounders were dropped from the model provided their deletion led to a gain in precision for the
effect estimate from examining confidence intervals.[44,45]
The primary analysis focused on confirmed ACT exposures in the 1st trimester and in the suggested
artemisinin embryo‐sensitive period (6‐12 weeks post‐LMP, as defined above under Exposure
Definition). In order to assess potential exposure misclassification bias, alternative models were run
using the following exposure definitions:
1. Probable exposure defined as ACT exposure detected from at least one of the 3 data sources
regardless of whether they were confirmed by another data source.
2. ACT exposure excluding those with possible exposures within the error margin of gestation time.
That is, to address the uncertainty related to the assessment of gestational age, we excluded
pregnancies with exposure that could have been before conception or in the 2nd trimester due to
possible error in gestational age assessment (for example: an ACT exposure at 13 weeks
estimated using the Ballard Score method has an estimated error margin of 2 weeks therefore
Chapter 8
148
the true gestational age most probably lies between 11 and 15 weeks. Since the upper margin for
this exposure would fall outside the 1st trimester, such cases were excluded from the exposure
group).
To account for the possibility of confounding by indication (that is, the fact that malaria, for which
ACT was taken, itself may also cause miscarriage), a separate Cox regression model was run
comparing:
3. Pregnancies exposed to quinine (not known to cause miscarriage) in the 1st trimester to those
exposed to ACT in the 1st trimester.
4. Pregnancies with possible malaria infection (including both self‐report and OPD records of
confirmed malaria) in the 1st trimester compared to those without evidence of malaria or
antimalarial treatment. The PBIDS data were not included in this analysis as information on
malaria diagnosis was not captured through PBIDS system which only collected information on
symptoms, health seeking behaviour and treatment
EthicalreviewandconsentThe EMEP study protocol was reviewed and approved by the institutional review boards of CDC (No.
5889), KEMRI (No. 1752) and the Liverpool School of Tropical Medicine (No. 09.70). Written
informed consent or assent was obtained from each participant.
Results
ParticipantcharacteristicsOut of 5911 WOCBA approached, 5536 (94%) consented to participate and 1,453 pregnancies were
detected. Out of these, 1,126 (77.5%) were included in the data analysis; 327 were excluded because
pregnancy detection occurred beyond 28 weeks gestation (233) or at the time of pregnancy
outcome (33), lack of information on gestational age of exposure (15), loss to follow up immediately
after pregnancy detection (41), or inconsistent pregnancy end dates (5). The 1,126 pregnancies
involved a total of 951 women, 86 of whom had 2 pregnancies and 1 who had 3 pregnancies during
the study period. The mean gestational age at time of pregnancy detection was 13.3 weeks (SD=6.9).
Overall, 62% of deliveries took place at a health facility, and 25% of miscarriages took place at a
health facility. 67% of pregnancy outcomes were captured less than 1 week after the end of
pregnancy; however, for miscarriage this proportion was only 20%. The median number of days
between outcome and follow up was 3 overall (range: 0‐755) and for miscarriage was 24 (range: 0‐
602). This reflects the fact that follow ups were arranged at the convenience of participants and to
ensure suitable amount of time between miscarriages and follow up. Table 2 provides a description
of participants overall and by ACT exposure status.
Association between artemisinin exposure in the first trimester and miscarriage
149
Table 11 Characteristics of 1,126 pregnancies by ACT exposure status (n (%) otherwise stated).
Total
(N=1126)
No ACT 1st
trimester (N=830)
Unconfirmed ACT 1st trimester (N=221)
Confirmed ACT 1st trimester
(N=75)
P‐values*
Demographic and Obstetric Characteristics
Gestational week at pregnancy detection (mean (SD; range))**
13.3 (6.9; 0‐27.9)
13.3 (6.9; 0‐27.9)
13.0 (6.8; 0.3‐ 27)
13.4 (6.9; 2.4‐ 27.4)
0.794
Age in years (mean (SD; range)) 26.2 (6.8; 15‐47)
26.1 (6.7; 15‐45)
26.6 (7.2; 15‐47)
25.1 (6.4; 16‐41)
0.261
Age categories 0.743
15‐20 280 (24.9) 202 (24.3) 55 (24.9) 23 (30.7)
21‐25 286 (25.4) 218 (26.3) 49 (22.2) 19 (25.3)
26‐30 252 (22.4) 188 (22.7) 47 (21.3) 17 (22.7)
31‐35 182 (16.2) 130 (15.7) 42 (19) 10 (13.3)
>35 126 (11.2) 92 (11.1) 28 (12.7) 6 (8)
Gravidity Missing n= 17 Missing n= 14 Missing n= 2 Missing n= 1 0.075
Primigravidae 212 (19.6) 145 (17.8) 47 (21.5) 20 (27.0)
1‐3 pregnancies 511 (47.1) 405 (49.6) 90 (41.1) 30 (40.5)
4+ pregnancies 361 (33.3) 266 (32.6) 82 (37.4) 24 (32.4)
Previous pregnancy loss 150 (13,8), Missing n=37
109 (13.7), Missing n=34
30 (13.7), Missing n=2
11 (14.9), Missing n=1
0.961
Education level Missing n= 42 Missing n= 39 Missing n= 2 Missing n= 1 0.505
None/ Primary not completed 478 (44.1) 345 (43.6) 95 (43.4) 38 (51.4)
Primary completed 519 (47.9) 383 (48.4) 103 (47.0) 33 (44.6)
Secondary completed 87 (8.0) 63 (8.0) 21 (9.6) 3 (4.1)
Occupation Missing n= 54 Missing n= 50 Missing n= 2 Missing n= 2 0.191
Not working 361 (33.7) 265 (34.0) 67 (30.6) 23 (39.7)
Farming 361 (33.7) 261 (33.5) 79 (36.1) 21 (28.8)
Small business/ Skilled Labour 331 (30.9) 243 (31.2) 65 (29.7) 23 (31.5)
Other 19 (1.8) 11 (1.4) 8 (3.7) 0
Marital Status Missing n= 42 Missing n= 39 Missing n= 2 Missing n= 1 0.161
Single 227 (20.9) 156 (19.7) 50 (22.8) 21 (28.4)
Married 857 (79.1) 635 (80.3) 169 (77.2) 53 (71.6)
Antenatal Care Summary by the End of Pregnancy
Gestational week at 1st ANC study visit (mean (SD; range)
20.8 (7.8; 1.7‐41.0)
21.31 (7.8; 2.7‐37.1)
19.6 (7.6; 1.7‐41.0)
19.4 (7.8; 3.4‐37.0)
0.014
Number of ANC visit Missing n=39 Missing n=31 Missing n=5 Missing n=3 0.121
none 90 (8.3) 65 (8.1) 21 (9.7) 4 (5.6)
1 88 (8.1) 61 (7.6) 23 (10.7) 4 (5.6)
2 151 (13.9) 119 (14.9) 24 (11.1) 8 (11.1)
3 244 (22.5) 192 (24.0) 39 (18.1) 13 (18.1)
4+ 514 (47.3) 362 (45.3) 109 (50.5) 43 (59.7)
IPTp doses (in HIV negative) Missing n= 279 Missing n= 210 Missing n= 53 Missing n=16 0.579
none 243 (28.7) 184 (29.7) 43 (25.6) 16 (27.1)
1 93 (11.0) 59 (9.5) 27 (16.1) 7 (11.9)
2 170 (20.1) 126 (20.3) 32 (19.1) 12 (30.3)
3 222 (26.1) 163 (26.3) 42 (25.0) 17 (28.8)
4 119 (14.1) 88 (44.2) 24 (14.3) 7 (11.9)
Vaginal Bleeding 20 (2.4),
Missing n= 298 18 (3.0),
Missing n= 238 3 (1.7), Missing
n= 53 2 (2.9), Missing
n=7 0.680
Chapter 8
150
Total (N=1126)
No ACT 1st trimester (N=830)
Unconfirmed ACT 1st trimester (N=221)
Confirmed ACT 1st trimester
(N=75)
P‐values*
ANC Laboratory Profile from 1st visit
HIV positive 244 (24.5) Missing n=129
183 (24.9) Missing n=97
48 (24.7) Missing n=27
13 (18.6) Missing n=5
0.491
Hb (mean (SD; range)) 11.2 (1.9; 4.3‐17.2)
Missing n=309
11.1 (1.9; 4.3‐17.2)
Missing n=247
11.6 (2.0; 5.1‐16.6) Missing
n=55
11.6 (1.9; 6.9‐16.7)
Missing n=7
0.002
Anemia (Hb<11g/dl) 346 (42.4) Missing n=309
263 (45.1) Missing n=247
56 (33.7) Missing n=55
27 (39.7) Missing n=7
0.029
Malaria slide positive 114 (13.9) Missing n=306
72 (12.3) Missing n=244
32 (19.3) Missing n=55
10 (14.7) Missing n=7
0.070
Syphilis reactive test 71 (7.9) Missing n=225
49 (7.5) Missing n=176
17 (9.4) Missing n=41
5 (7.5) Missing n=8
0.684
*P values refer to Pearson Chi‐square test for categorical variables and ANOVA test for continuous variables ** Gestational age lowest estimate include 0 which reflects inaccuracy in the gestational age measurements Acronyms: ACT, artemisinin combination therapy; Hb, haemoglobin; SD, Standard deviation
Prevalenceof1sttrimesterACTexposuresOverall, 296 (26.3%) of the 1,126 pregnancies had evidence of an ACT exposure during the 1st
trimester as determined by at least 1 of the 3 exposure data sources (probable exposure) and 75 of
these (25.3% of exposures and 6.7% of all pregnancies) by at least 2 of the 3 sources (confirmed
exposure). The number reduced to 54 (18.4%, 4.8% of pregnancies) when pregnancies with
exposures within the potential gestational age error margin were excluded. Two‐hundred and one
(17.9%) pregnancies had evidence of probable exposure in the suggested artemisinin embryo‐
sensitive period but only 42 were confirmed exposures. Of these, 21 (2% of pregnancies) were within
the estimated gestational age confidence interval. Table 3 below depicts the different exposure
categories.
Table 12 Breakdown of antimalarial and ACT exposure status in the 1st trimester
Overall n (%)
N=1,126
Within gestational age confidence interval n (%)*
N=1,063
No antimalarial in 1st trimester 812 (72.1) 812 (76.4)
Probable non‐ACT antimalarial in 1st trimester**
18 (1.6) 13 (1.2)
Probable ACT exposure in 1st trimester
296 (26.3) 238 (23.0)
Confirmed ACT exposure in 1st trimester
75 (6.7) 54 (5.1)
Probable ACT exposure in 6‐12 weeks gestational age
201 (17.9) 116 (10.9)
Confirmed ACT exposure in 6‐12 weeks gestational age
42 (3.7) 21 (2.0)
* 63 cases have undefined exposure status as they fall in the potential error margin for gestational age.
** including 12 exposed to quinine, 5 to sulphadoxine‐pyrimethamine and 1 to amodiaquine.
RateofmiscarriagepergestationalweekThere were 89 (7.9%) miscarriages among the 1,126 pregnancies included in the analysis. The mean
gestational age at the time of miscarriage was 14.4 weeks (SD: 5.7) and the median was 13 weeks
(range: 4.3‐28); 75% of miscarriages occurred by 18 weeks. Accounting for late pregnancy detection,
the rate of miscarriage over the 1st 28 weeks of pregnancy calculated by survival analysis with left
truncation was 16.5 per 100 pregnancy‐weeks (95% CI: 13.4‐ 20.3) and the cumulative probability of
Association between artemisinin exposure in the first trimester and miscarriage
151
miscarriage was 19.2% (95% CI: 14.7‐ 24.9). The weekly miscarriage rate declined steadily with
increasing gestation (see Figure 1 and Table s1 in Appendix for miscarriage weekly rates and
probabilities) until approximately 16 to 20 weeks and then remained steady at approximately 3 per
1000 pregnancy‐weeks. Figure 2 shows the cumulative pregnancy survival probabilities per gestation
week.
Figure 1. Miscarriage rate by week of gestation with upper and lower estimates of 95% confidence interval.
Figure 2. Miscarriage Kaplan Meier survival curve by gestational week.
0
5
10
15
20
25
30
35
40
4 6 8 10 12 14 16 18 20 22 24 26
Rate per 1000 pregnan
cy‐w
eek
Gestational Week
Upper CI:117
6361
0.70
0.80
0.90
1
Per
cent
age
pre
gnan
cies
no
t yet
mis
carr
ied
46 297 514 675 809 916 1009 Number at risk
4 8 12 16 20 24 28Gestational Weeks
Chapter 8
152
RiskfactorsformiscarriageFactors associated with miscarriages included maternal age (increasing risk with increasing age);
gravidity (higher risk for primi and gravidity of 4 or more); previous pregnancy loss; occupation (with
highest risk for those doing physical labour/farming); reporting any vaginal bleed during the index
pregnancy; haemoglobin levels (1.2 g/dl higher Hb level observed in cases who miscarried compared
to other pregnancy outcomes which include stillbirths and live‐births); being HIV positive (with over
half of the miscarriage cases being HIV positive). Reported use of traditional remedies during
pregnancy was associated with a lower risk of miscarriage. Gestational age at first study ANC visit,
the number of ANC visits as well as the number of IPTp doses received had a significant protective
effect on miscarriage but this could reflect the fact that the majority of pregnancies ending in
miscarriage did not survive long enough to receive antenatal care since the average gestational age
for starting ANC in the overall cohort was 23 weeks, whereas the mean gestational age at the time of
miscarriage was 14 weeks.
Table 4. Distribution (n (%) or otherwise stated), rates of miscarriage and hazard ratios for covariates of interest.
Miscarriage
(N=89)
Other Pregnancy Outcomes (n=1037)
P‐values*
Rate of Miscarriage per 1000 pregnancy‐
weeks (95%CI)
Hazard Ratio (95%CI) §
Age in years (mean (SD)) 29.5 (7.9) 25.9 (6.6) P<0.001 NA 1.08 (1.04‐ 1.11)
Age categories P<0.001
15‐20 14 (15.7) 266 (25.7) 3.80 (2.29‐ 6.76) Ref
21‐25 14 (15.7) 272 (26.2) 3.50 (2.04‐ 6.51) 0.91 (0.43‐ 1.92)
26‐30 16 (18.0) 236 (22.8) 4.56 (2.83‐ 7.78) 1.15 (0.57‐ 2.31)
31‐35 21 (23.6) 161 (15.5) 8.85 (5.90‐ 13.85) 2.34 (1.22‐ 4.51)
>35 24 (26.8) 102 (9.8) 15.76 (10.45‐ 24.61) 4.14 (2.14‐ 7.98)
Gravidity Missing n= 1 Missing n= 16 P<0.001
Primigravidae 17 (19.3) 195 (19.1) 6.33 (3.98‐ 10.63) Ref
1‐3 pregnancies 23 (26.1) 502 (49.2) 3.12 (2.10‐ 4.83) 0.49 (0.26‐ 0.91)
4+ pregnancies 48 (54.6) 324 (31.7) 10.11 (7.64‐ 13.65) 1.61 (0.94‐ 2.77)
Previous pregnancy loss 19 (23.2),
Missing n= 7 131 (13.0) , Missing n= 30
P=0.004 10.17 (6.57‐ 16.46) 2.11 (1.28‐ 3.49)
Gestational week at detection (mean (SD))
8.0 (4.7) 13.8 (6.9) P=0.058 NA 0.94 (0.88‐ 1.01)
Education level Missing n= 10 Missing n= 32 P=0.541
Primary not completed 32 (40.5) 446 (44.4) 4.90 (3.50‐ 7.07) Ref
Primary completed 42 (53.2) 477 (47.5) 6.16 (4.59‐ 8.45) 1.21 (0.77‐ 1.91)
Secondary completed 5 (6.3) 82 (8.2) 4.16 (1.41‐ 17.36) 0.8 (0.27‐ 2.38)
Occupation Missing n= 11 Missing n= 43 P=0.002
Not working 22 (28.2) 339 (34.1) 4.90 (3.27‐ 7.66) Ref
Farming 33 (42.3) 328 (33.0) 6.35 (4.56‐ 9.10) 1.21 (0.71‐ 2.05)
Small business/Skilled Labour
18 (23.1) 313 (31.5)
4.03 (2.56‐ 6.67) 0.81 (0.44‐ 1.49)
Other 5 (6.4) 14 (1.4) 24.09 (9.43‐ 71.12) 4.54 (1.73‐ 11.91)
Marital Status Missing n= 10 Missing n= 32 P=0.140
Single 21 (26.6) 206 (20.5) 7.27 (4.81‐ 11.49) Ref
Married 58 (73.4) 799 (79.5) 4.97 (3.85‐ 6.54) 0.69 (0.42‐ 1.12)
Association between artemisinin exposure in the first trimester and miscarriage
153
Miscarriage (N=89)
Other Pregnancy Outcomes (n=1037)
P‐values*
Rate of Miscarriage per 1000 pregnancy‐
weeks (95%CI)
Hazard Ratio (95%CI) §
Antenatal Care Summary
Gestational week at 1st ANC study visit (mean (SD))
10.4 (4.9), Missing n=71
21.0 (7.7), Missing n=227
P<0.001 NA 0.85 (0.79‐ 0.91)
Number of ANC visit at end of pregnancy
Missing n= 0 Missing n=39 P<0.001
none 66 (74.2) 24 (2.4) 110 (84.06‐140) Ref
1 18 (20.2) 70 (7.0) 16.69 (10.54‐27.56) 0.17 (0.1‐ 0.3)
2 1 (1.1) 150 (15.0) 0.48 * 0 (0‐ 0.04)
3 3 (3.4) 241 (24.2) 0.92 (0.29‐4.45) 0.01 (0‐ 0.03)
4+ 1 (1.1) 513 (51.4) 0.13 * 0 (0‐ 0.01)
IPTp doses (HIV negative) at end of pregnancy
Missing n= 15 Missing n= 264 P<0.001
none 73 (98.7) 171 (22.0) 24.41 (19.27‐ 31.16) Ref
1 1 (1.4) 92 (11.9) 0.86* 0.04 (0.01‐ 0.32)
2 0 170 (22.0) 0 0 (0‐ 0)
3 0 222 (28.7) 0 0 (0‐ 0)
4 0 119 (15.4) 0 0 (0‐ 0)
Vaginal Bleeding 4 (22.2),
Missing n= 71 19 (2.4),
Missing n= 227 P<0.001 14.03(5.30‐46.92) 11.45 (3.96‐ 33.12)
ANC Laboratory Profile at 1st visit
BMI Missing n=73 Missing n=242 P=0.205
<18 3 (18.8) 47 (5.9) 3.74 (1.19‐17.48) Ref
18‐25 12 (75.0) 616 (77.5) 1.37 (0.80‐2.59) 0.4 (0.11‐ 1.4)
>25 1 (6.3) 132 (16.6) 0.60* 0.19 (0.02‐ 1.72)
HIV positive 15 (53.6),
Missing n=61 229 (23.6), Missing n=68
P<0.001 4.64 (2.85‐ 8.04) 3.83 (1.84‐ 7.96)
Hb (mean (sd)) 12.4 (1.9)
Missing n=72 11.2 (1.9)
Missing n=237 P=0.025 NA 1.31 (1.05‐ 1.63)
Anemia (Hb<11g/dl) 4 (23.5)
Missing n=72 342 (42.8)
Missing n=237 P=0.177 0.88 (0.33‐3.15) 0.47 (0.15‐ 1.46)
Malaria slide positive 2 (11.1)
Missing n=71 112 (14.0)
Missing n=235 P=0.737 1.25 (0.27‐12.32) 0.78 (0.18‐ 3.42)
Syphilis reactive test 1 (4.8)
Missing n=68 70 (8.0)
Missing n=157 P=0.635 1.05 * 0.62 (0.08‐ 4.51)
Traditional Remedies 6 (6.7),
Missing n=0 187 (18.3), Missing n=16
P=0.011 2.33 (1.06‐ 6.08) 0.36 (0.16‐ 0.81)
*P value refers to log‐rank test for categorical variables and to univariate Cox proportional hazard regression for continuous variables. **Confidence intervals are missing because of an insufficient number of failures.
Chapter 8
154
Associationbetweenmiscarriageand1sttrimesterexposuretoACTsorquinineNo difference in the risk of miscarriage was apparent when confirmed ACT exposures were
considered only (adjusted HRs for ACT exposures in 1st trimester 1.60 95% CI [0.70‐3.68] and
exposure in the embryo‐sensitive period 0.85 95% CI [0.22‐3.33]) compared to pregnancies without
evidence of malaria or antimalarial exposure in the 1st trimester. The risk of miscarriages was higher
among the pregnancies with probable exposure to ACTs in the 1st trimester (29/296, 7.8/1000
pregnancy‐weeks) compared to unexposed pregnancies (57/790, 5.2/1000 pregnancy‐weeks) by
crude (HR 1.68, 95% CI [1.07‐2.64]) and adjusted analysis (HR 1.79, 95% CI [1.10‐2.91]). Similar
findings were evident for probable ACT exposures restricted to the embryo‐sensitive period
although these were not statistically significant at the 5% level (table 5).
There was no statistically significant difference between pregnancies exposed to quinine (excluding
those with concomitant ACT exposure) during the 1st trimester compared to pregnancies not
exposed to antimalarials (crude HR 1.53 95% CI [0.19‐12.40] and adjusted 2.64 95% CI [0.32‐21.90]).
For comparison purposes, exposures during the suggested artemisinin embryo‐sensitive period (6‐12
weeks) are presented. Although the crude and adjusted HRs are slightly higher than for the whole 1st
trimester, these are not statistically significant. Comparison of pregnancies exposed to quinine in the
1st trimester to those with a confirmed exposure to ACT in the same period (HR=1.38 95%CI [0.19‐
9.95]) and for the embryo sensitive period (HR=0.79 95%CI [0.10‐ 6.54]) showed no statistical
difference although the number in the quinine comparison group were very small (n= 12 and 8
respectively).
The sensitivity analysis assessing the effect of using conservative estimates of gestational age
measurement for determination of 1st trimester showed no evidence of increase in the effect
estimates compared to exposures including those in the possible error margin for 1st trimester
gestational age measurement (crude HR 0.85 95%CI [0.28‐ 2.56] and adjusted HR 1.01 95%CI [0.32‐
3.15]).
Our study did not allow for an analysis of the risk of miscarriage by confirmed vs. unconfirmed
malaria as the use of diagnostic tests, or their results, were not systematically available. However,
pregnancies that reported malaria disease had a two‐fold higher hazard of miscarriage than
pregnancies without evidence of malaria disease or use of antimalarials, both in the crude and
adjusted models (HRs 2.06 95%CI [1.24‐ 3.41] and 2.32 95%CI [1.36‐ 3.94] respectively, table 6).
Asso
ciation betw
een artem
isinin exp
osure in
the first trim
ester and m
iscarriage
155
Table 5. M
iscarriage rate
, unad
juste
d an
d ad
juste
d hazard
rates fo
r the asso
ciation betw
een diffe
rent an
timalarial e
xposure cate
gorie
s and m
iscarriage.
a Th
e unexp
osed
pregn
ancies w
ere defin
ed as th
ose with
out h
istory o
f malaria d
isease or an
timalarial u
se in
the 1
st trimester acco
rding to
any o
f the 3 data so
urce. N
ote th
at the numerato
r and den
ominato
r for th
is unexp
osed
group re
main
the sam
e for all co
mpariso
ns, ye
t the rate
of m
iscarriage dep
icted in
the next co
lumn varies sligh
tly betw
een ro
ws. Th
is is becau
se antim
alarial exposure was fitted
as a time‐d
epen
dant variab
le an
d th
erefore th
e unexp
osed
group co
nsists o
f a combinatio
n of p
regnancies n
ever exp
osed
(e.g. 7
90) an
d of p
regnancies th
at were
eventually exp
osed
, but co
ntrib
uted
‘unexp
osed
pregn
ancy‐w
eeks’ to th
e unexp
osed
category u
ntil th
e day o
f exposure, after w
hich
they sw
itched
to th
e exp
osed
category.
b Covariates in
clude m
aternal age an
d occu
patio
n; c Exclu
ding th
ose also
exposed
to ACT in
1st trim
ester; d Confid
ence intervals co
uld not b
e calcu
lated becau
se of an
insufficien
t number o
f even
ts in th
e dataset; e Exp
osures are treated
here as tim
e‐indep
enden
t (a pregn
ancy is co
nsid
ered exp
osed fo
r the whole perio
d under o
f observatio
n disregard
ing th
e tim
e of exp
osure) an
d
the ACT gro
up exclu
des an
y quinine exp
osures. A
cronym
s: CI, C
onfid
ence Interval; A
CT, A
rtemisin
in Combinatio
n Th
erapy.
Exp
ose
d
Misca
rriage
ra
te p
er 1000
pre
gn
an
cy-we
eks
(95%
CI)
Une
xpo
sed a
Misca
rriage
rate
per 1
000
p
regn
ancy-w
eeks
(95
%C
I)
Exp
ose
d
#M
iscarria
ge/
#O
verall
Un
exp
ose
d a #M
iscarriag
e/
#O
verall
Un
ad
juste
d
Ha
zard
Ra
tio (9
5%
CI)
Adju
sted b
Ha
zard
R
atio
(95
% C
I) P
-valu
eP
-valu
e Antim
alarial Exposure Categories
Quinine c
1st trim
este
r (prob
able
)
1st trim
este
r (confirm
ed
)
6-1
2 w
eeks g
esta
tion (p
rob
able
)
6-1
2 w
eeks g
estatio
n (co
nfirm
ed)
1/ 1
2 0
/ 3
1/ 8
0
/ 1
57/7
90
5.3
7 (4
.13
, 7.1
0)
5.3
8 (4
.14
, 7.1
2)
5.3
7 (4
.13
, 7.1
1)
1.5
3 (0
.19
, 12
.40)
0.6
91
2.6
4 (0
.32
,21
.90)
0.3
68
6.3
1 d
0.0
0
57/7
90
57/7
90
57/7
90
2.0
2 (0
.24
, 17
.10)
3.8
7 (0
.44
, 33
.76)
0.5
20
0.2
21
9.5
6 d
0.0
0
5.3
8 (4
.14
, 7.1
2)
AC
T
1st trim
este
r (prob
able
)
1st trim
este
r (confirm
ed
)
1st trim
este
r (confirm
ed
& g
esta
tion C
I restricted
)
6-1
2 w
eeks g
esta
tion (p
rob
able
)
6-1
2 w
eeks g
estatio
n (co
nfirm
ed)
6-1
2 w
eeks g
esta
tion (co
nfirm
ed &
gesta
tion C
I restricte
d)
29/ 2
96
6/ 7
5
3/ 5
4
7.7
8 (5
.45
, 11
.47)
6.2
(2.8
6, 1
5.9
3)
57/7
90
57/7
90
57/7
90
57/7
90
57/7
90
5.2
4 (4
.03
, 6.9
4)
5.3
6 (4
.12
, 7.0
9)
1.6
8 (1
.07
, 2.6
4)
1.2
4 (0
.56
, 2.7
3)
0.8
5 (0
.28
, 2.5
6)
1.5
5 (0
.91
, 2.6
3)
0.8
0 (0
.21
, 3.0
8)
0.0
25
1.7
9 (1
.10
, 2.9
1)
1.6
0 (0
.7, 3
.68)
1.0
1 (0
.32
, 3.1
5)
1.6
3 (0
.93
, 2.8
5)
0.0
18
0.6
01
0.7
72
0.1
07
0.7
48
0.2
67
0
.985
0
.086
0
.818
4.1
6 (1
.34
, 19
.31)
7.2
4 (4
.62
, 11
.94)
3.7
5 (0
.84
, 34
.46)
0.0
0
5.3
6 (4
.13
, 7.1
0)
5.2
8 (4
.06
, 6.9
8)
5.3
6 (4
.13
, 7.1
0)
5.3
8 (4
.14
, 7.1
2)
18/ 2
01
2/ 4
2
0/ 2
1
0.8
5 (0
.22
, 3.3
3)
57/7
90
ACT vs. Quinine e
1st trim
este
r (prob
able
)
1st trim
este
r (confirm
ed
)
1st trim
este
r (confirm
ed
& g
esta
tion C
I restricted
)
6-1
2 w
eeks g
esta
tion (p
rob
able
)
6-1
2 w
eeks g
estatio
n (co
nfirm
ed)
6-1
2 w
eeks g
esta
tion (co
nfirm
ed &
gesta
tion C
I restricte
d)
27/ 2
74
5/ 6
8
2/ 4
9
16/ 1
81
1/ 3
6
7.2
3 (5
.01
, 10
.78)
5.4
6 (2
.33
, 15
.79)
2.8
9 (0
.65
, 26
.77)
6.5
4 (4
.08
, 11
.11)
1/ 1
2 1
/ 12
1.3
8 (0
.19
, 9.9
5)
1.0
2 (0
.12
, 8.5
4)
0.7
48
0.9
83
0.5
87
1.3
4 (0
.22
, 8.0
8)
1.3
6 (0
.18
, 10
.31)
0.7
49
0
.768
0
.521
5.3
6 d
5.3
6 d
5.3
6 d
8.4
3 d
8.4
3 d
8.4
3 d
1/ 1
2 1
/ 8
1/ 8
1
/ 8
0.5
4 (0
.06
, 4.9
8)
0.7
9 (0
.10
, 6.5
4)
0.2
5 (0
.02
, 2.8
2)
0.5
4 (0
.08
, 3.5
2)
0.7
2 (0
.10
, 5.1
7)
0.8
27
0.2
61
0.7
47
2.0
2 d
0.0
0
0/ 2
0
.05
15
Chapter 8
156
Table 6. Hazard ratio for the association between malaria in the 1st trimester and miscarriage.
#Miscarriages/ #Overall
Miscarriage rate per 1000
pregnancy‐week (95%CI)
Crude HR (95%CI) Adjusted§ HR (95%CI)
No antimalarial or reported malaria in 1st trimester
57/790 5.33 (4.10‐ 7.05) Ref Ref
Malaria in 1st trimester* 23/197 9.71 (6.48‐ 15.12) 2.04 (1.25‐ 3.35) 2.28 (1.36‐ 3.84)
§ Covariates include maternal age and occupation. * Malaria exposure in 1st trimester was derived from self report through interviews at ANC and pregnancy outcome study visits and from Lwak outpatient records on malaria diagnosis based on slide microscopy confirmation.
DiscussionThis study found that there was no statistically significant association between the risk of
miscarriage and confirmed ACT exposures in the 1st trimester of pregnancy or exposures in the
suggested artemisinin embryo‐sensitive period. Miscarriage is a relatively frequent endpoint and
although the sample size was limited to approximately 1000 pregnancies, our study had the power
to exclude a 3.7 fold or greater increased risk of miscarriage associated with artemisinin use in the
1st trimester as indicated by the upper bound of the 95% confidence interval (table 5). Sensitivity
analysis on the effect of using a less stringent definition of ACT exposure (i.e. probable exposure,
defined as exposure in at least one data source) in the first trimester showed a statistically
significant increase in the hazard of miscarriages of 1.8 compared to women who did not receive
antimalarials.
Several factors indicate that this increase in risk of miscarriage observed in the ACT exposed women
may be explained by the underlying malaria infection (i.e. potential confounding by indication).
Firstly, the ACT embryotoxic effect is expected to occur within 6‐12 weeks gestation when primitive
fetal erythroblasts, the target for ACT embryotoxicity, are in circulation.[2,6,46,47] The artemisinins
have very short half‐lives and are eliminated within hours. Therefore if ACTs were causing
miscarriage, the effect size would be expected to be highest for exposures restricted to that embryo‐
sensitive period compared to effect size of exposures encompassing the whole 1st trimester. We
observed no such trend. Secondly, although the effect estimate of quinine should be interpreted
with caution due to the small numbers of quinine‐only exposed women (12), the rates of miscarriage
in the quinine‐only and ACTs exposed pregnancies were similar. For example during the suggested
embryo‐sensitive period for the artemisinins the miscarriage rate was 7.2/1000 pregnancy weeks for
ACTs compared to 9.6/1000 pregnancy weeks for quinine. This is supported by the findings from
McGready et al in Thailand, indicating a higher risk of miscarriage associated with malaria in the 1st
trimester.[11] In this area of low malaria transmission, they found that asymptomatic malaria in 1st
trimester increased the odds of miscarriage by nearly 3 fold and symptomatic infections by 4 fold.
They found no difference in the proportions of pregnancies ending in miscarriages between women
treated with chloroquine (26%), quinine (27%) or artesunate (31%).
The small number of quinine exposures in the first trimester was surprising, but consistent with a
recent study on malaria in pregnancy treatment guidelines knowledge and prescribing practice
carried out in the same area of western Kenya (Riley et al, unpublished, chapter 7) and a study from
Jinja, Uganda.[48] The Kenya study found low adherence to treatment guidelines for women in their
1st trimester regardless of the provider’s malaria treatment guidelines knowledge (Riley et al,
Association between artemisinin exposure in the first trimester and miscarriage
157
unpublished). Very few women in their 1st trimester who had uncomplicated malaria (0% in drug
outlets and 68% in health facilities) were prescribed oral quinine (none as a result of stock out)
which draws attention to the need to assess reasons for poor adherence with quinine prescriptions
and malaria treatment guidelines. Quinine poor tolerability and poor compliance to its 7 days
regimen are known problems. [49]
This study provides the first description of the miscarriage rate in this rural Kenyan population in the
context of high malaria and HIV prevalence; there is very little data on miscarriage background rate
for sub‐Saharan Africa in general. The cumulative probability of miscarriages by 28 weeks gestation
accounting for staggered pregnancy detections in our study population was 19.2%, and the
probability by week declined from 16 weeks onward. The true rate is likely to be higher as
information from very early pregnancies (e.g. <6 weeks gestation) was not captured and the average
gestational age of pregnancy detection was 13.3 weeks, which meant that only 57% of pregnancies
were detected during the highest risk period for miscarriage (the first 12 weeks). However, the rate
of 19.2% is similar to that reported by McGready et al from the Thai‐Burmese border (20%)[11] and
consistent with that observed in other prospective studies in non malarious areas, which ranges
from 10% to 22%. [33,50,51,52]
This study had several limitations that should be considered. First, because only few women were
treated with quinine (the recommended 1st line malaria treatment in the 1st trimester) our ability to
compare ACT exposed pregnancies to a purportedly ‘control’ drug not known to be linked to
miscarriages was limited. Second, we could not control for confounding by indication (i.e. the
disease itself) because malaria was partly self‐reported (including self‐diagnosis, clinical diagnosis
and laboratory confirmed diagnosis) and partly laboratory confirmed. Controlling for laboratory
confirmed malaria diagnosis is important, as malaria itself has been suggested to reduce the
potential risk of embryotoxicity with artemisinin as was found in rat models. It was recently
suggested that in human pregnancies malaria may protect against artemisinin‐induced decreases in
reticulocyte count (a marker for embryotoxicity) by reducing the tissue levels of active drug and/or
ferrous iron which activates the drug to toxic free radicals, so that pregnant women without malaria
would be at greater risk of artemisinin‐induced embryotoxicity.[53] Third, there could have been
misclassification of miscarriage as an outcome since induced abortions are illegal in Kenya, they can
result in differential loss to follow‐up as well as misclassification of induced abortions as
miscarriages. However since inadvertent ACT exposure is very common and not perceived as an
indication for induced abortion, it is unlikely that such misclassification would differ according to
exposure status and therefore affect the ACT exposure effect estimate. Lastly, we found little
agreement between data sources regarding ACT exposures. Partly this is due to the fact that by
design, the PBIDS and Lwak OPD datasets could not capture all exposures. The Lwak OPD data
source was limited to participants receiving treatment from this facility; only 68 of the 169 (40%)
EMEP ACT exposures in the 1st trimester were reported to be from Lwak health facility and 48% of
cases reporting taking an antimalarial at a PBIDS home visit also sought care from Lwak health
facility. Further the antimalarial drug intake was not observed in most instances and it is unknown
what proportion of women complied with the prescribed regimen. The limitation of the PBDS (bi‐
)weekly home visits is in part missingness of data when no one was at home and the use of proxy
reporting where the participants herself was not at home on the day of interview and information
was provided by a relative responding on her behalf. The EMEP data source is limited by the
Chapter 8
158
participant recall, however pictorial aids and calendars were used to facilitate better recall. This data
source was also more prone to interviewer bias and reporter desirability bias.[54] Nevertheless the
group of ‘confirmed’ exposures is less likely to be misclassified as this was defined as evidence for
exposure in at least 2 of the 3 exposure data sources. Furthermore, the unexposed group was also
strictly defined as it excluded all possible antimalarial treatment according to any of the 3 data
sources. Another potential source of exposure misclassification is gestational age measurement
error. Our sensitivity analysis restricting the definition of ACT exposures to women whose exposure
fell within the potential error margins for gestational age measurement suggested this was not an
effect modifier. Also potential bias would only have been introduced if the error in gestational age
differed by exposure status.[55] We cannot account for exposure misclassification due to counterfeit
antimalarials, which is getting more common in sub‐Saharan Africa.[56]
Our study shows some of the challenges to conduct drug safety studies in resource poor settings. It
is very difficult to obtain reliable antimalarial exposure data outside a controlled environment such
as a clinical trial, as seen by the low level of agreement between the 3 data sources used for drug
ascertainment. In areas of high malaria transmission, having a febrile episode is not a particularly
memorable event and thus recalling the date is difficult. Furthermore, in rural settings where
illiteracy is common (estimated at 15% in Nyanza province [57]), reliable reports of drug name and
date are difficult to obtain. Antimalarials are widely available over‐the‐counter, and data from health
facility or clinic books do not capture history of drug intake from other sources or over‐the‐counter
exposures. Nevertheless, observational studies are essential to monitor drug safety in the post‐
marketing phase particularly for pregnant women and are often the only source of information
when administering a drug through a controlled clinical trial is considered unethical, or when the
outcome is rare or can occur a long time after the exposure. Such studies should be conducted in
sites where drug exposures can be confirmed, such as presented here, and be captured as
comprehensively as possible.
ConclusionThis prospective cohort study in women of childbearing age provides one of the first estimates of
weekly miscarriage rates in a rural African setting in the context of high malaria and HIV prevalence.
This information should be valuable to researchers and program managers for resource planning, to
monitor trends and impact of interventions as well as to clinicians in gauging miscarriage rate at a
given gestational week. Additionally, it adds to the limited information available on the use of
artemisinins in the 1st trimester of pregnancy. Our results are consistent with two previous
observational studies showing that the rate of miscarriage for pregnancies exposed to ACTs is similar
[11] or lower compared to quinine [12] and as such provides further reassurance regarding their
inadvertent or advertent use in early pregnancy for the treatment of acute malaria. Lastly, our
results suggest that ACT use in the 1st trimester is much more common than oral quinine, consistent
with our survey in clinics and drug outlets (Riley et al, unpublished). The lack of observed risk
associated with ACTs use in early pregnancy to date, and the limited compliance to treatment with
quinine suggests a trial comparing ACT vs. oral quinine for the treatment of uncomplicated malaria
in the 1st trimester may be merited. Before such a trial is considered, further safety data on the
association between ACT and congenital malformations that is forthcoming from studies conducted
by the Malaria in Pregnancy Consortium and WHO should be reviewed. An updated review to assess
the pooled evidence of the risk and benefits of artemisinin use in pregnancy is needed to potentially
Association between artemisinin exposure in the first trimester and miscarriage
159
revise the recommendations for the treatment of malaria in pregnancy by WHO. To prevent
exposure to malaria or antimalarial drug in the 1st trimester of pregnancy, malaria prevention
strategies should target women as early as possible in pregnancy.
AcknowledgementsThis research was conducted through the ongoing KEMRI and CDC collaboration. We are very
grateful to all participants and their babies for taking part in the study. We wish to thank the EMEP
study team for their perseverance and hard work. We are grateful to the International Emerging
Infection Program (IEIP) team for their help and collaboration. Furthermore we wish to thank the
Asembo District health and medical team and the Lwak Mission Hospital Board for their support. We
also wish to thank John Williamson and Jane Bruce for the statistical support and advice. KEMRI/CDC
HDSS is a member of the INDEPTH Network. The findings and conclusions in this paper are those of
the authors and do not necessarily represent the views of the US Centers for Disease Control and
Prevention. This paper is published with the permission of KEMRI Director.
References1. WHO (2013) World Malaria Report: 2013. Geneva: World Health Organisation. 2. Clark R, Kumemura M, Makori N, Nakata Y, Bernard F, et al. (2006) Artesunate: Developmental toxicity in
monkeys. . Bir Birth Defects Research (Part A) 76: 329. 3. Clark RL (2009) Embryotoxicity of the artemisinin antimalarials and potential consequences for use in
women in the first trimester. Reprod Toxicol 28: 285‐296. 4. WHO, TDR (2007) Assessment of the safety of artemisinin compounds in pregnancy: report of two joint
informal consultations convened in 2006. Geneva: WHO. WHO/GMP/TDR/Artemisinin/07.1. WHO/GMP/TDR/Artemisinin/07.1.
5. Kelemen E, Calvo W, Fliedner T (1979) Atlas of Human Hemopoietic Development; Springer‐Verlag, editor. Berlin.
6. White TE, Clark RL (2008) Sensitive periods for developmental toxicity of orally administered artesunate in the rat. Birth Defects Res B Dev Reprod Toxicol 83: 407‐417.
7. Adam I, Elhassan EM, Omer EM, Abdulla MA, Mahgoub HM, et al. (2009) Safety of artemisinins during early pregnancy, assessed in 62 Sudanese women. Ann Trop Med Parasitol 103: 205‐210.
8. Deen JL, von Seidlein L, Pinder M, Walraven GE, Greenwood BM (2001) The safety of the combination artesunate and pyrimethamine‐sulfadoxine given during pregnancy. Trans R Soc Trop Med Hyg 95: 424‐428.
9. Dellicour S, Hall S, Chandramohan D, Greenwood B (2007) The safety of artemisinins during pregnancy: a pressing question. Malar J 6: 15.
10. Manyando C, Mkandawire R, Puma L, Sinkala M, Mpabalwani E, et al. (2010) Safety of artemether‐lumefantrine in pregnant women with malaria: results of a prospective cohort study in Zambia. Malar J 9: 249.
11. McGready R, Lee SJ, Wiladphaingern J, Ashley EA, Rijken MJ, et al. (2012) Adverse effects of falciparum and vivax malaria and the safety of antimalarial treatment in early pregnancy: a population‐based study. Lancet Infect Dis 12: 388‐396.
12. Mosha D, Mazuguni F, Mrema S, Sevene E, Abdulla S, et al. (2014) Safety of artemether‐lumefantrine exposure in first trimester of pregnancy: an observational cohort. Malar J 13: 197.
13. Rulisa S, Kaligirwa N, Agaba S, Karema C, Mens PF, et al. (2012) Pharmacovigilance of artemether‐lumefantrine in pregnant women followed until delivery in Rwanda. Malar J 11: 225.
14. WHO (2003) Assessment of the safety of artemisinin compounds in pregnancy. Report of two informal consultations convened by WHO in 2002. Geneva: WHO.
15. Dellicour S, ter Kuile FO, Stergachis A (2008) Pregnancy exposure registries for assessing antimalarial drug safety in pregnancy in malaria‐endemic countries. PLoS Med 5: e187.
16. Desai M, ter Kuile FO, Nosten F, McGready R, Asamoa K, et al. (2007) Epidemiology and burden of malaria in pregnancy. Lancet Infect Dis 7: 93‐104.
Chapter 8
160
17. Huynh BT, Fievet N, Gbaguidi G, Dechavanne S, Borgella S, et al. (2011) Influence of the timing of malaria infection during pregnancy on birth weight and on maternal anemia in Benin. Am J Trop Med Hyg 85: 214‐220.
18. Cottrell G, Mary JY, Barro D, Cot M (2007) The importance of the period of malarial infection during pregnancy on birth weight in tropical Africa. Am J Trop Med Hyg 76: 849‐854.
19. Walker PG, Ter Kuile FO, Garske T, Menendez C, Ghani AC (2014) Estimated risk of placental infection and low birthweight attributable to Plasmodium falciparum malaria in Africa in 2010: a modelling study. Lancet Glob Health 2: e460‐e467.
20. Brabin BJ, Romagosa C, Abdelgalil S, Menendez C, Verhoeff FH, et al. (2004) The sick placenta‐the role of malaria. Placenta 25: 359‐378.
21. Desai M, Dellicour S (2012) Effects of malaria and its treatment in early pregnancy. Lancet Infect Dis 12: 359‐360.
22. Bigogo G, Audi A, Aura B, Aol G, Breiman RF, et al. (2010) Health‐seeking patterns among participants of population‐based morbidity surveillance in rural western Kenya: implications for calculating disease rates. Int J Infect Dis 14: e967‐973.
23. Feikin DR, Audi A, Olack B, Bigogo GM, Polyak C, et al. (2010) Evaluation of the optimal recall period for disease symptoms in home‐based morbidity surveillance in rural and urban Kenya. Int J Epidemiol 39: 450‐458.
24. Adazu K, Lindblade KA, Rosen DH, Odhiambo F, Ofware P, et al. (2005) Health and demographic surveillance in rural western Kenya: a platform for evaluating interventions to reduce morbidity and mortality from infectious diseases. Am J Trop Med Hyg 73: 1151‐1158.
25. Odhiambo FO, Laserson KF, Sewe M, Hamel MJ, Feikin DR, et al. (2012) Profile: the KEMRI/CDC Health and Demographic Surveillance System‐‐Western Kenya. Int J Epidemiol 41: 977‐987.
26. Sasidharan K, Dutta S, Narang A (2009) Validity of New Ballard Score until 7th day of postnatal life in moderately preterm neonates. Arch Dis Child Fetal Neonatal Ed 94: F39‐44.
27. WHO (1970) Spontaneous and Induced Abortion: Report of a WHO scientific group. Geneva. 28. Mitchell AA, Cottler LB, Shapiro S (1986) Effect of questionnaire design on recall of drug exposure in
pregnancy. Am J Epidemiol 123: 670‐676. 29. de Jong‐van den Berg LT, Waardenburg CM, Haaijer‐Ruskamp FM, Dukes MN, Wesseling H (1993) Drug use
in pregnancy: a comparative appraisal of data collecting methods. Eur J Clin Pharmacol 45: 9‐14. 30. Gama H, Correia S, Lunet N (2009) Questionnaire design and the recall of pharmacological treatments: a
systematic review. Pharmacoepidemiol Drug Saf 18: 175‐187. 31. Howards PP, Hertz‐Picciotto I, Poole C (2007) Conditions for bias from differential left truncation. Am J
Epidemiol 165: 444‐452. 32. Meister R, Schaefer C (2008) Statistical methods for estimating the probability of spontaneous abortion in
observational studies‐‐analyzing pregnancies exposed to coumarin derivatives. Reprod Toxicol 26: 31‐35. 33. Avalos LA, Galindo C, Li D‐K (2012) A Systematic Review to Calculate Background Miscarriage Rates using
Life Table Analysis. Birth Defects Research (Part A) 94. 34. Xu R, Luo Y, Chambers C (2012) Assessing the effect of vaccine on spontaneous abortion using time‐
dependent covariates Cox models. Pharmacoepidemiol Drug Saf 21: 844‐850. 35. Nybo Andersen AM, Wohlfahrt J, Christens P, Olsen J, Melbye M (2000) Maternal age and fetal loss:
population based register linkage study. BMJ 320: 1708‐1712. 36. Smith KE, Buyalos RP (1996) The profound impact of patient age on pregnancy outcome after early
detection of fetal cardiac activity. Fertil Steril 65: 35‐40. 37. Kline J (1978) I. An epidemiological review of the role of gravidity in spontaneous abortion. Early Hum Dev
1: 337‐344. 38. Regan L, Braude PR, Trembath PL (1989) Influence of past reproductive performance on risk of
spontaneous abortion. BMJ 299: 541‐545. 39. Banerjee B, Dey TK, Chatterjee P (2005) Work related physical exertion and spontaneous abortion. Indian J
Public Health 49: 248‐249. 40. Kline JK (1986) Maternal occupation: effects on spontaneous abortions and malformations. Occup Med 1:
381‐403. 41. D'Ubaldo C, Pezzotti P, Rezza G, Branca M, Ippolito G (1998) Association between HIV‐1 infection and
miscarriage: a retrospective study. DIANAIDS Collaborative Study Group. Diagnosi Iniziale Anomalie Neoplastiche AIDS. AIDS 12: 1087‐1093.
42. Brocklehurst P, French R (1998) The association between maternal HIV infection and perinatal outcome: a systematic review of the literature and meta‐analysis. Br J Obstet Gynaecol 105: 836‐848.
Association between artemisinin exposure in the first trimester and miscarriage
161
43. Holder WR, Knox JM (1972) Syphilis in pregnancy. Med Clin North Am 56: 1151‐1160. 44. Sonis J (1998) A closer look at confounding. Fam Med 30: 584‐588. 45. Kleinbaum DG, Klein M (2010) Chapter 6: Modeling Strategy Guidelines. Logistic Regression: A Self‐Learning
Text. 3rd edition ed: Springer. 46. Longo M, Zanoncelli S, Manera D, Brughera M, Colombo P, et al. (2006) Effects of the antimalarial drug
dihydroartemisinin (DHA) on rat embryos in vitro. Reprod Toxicol 21: 83‐93. 47. Longo M, Zanoncelli S, Torre PD, Riflettuto M, Cocco F, et al. (2006) In vivo and in vitro investigations of the
effects of the antimalarial drug dihydroartemisinin (DHA) on rat embryos. Reprod Toxicol 22: 797‐810. 48. Sangaré L, Weiss N, Brentlinger P, Richardson B, Staedke S, et al. (2011) Patterns of antimalarial drug
treatment among pregnant women in Uganda. Malaria Journal 10: 152. 49. Achan J, Talisuna AO, Erhart A, Yeka A, Tibenderana JK, et al. (2011) Quinine, an old anti‐malarial drug in a
modern world: role in the treatment of malaria. Malar J 10: 144. 50. Ellish NJ, Saboda K, O'Connor J, Nasca PC, Stanek EJ, et al. (1996) A prospective study of early pregnancy
loss. Hum Reprod 11: 406‐412. 51. Wang X, Chen C, Wang L, Chen D, Guang W, et al. (2003) Conception, early pregnancy loss, and time to
clinical pregnancy: a population‐based prospective study. Fertil Steril 79: 577‐584. 52. Wilcox AJ, Weinberg CR, O'Connor JF, Baird DD, Schlatterer JP, et al. (1988) Incidence of early loss of
pregnancy. N Engl J Med 319: 189‐194. 53. Clark RL (2012) Effects of artemisinins on reticulocyte count and relationship to possible embryotoxicity in
confirmed and unconfirmed malarial patients. Birth Defects Res A Clin Mol Teratol 94: 61‐75. 54. Delgado‐Rodriguez M, Llorca J (2004) Bias. J Epidemiol Community Health 58: 635‐641. 55. Howards PP, Hertz‐Picciotto I, Weinberg CR, Poole C (2006) Misclassification of gestational age in the study
of spontaneous abortion. Am J Epidemiol 164: 1126‐1136. 56. Newton PN, Green MD, Mildenhall DC, Plancon A, Nettey H, et al. (2011) Poor quality vital anti‐malarials in
Africa ‐ an urgent neglected public health priority. Malar J 10: 352. 57. Kenya National Bureau of Statistics (2013) Nyanza Province Multiple Indicator Cluster Survey 2011, Final
Report. Nairobi.
162
Chapter8Appendix&SupportingInformation
Text S1. Detailed description of study site and methodology.
Table S1. Weekly miscarriage rate, cumulative probabilities of survival and remaining risk of
miscarriage by gestational week.
Association between artemisinin exposure in the first trimester and miscarriage
163
TextS1:Detaileddescriptionofstudysiteandmethodology.
ThestudysiteandpopulationThe study area was located in Rarieda District in Siaya County, lying northeast of Lake Victoria in Nyanza Province, western Kenya. The cohort was carried out in 33 adjacent villages in the Asembo area (figure 1). The study area comprises one mission hospital, 2 government‐run health centres and 4 dispensaries. The provincial teaching hospital is located in Kisumu town approximately 50 km away and the closest district hospital is Bondo District Hospital 13km away. This area is under continuous health and demographic surveillance (HDSS) by Kenya Medical Research Institute (KEMRI) and the Centers for Disease Control and Prevention (CDC) Research and Public Health Collaboration since 2001.[1,2] Through quarterly household visits, information on births, deaths, and migrations are collected. Cause of death is established through verbal autopsy. In addition to the HDSS, the 33 villages are under enhanced morbidity surveillance since 2005 through the KEMRI/CDC International Emerging Infection program (IEIP) to investigate major infectious disease syndromes. This program has been described in details elsewhere.[3] As part of this population‐based infectious disease surveillance project (PBIDS), trained fieldworkers visit households regularly (on a weekly basis from January 5, 2010 to May 26, 2011 and then bi‐weekly basis from May 27, 2011 onwards) collecting information on all symptoms since the previous visit, the source of care and any medication taken, including specific antimalarial medication. In addition all visits for infectious symptoms Lwak Mission Hospital, the referral health facility for PBIDS, are recorded (including diagnosis made and prescriptions given) using the HDSS ID for record linkage. All care for acute illness is free at this health facility for PBIDS study participants, thereby enhancing health utilization. The morbidity data is linked to socio‐economic and demographic data collected by the HDSS. The HDSS and IEIP platform provided a unique opportunity to monitor drug use in pregnancy and develop/validate methodology for pharmacovigilance throughout the pregnancy period.
Rainfall is bimodal with long rains between March‐June and short rains in October‐December. Average temperature ranges between 18C and 36C at a mean altitude of approximately 1070 metres above sea level. Over 95 percent of the inhabitants are from the Luo ethnic group. The majority of the population is Christian and there is a small Muslim community. Polygamy is common with about 21% of women in Nyanza province in a polygamous relationship.[4] The main income generating activities are fishing, subsistence farming and cattle keeping and small scale businesses. Houses are typically made out of mud, brick or cement and roof are either made of thatched grass or iron sheets. The majority have completed primary school education but only 23% finished secondary school. The total population in the Asembo area of the HDSS was 64,491 with 22% women of childbearing age (15‐49 yrs) in 2010. The total fertility rate for 2009 was estimated at 4.3 in this area. Migrations are significant and the rates of out‐ and in‐migration for females in Asembo were 127 and 111 per 1000 person year respectively in 2009.[5]
Nyanza province has a high burden of disease and worst health indicators compared to the overall Kenyan national statistics.[6] Malaria transmission is perennial and holo‐endemic with peaks following the 2 rainy seasons. Annual cross‐sectional survey in this area showed parasitaemia of 42% in under‐5 years old, 60% in 5‐14 years old and 20% in over 14 year old (unpublished KEMRI/CDC data for 2010). Whereas the national HIV prevalence is 6% (4% for men and 8% for women), the prevalence for Nyanza province is close to double around 14% (11% for men and 16% for women).[4] In Asembo, infant mortality was 94 per 1000 live‐births and under five mortality ratio was 181 per 1000 live‐births (unpublished HDSS fact sheet for 2009).
Communitymobilizationandformativeresearch
The success and acceptability of any study is highly dependent on efforts made to involve the community and provide enough information in an accessible manner. Community mobilisation activities started in September 2010. A series of meetings were held with District Medical Officer for Health (DMOH), the village chiefs, district officers and counsellors, the community advisory board
Chapter 8 Appendix
164
(CAB) set up by KEMRI/CDC and community members to introduce and get feedback on the proposed study plans. “Barazas” (community meetings) where held in the pilot villages in November and in each IEIP study villages in February and March 2011. Study brochures were also distributed at the community meetings and at the central health facility. Formative research involving 10 focus group discussions was carried out in September 2010 with the aim to explore the socio‐cultural context around pregnancy and to investigate acceptability of proposed study procedures. A pilot study was subsequently carried out in 3 villages in November‐December to assess implementability and identify bottlenecks before wider implementation. Overall positive feedback was received through community meetings and acceptability during the pilot was high. Some concerns were raised regarding the need for the study to provide free ANC, delivery and referral services. Lessons learned in terms of field logistics were incorporated for the study implementation which started on the 15th of February 2011. Free ANC services were offered at the study health facility but covering delivery and full referral services was beyond the scope of this study and this was clarified at Barazas before implementation. Participants presenting to Lwak Mission Hospital for their ANC received a free delivery kit which entitled them to reduced costs of delivery at the same health facility (from KES 900 to KES 500 for a standard delivery).
Figure s8. Map of study area
StaffingandfieldinfrastructureThe core study team comprise of a study coordinator, part‐time field coordinator (shared with the IEIP surveillance activities), a VR supervisor, 2 field assistants and 3 study nurses. All went through 1 week training covering Good Clinical Practice (GCP), research ethics and consenting procedures, fieldwork etiquette and interview techniques, as well as protocol and study tools (including training
Association between artemisinin exposure in the first trimester and miscarriage
165
on the use of the netbook for data collection). In addition to the overall protocol training, study nurses were also trained on newborn examination for recognition of congenital malformations detectable by surface examination and Ballard score for gestational assessment (2 days) using training documents developed by the Malaria in Pregnancy Consortium (MIPc), use of ultrasound for gestational assessment (5 days) and a 2‐week motorbike training for home follow ups.
A team of 40 village based staff called “village reporters” or VRs were involved for the community mobilisation, enrolment and pregnancy detection activities. VRs usually act as community liaison officers for various KEMRI/CDC projects, some of them have been involved in research since the pivotal bednet trial which was carried out in Asembo and Gem in 1998‐2001.[7] All the 40 VRs involved in the study were females. They received a 3‐day training giving an overview of research ethics, the study (background, objective, procedure, expected output and role and responsibilities), consenting process and pregnancy test procedure and counselling. A 1 day refresher was provided before initiation of each village and each VR was accompanied by the supervisor or field assistants during the first few visits.
Datacollection
EnrolmentFollowing community mobilisation, door‐to‐door enrolment was carried out to provide information about the study to eligible women of childbearing age (15‐49 years). All female age 15‐49 years residents of households within the defined catchment area and participating in the population‐based infectious disease surveillance project (PIBDS) [3] were eligible for enrolment in the study. Women were excluded if they refused to participate; were unable to provide informed consent due to mental, physical or social inability or if they refused to be followed up at the end of the pregnancy. A list of all PIBDS female participants within the specified age group was provided to the VRs. Women above 18 years of age and mature minors (defined as an under 18 year old who is either married, head of a household or already a parent) could provide consent for themselves. Parental consent and assent were sought for minors (<18 years of age not qualifying as a mature minor). Information on screening and enrolment was collected on simple scannable forms in addition to the signed informed consent forms. Enrolment was active throughout the study period (from mid February 2011 to mid February 2013) whereby newly eligible women (turned 15 years of age or in‐migrant joining BIPDS) could be enrolled at any time. Of the women screened (not including those not at home after 3 attempts or migrated, N=1102), 92% (N= 5536) consented or assented to be part of the study. About 2% were not eligible to participate (68 were incapacitated, 17 were non‐PIBDS, 12 under 15 years of age, 1 was over 49 years of age and 1 was a male). Refusal rate was low at 6% (375) of screened participants. The majority of refusals (274/375) did not provide a reason as to why they dislike the study, of those who provided a reason 57% refused because they couldn’t become pregnant anymore (some felt too old and had already gone through menopause and some had done bilateral tubal ligation), for 1% the parents refused for their daughter to be part of the study and for less than 1% the husbands refused. Figure 2 depicts the participant flow from screening to inclusion in the data analysis. For girls under 18 years of age, 10% were mature minor and refusal rate was similar to the overall cohort (i.e. 6%). Participants were free to withdraw at anytime during the study, 1% (N=60) did so for various reason: the majority reported the study was not beneficial to them, 10% reported to dislike pregnancy test, 10% were later assessed as incapacitated and unable to answer study questions, 1 participants husband requested for his wife to be withdrawn, 2 participants felt too old and 4 refused to be followed up after identification of a pregnancy.
Chapter 8 Appendix
166
Figure s9. Study participant flow diagram from screening to inclusion in data analysis for miscarriage.
Analysis
Excluded from analysis (n=53) Detected at outcome (n=33) No GA information (n=15) Pregnancy end date error (n=5)
Assessed for eligibility N=6010
Excluded (n=474) Not meeting inclusion criteria (n=99) Declined to participate (n=375)
Consented (n=5536)
Pregnancies (n=1453)
Number of pregnancies per participants: Single pregnancy (n=1266) Two pregnancies (n=92) Three pregnancies (n=1)
Follow‐Up
Enrolment
Loss to follow up (n=85) Migrated (n=67) Withdrawal or refused follow up (n=13) Death (n=5)
Pregnancy Outcomes (n=1368)
Overall Pregnancy Outcomes (n=1315)
Miscarriage (n=1126)
Excluded from analysis (n=327) Detected at outcome (n=33) No GA information (n=15) Pregnancy end date error (n=5) No follow up (n=41) Entered after 28 weeks (n=233)
Association between artemisinin exposure in the first trimester and miscarriage
167
PregnancydetectionstrategyAt the time of enrolment, all consented participants were asked about their pregnancy status and offered to take a pregnancy test if they were not visibly pregnant. About 10% of participants were detected as pregnant at the time of enrolment. This was lower that the estimated proportion of WOCBAs pregnant 13% based on the fertility rate in this area (TFR 4.3)3.
Following the initial enrolment period, VRs (who were trained on pregnancy testing and counselling) visited all study participants every 3 months at their home to collect information on their pregnancy status and offer pregnancy testing4. Refusals remained very low at less than 2% of participants refusing the VRs visit every round. Although the number of women reporting to be menstruating at the time of the 3 monthly visit and in no need of taking a pregnancy test (7%) was higher than expected (about 2% (5/30 days) would be expected to be menstruating on any 1 day). This suggests some participant used menstruation as an excuse to avoid a pregnancy test. In between home visits the VRs also kept a stock of test strips, gloves and urine cups and participants were made aware that they could access pregnancy tests through the VRs or the study health facility any time they wished. These were accessed by 237 participants over the study period and only 12% (n=179) of pregnancy were detected that way. Although some pregnancies (n=30) were detected at delivery, about half (45%) were detected in their first trimester of pregnancy and 69% were detected using a pregnancy test before being visible. The average gestational age at the time of pregnancy detection varied according to the approach: pregnancies detected at the time of enrolment in the study had an average gestational age of 17 weeks (range 0‐42 weeks); pregnancies detected by pregnancy tests by VRs were detected around 16 weeks; whereas pregnancies detected at the ANC were on average 23 weeks (range 2‐41 weeks).
The graph below depicts the pregnancy detected by the different strategy and the line represent the average gestational age at detection. The latter decreased over the study period as the majority of pregnancies were detected in the first trimester. In the last rounds of active pregnancy detection gestational age was around 12 weeks gestation.
Figure s10 Graph of the number of pregnancies detected per round and average gestational age at detection over time.
3 This estimate of 13% was based on the formula below given a fertility rate of 4.3 and a pregnancy rate of 6.1 (adjusted for
pregnancy loss and with the proportion of live‐birth at 70.4% for Kenya):
Proportion pregnant at any year is 6.1/ (49‐15) = 0.18 and the proportion pregnant at any month in the year as 0.18 x (40/52) =0.13 4 The 3 monthly home visits started in October 2011
0
5
10
15
20
25
30
0
20
40
60
80
100
120
140
Feb
Mar
Apr
May Jun
Jul
Aug
Sep
Oct
Nov
Dec Jan
Feb
Mar
Apr
May Jun
Jul
Aug
Sep
Oct
Nov
Dec Jan
Feb
2011 2012 2013
Average Gestational Age (weeks)
Number of Pregnan
cies Detected
Outcome
ANC
Active detection
Passive detection
Enrolment
Chapter 8 Appendix
168
Characteristicsofthepregnanciesenrolledinthestudy
Around 72% of pregnancies were seen at least once at Lwak Mission Hospital before the end of their pregnancy. Overall 92% of women made at least 1 visit to ANC for the index pregnancy.
Table s1. Characteristics of pregnant participants.
Demographic characteristics N=1453
Age in years (mean (SD)) 26 (6.9)
Gravidity (n (%)) Missing n= 83
Primigravidae 298 (22%)
1‐3 pregnancies 627 (46%)
4+ pregnancies 445 (33%)
Gestational age at detection in weeks (mean (SD)) 17 (10.2) Missing n= 15
Education level (n (%)) § Missing n= 83
None 10 (1%)
Primary not completed 593 (43%)
Primary completed 652 (48%)
Secondary completed 94 (7%)
Higher education 21 (2%)
Occupation (n (%)) Missing n= 98
Not working 501 (37%)
Farming 415 (31%)
Salaried worker 37 (3%)
Skilled worker 55 (4%)
Small business 314 (24%)
Other 33 (3%)
Marital Status (n (%)) Missing n= 83
Single 298 (21%)
Married 1052 (72%)
Divorced 7 (1%)
Widowed 13 (1%)
PregnancyfollowupPregnant participants were referred for free antenatal care at Lwak Mission Hospital where trained study nurses were running the antenatal clinic. Participants were managed according to the Ministry of Health guidelines for ANC including a full ANC profile (covering blood group, syphilis, urinalysis, Hb and HIV) and in addition received ultrasound scan if they presented before 6 months of gestation. Following the ANC consultation the participant was asked a series of additional question for the study covering demographic characteristics, medical history and description of any illnesses and/or medication taken since the beginning of the index pregnancy. At each ANC revisits (scheduled according to MOH recommendation of minimum of 4 ANC visit per pregnancy), the participants were asked about any illnesses and treatment sought since their last ANC visits.
Association between artemisinin exposure in the first trimester and miscarriage
169
Table s2. Antenatal care summary for study participants.
Antenatal Care Summary
Gestational age at 1st study ANC visit in weeks (mean (SD)) 23 (8.7)
Number of ANC visit (n (%)) by end of pregnancy N= 1368, Missing n= 85
none 111/1368 (8%)
1 126/1368 (9%)
2 207/ 1368 (15%)
3 325/1368 (24%)
4+ 599/1368 (44%)
IPTp doses (HIV negative) (n (%))by end of pregnancy N= 1091, Missing n= 447
none 313 (29%)
1 136 (13%)
2 234 (16%)
3 270 (19%)
4 138 (10%)
ANC Profile
HIV positive (n (%)) 282/1235 (23%) missing n=218
Hb (mean (sd)) 11.1 (1.9) missing n=469
Anemia (Hb<11g/dl) 440/984 (45%)missing n=469
Malaria slide positive (n (%)) 133/987 (14%) missing n=466
Syphilis reactive test (n (%)) 79/1093 (7%) missing n=360
GestationalageassessmentGestational age was assessed using multiple methods and was determined using the most accurate measurement available for each participant. Ultrasound assessment of gestational age up to 24 weeks provides the most accurate prediction of the expected date of delivery, this was offered to all participants presenting at the study facility before 24weeks gestation. Out o f 995 pregnancies seen at the study health facility at least once for ANC, only 428 (43%) presented at or before 24 weeks and had a dating ultrasound completed. The data for 3 of those was unreliable and alternative method of gestational age assessment was used. Ballard score was performed on all live newborns seen by a study nurse within 7 days of birth (73% of 1221 live born babies). In case of a pregnancy loss, the participant was asked if she knew how many months pregnant she was at the time of the loss and 26% did and reported a gestational age. The first day of the last menstruation period (LMP) was asked to all participants at the time of pregnancy detection by trained VRs and was asked again during follow up visit by a study nurse either at ANC or at the time of outcome follow up (if the participant had not been seen at ANC at the study health facility). Only 19% of cases (which had both reported VR and ANC LMP) reported the same LMP date at the time of detection by village based staff and at the time of follow up by a study nurse. Forty four percent of the LMPs reported at these consecutive visits were within the same week. Eleven percent reported not to know the date of their LMP. Fundal height was also measured at each ANC visit at the study health facility and recorded by a trained study nurse. Only measurements done between 20 and 34 weeks gestation in singleton pregnancies were used and gestation estimated using the McDonald rule formula: # of cm x 8/7 = weeks gestation. There were 17 participants for whom no gestational age data was available as the participants didn’t know their LMP, had no ultrasound scan, fundal height or Ballard Score that could be used (including one which was lost to follow up).
The table below depicts the accuracy of each gestational age assessment method used and the corresponding expected error margin. The methods are listed in order of decreasing accuracy. In
Chapter 8 Appendix
170
addition if the LMP agreed within 10 days of ultrasound, fundal height or Ballard estimates the error margin was reduced to +/‐ 10 days. The last column represents the number of participants for which the specific method was used to assess gestational age in the data analysis.
Table s3. Accuracy correction for the gestational age assessment methods listed in order of decreasing accuracy.
Method of Gestational Age Assessment Accuracy Correction
Number of participants (% LMP confirmed)*
Mean gestational age at pregnancy detection in weeks
1. Ultrasound before 14 gestational weeks ± 5 days[8] 153 (68%) 8.4
2. Ultrasound between 14 and 24 gestational weeks
± 2 weeks[8] 273 (56%) 14.8
3. Ballard score within 7 days of birth ±2 week[9] 574 (41%) 21.8
4. LMP/Self report GA at end of pregnancy ± 4 weeks 431 (NA) 14.9
5. Fundal Height ± 6 weeks‐ when multiple measurements
[10]
5 (0%) 17.1
*17 had no gestational age information
SourceofinformationondrugexposureDrug exposure data was captured using 3 approaches: a) interview of pregnant participants seen at the study facility antenatal clinic and at the time of pregnancy outcome follow up (from here on referred to as EMEP data); b) record linkage of data on all drugs prescribed to WOCBAs at outpatient department in Lwak hospital (from here on referred to as Lwak OPD data) and lastly c) via an active approach involving regular home visits by field workers as part of the PBIDS.
a) EMEP self‐reported drug exposure: Pregnant participants attending ANC in Lwak were asked about drug exposure since the beginning of their pregnancy and at subsequent ANC visit about any drugs taken since their last visit. At the time of the pregnancy outcome follow up visit, participants were interviewed by a study nurse regarding drugs used in pregnancy and other medically relevant information. The limitation of drug histories is accuracy of recalled information. Visual aids depicting photographs of all antimalarial drugs found in the study area were used to facilitate recognition of drug names. Calendar marking public holidays and school closures was also used to enhance recall of dates.
b) Lwak OPD Health Facility records: The existing IEIP system in Lwak Mission Hospital is set up to document diagnosis and related treatments in all age‐groups at out‐patient department (OPD). All participants were already enrolled in the IEIP program and had access to free care at Lwak Mission Hospital for infectious syndrome.
c) PBDS (bi‐)weekly home visits: This active approach involved home visits by trained fieldworkers on a weekly (from January 5, 2010 to May 26, 2011) and then bi‐weekly basis (May 27, 2011 onwards), before a woman might becomes aware of her pregnancy. The interview component on drug exposure was also enhanced by introducing visual aids for recall (same as the one mentioned above under EMEP self‐reported drug exposure). This approach should limit drug exposure recall bias as the information is collected continuously (concurrently to when exposure occurs or close to this time) before a woman might become aware of her pregnancy. However a limitation of this approach is that interview data is incomplete for weeks when no one was found at home (54% of times) and on some occasion the interview was completed by a proxy (family member available on the day of the interview) instead of the participant herself.
Figures s4 and s5 below depict the number of ACT exposures during the 1st trimester of pregnancy and embryo‐sensitive period detected by the 3 different approaches as well as agreement between
Association between artemisinin exposure in the first trimester and miscarriage
171
the data. Overall there are 352 probable ACT exposure in the 1st trimester of pregnancy out of the 1453 pregnancies (24%), however only 88 (25% of the 352 possible exposures) were confirmed by at least 2 data sources. Similarly for exposures occurring in the embryo‐sensitive period, out of the 240 possible exposures only 49 (20% of these exposures) were confirmed by at least 2 of the data sources. After excluding exposures which fall within gestational age estimate error margin (for example: an ACT exposure at 13 weeks estimated using the Ballard Score method has an error margin of 2 weeks therefore the true gestational age most probably lies between 11 and 15 weeks. Since the upper margin for exposures falls outside the 1st trimester of pregnancy such cases are excluded from the exposure group), there are 284 probable 1st trimester exposure and 114 probable exposures within the embryo‐sensitive period of which 62 (22%) and 24 (21%) are confirmed by at least 2 data sources respectively. The 24 confirmed (by 2 data sources and within probable gestational age estimation margins) exposures during the embryo sensitive period represent 1.7% of the pregnancies which is lower than the anticipated 3% inadvertent exposure based on the assumption of 1 antimalarial treatment every 2 years for this population5.
Overall 34% of the EMEP ACT 1st trimester exposures were confirmed by one of the other data source and 87% (1028/1176) of EMEP unexposed were unexposed for all data sources. However, there is no gold standard to assess EMEP approach for detecting drug exposures. The Lwak OPD data source is limited to participants receiving treatment from this facility (only 68 of the 168 (40%) EMEP ACT exposures in the 1st trimester were reported to be from Lwak health facility and 48% of cases reporting taking an antimalarial at a 2 weekly home visit also sought care from Lwak health facility). Thus LWAK OPD data source is incomplete in terms of ACT received from other health facilities, drug outlets or relatives/friends and maybe misleading in case a woman was prescribed an antimalarial but did not take it. The limitation of the PBIDS home visits is in part the completeness of the data as 32% of participant did not have all home visit for the period of their 1st trimester of pregnancy (due to instances when no one was at home) and on the other hand 26% of first trimester visits were completed by proxy (i.e. a relative responding on behalf of the participant) which might not be as accurate and complete compare to interview completed by the woman herself. In order to estimate sensitivity of the EMEP interviews, EMEP ACT 1st trimester exposures were compared to exposures confirmed by both PBIDS home visits and Lwak OPD. Only 17 of the 46 PBIDS‐LWAK confirmed ACT 1st trimester exposures were detected by EMEP reflecting a low sensitivity of around 37%. The positive predictive value for this sub‐group shows than only 10% of cases defined as exposed according to EMEP data source are truly exposed. For specificity, EMEP ACT 1st trimester unexposed cases were compared to unexposed as defined by both PBIDS (only including participants with all visits during their 1st trimester of pregnancies 469/1453) and Lwak OPD. Specificity was also poor, with 66% (342 of the 519) unexposed pregnancies being correctly identified by EMEP as unexposed. Cohen's κappa statistic was run to determine if there was agreement between the different data sources and there was only slight (according to the commonly cited scale[12]) agreement between the EMEP and both PBIDS and Lwak OPD data sources, κ = 0.177 (p < 0.001) and κ = 0.103 (p < 0.001) respectively. Kappa agreement remained slight for comparison limited to pregnancies with complete PBIDS visit in their 1st trimester (κ = 0.138) and comparison limited to exposures reportedly obtained from LWAK OPD (κ = 0.208).
5 The average pregnancy is 266 days (38 weeks) from conception (280 days or 40 weeks from LMP). The average number of treatments with ACTs in adults in the study areas is approximately 0.5 treatments per year (1 every 2 years) and the total fertility rate is estimated at 5.5. Under these conditions and using the model developed by Dr. Ian Hastings, we estimate the probability that an embryo will encounter artemisinins inadvertently during the critical 42 day (6 weeks) period of its development (week 4 to week 9 inclusive, from conception) is about 6.0% [11. Ward SA, Sevene EJ, Hastings IM, Nosten F, McGready R (2007) Antimalarial drugs and pregnancy: safety, pharmacokinetics, and pharmacovigilance. Lancet Infect Dis 7: 136‐144.] Under these circumstances the potential ratio of exposed versus unexposed pregnant women would be 1:16. In Kenya, we anticipated that the probability of inadvertent exposure will reduce significantly and maybe halved to 3% following the introduction of pregnancy testing at the household level.
Chapter 8 Appendix
172
Table s4. Validity of EMEP approach for detection of ACT 1st trimester exposures in terms of sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), % agreement and % positive agreement with PBDS and Lwak OPD data sources.
ACT 1st Trimester Exposures
EMEP‐PBIDS
EMEP‐PBIDS (Full
1st Trimester1)
EMEP‐Lwak
EMEP‐Lwak Source Only2
EMEP‐PBIDS& Lwak CONF3
EMEP‐PBIDS(Full 1st
Trimester) OR LWAK4
Sensitivity 25% 19% 29% 25% 37% 65%
Specificity 91% 92% 89% 96% 89% 66%
PPV 33% 32% 12% 25% 10% 24%
NPV 87% 85% 96% 96% 98% 92%
% Agreement 81% 80% 86% 93% 87% 66%
% Positive Agreement 17% 14% 9% 14% 9% 21%
Kappa‐statistic 0.177 0.138 0.103 0.208 0.116 0.181
Kappa p‐values <0.001 0.002 <0.001 <0.001 <0.001 <0.001 1 Comparison between EMEP exposures and PBIDS exposures restricted to participants who had all PBIDS home visits for the period of their 1st trimester 2 Comparison between EMEP exposures and Lwak OPD exposures restricted to participants who reported that the ACT exposure was obtained from Lwak health facility at the time of EMEP interview 3 Comparison between EMEP exposures and PBIDS exposures confirmed in the Lwak OPD data source 4 Comparison between EMEP exposures and exposures defined by either PBIDS or Lwak OPD while unexposed are defined by PBIDS (only including participants with all visits during their 1st trimester of pregnancies 469/1453) and confirmed by Lwak OPD.
Table s4 summaries the relationship between potential predictors of EMEP data source ACT
exposure confirmation in the 1st trimester by another data sources (either PBIDS or Lwak OPD).
Overall demographic and obstetric characteristics were not significantly different between cases
with confirmed exposures detected by EMEP and those that weren’t detected. Factors associated
having a true positive ACT 1st trimester exposed according to EMEP data source, included having an
early 1st EMEP visit in the index pregnancy, the number of weeks between the exposure and the next
EMEP visit (cases with confirmed exposures having shorter time interval), the number of EMEP visits
(the higher the number the more likely to be detected), and having a follow up visit during
pregnancy as well as after the end of pregnancy follow up visit.
Asso
ciation betw
een artem
isinin exp
osure in
the first trim
ester and m
iscarriage
173
ACT Exp
osure in
1st
trimeste
r any d
ata source
N=353
Exposure co
nfirm
ed
by 2
data so
urce
s N=88 (2
5%)
EMEP
N=169
LWAK
N=70
PBIDS
N=219
30
3
17
38
111
134
20
A.
EMEP
N=131
LWAK
N=49
PBIDS
N=179
18
2
13
29
87
119
16
ACT Exp
osure in
1sttrim
este
r corre
cted fo
r gestatio
nal age
erro
r any d
ata
source
N=284
Exposure co
nfirm
ed
by 2
data so
urce
s N=62 (2
2%)
B.
Figure s1
1. V
enn diagram
of A
rtemisin
in Combinatio
n Th
erapy (A
CT) exp
osure in
1st trim
ester of p
regnancy acco
rding to
data so
urce. A
. All A
CT exp
osures. B
. ACT
exposures n
ot w
ithin possib
le error m
argins fo
r gestational age assessm
ent.
EMEP
N=101
LWAK
N=41
PBIDS
N=158
Exposure co
nfirm
ed
by 2
data so
urces
N=49 (2
0%)
19
210
18
70
111
11
ACT Exp
osure in
em
bryo
‐sensitive
perio
d an
y data
source N
=241
C.
ACT Exp
osure in
embryo
‐sen
sitive perio
d
corre
cted fo
r gestatio
nal age
erro
r any d
ata source
N=115
Exposure co
nfirm
ed
by 2
data so
urce
s N=24 (2
1%)
EMEP
N=48
LWAK
N=17
PBIDS
N=95
7
0
5
12
32
55
5
D.
Figure s1
2. V
enn diagram
of A
CT exp
osures in
embryo
‐sensitive p
eriod (6
‐12 weeks p
ost LM
P) acco
rding to
data so
urce. C
. All A
CT exp
osures. D
. ACT exp
osures n
ot w
ithin
possib
le erro
r margin
s for gestatio
nal age assessm
ent.
Chapter 8 Appendix
174
Table s5. Summary of factors associated with ACT exposure in 1st trimester of pregnancy confirmation by EMEP data source of exposures confirmed by prospective home visits (PBDS) and Lwak outpatient records.
Overall LWAK or PBIDS ACT 1st Trimester (N=233)
EMEP ACT 1st Trimester Unexposed
(N=177)
EMEP ACT 1st Trimester
Confirmed (N=56)
Chi‐Square P‐values
Age in years* 25.6 7.1 25.6 7.4 25 6.4 0.877
Gravidity 0.652
Primigravidae 62 27% 49 28% 13 23%
1‐3 pregnancies 88 38% 64 36% 24 43%
4+ pregnancies 82 35% 63 36% 19 34%
Previous pregnancy loss 33 14% 25 14% 8 14% 0.988
EMEP pregnancy number 0.703
1 220 94% 166 94% 54 96%
2 12 5% 10 6% 2 4%
3 1 0% 1 1% 0 0%
Any report of bleeding 3 1% 1 1% 2 4% 0.133
Gestational week at detection * 16.0 9.7 16.0 9.7 15.9 9.9 0.944
Education level 0.229
None or Primary not completed 114 49% 90 51% 24 43%
Primary completed 103 44% 73 41% 30 54%
Secondary completed 15 6% 13 7% 2 4%
Occupation 0.130
Not working 94 40% 68 38% 26 46%
Farming 71 30% 60 34% 11 20%
Small business or Salaried 62 27% 44 25% 18 32%
Other 4 2% 4 2% 0 0%
Marital Status 0.446
Single 63 27% 50 28% 13 23%
Married 169 73% 126 71% 43 77%
Pregnancy Outcome 0.601
Live birth 207 89% 155 88% 52 93%
Induced Abortion 2 1% 1 1% 1 2%
Miscarriage 16 7% 13 7% 3 5%
Stillbirth 1 0% 1 1% 0 0%
Neonatal death 7 3% 7 4% 0 0%
Gestational week at 1st EMEP visit* 21.7 8.6 22.4 8.6 20 8.5 0.077
Number of EMEP visit <0.001
1 51 22% 46 26% 5 9%
2 49 21% 42 24% 7 13%
3 34 15% 25 14% 9 16%
4+ 99 42% 64 36% 35 63%
EMEP visit at outcome only 51 22% 46 26% 5 9% 0.007
Weeks between EMEP interview & exposure*
19.3 14 20.8 14.9 15 9.7 0.003
*(mean (SD))
Association between artemisinin exposure in the first trimester and miscarriage
175
PregnancyoutcomeassessmentA combination of facility‐based and home‐based follow‐up systems were used to ascertain
pregnancy outcomes because the majority of deliveries occur in the home setting (only 44%
delivered in a health facility according to KDHS 2008/9). This is particularly important for
miscarriages which occur outside health facilities for the vast majority. Therefore, a home‐based
follow up system was developed, using VRs that alerted their supervisor by phone regarding any
pregnancy outcome. VRs were incentivised to report any pregnancy outcome as soon as possible
after delivery (they received Ksh120 for pregnancy outcomes reported within 4 days and Ksh 90
otherwise based on the rate paid by HDSS for birth notification at the time). Simple scannable forms
including the expected date of delivery (based on ultrasound EDD if available otherwise on LMP)
were distributed to the field team as a prompt that a delivery was expected. Participants were
encouraged and counselled on the benefits of delivering at a health facility although the study did
not cover delivery costs. At their last ANC visit, participants received a simple delivery kit (including
sterile gloves, gauze, cord clamp, soap, razor blade, maternity pads and clear plastic sheet) to bring
to the health facility which got them a discount on the delivery cost. Once notified about a
pregnancy outcome the VR supervisor would coordinate the follow up to be done by a study nurse
either at home or at the health facility. Overall 62% of deliveries took place at a health facility (70%
excluding pregnancies ending in miscarriages). A detailed questionnaire about the delivery, outcome
and newborn examination findings was administered in addition to the questions regarding
illnesses/medication used during pregnancy. This took approximately 60min. If the mother
consented, the baby was photographed. Any case of suspected malformation or abnormality was
reviewed by the study paediatrician either by going through the newborn exam data and the photos
or by full physical review if the latter information was inconclusive. Cases needing further
investigation or intervention were referred to the appropriate facilities/ specialist.
Sixty four percent of deliveries occurred at a health facility. It can be common for women to travel to
deliver close to the husband if he is working outside of the study area or to be with family. Overall
33% of pregnancy outcomes were captured over 1 week after the end of pregnancy and 16% of
outcomes were captured over 1 month later. The average number of days between outcome and
follow up was 28 range (0‐755).
Table s6. Summary of pregnancy outcomes
Pregnancy Outcome§ N (%)
Live‐Birth 1208 (88.3%)
Neonatal Death* 23 (1.6%)
Miscarriage (<=28 weeks)** 105 (7.7%)
Induced abortions 8 (0.6%)
Stillbirth (>28 weeks) 26 (1.9%)
§ Including 2 twin pregnancies which had 1 livebirth and 1 neonatal death/ 1 stillbirth. There were 85 (6%) loss to follow up.
* Babies were not followed up to 28 days, these are the numbers detected by the time of follow up.
**including 5 suspected induced abortions
Chapter 8 Appendix
176
Qualitycontrol
DatamanagementEnrolment and pregnancy status
All scannable forms were verified in the field by the study assistants when returned by the VRs
before being sent for scanning at the KEMRI/CDC offices in Kisian. Once the forms arrived in Kisian
they were scanned by the data management team who verified the entries on the computer screen
before uploading the data to the database. The forms were filled in secure office space for easy
retrieval in case of any query or data correction needed.
ANC visits and pregnancy outcome follow up
All data collected by the study nurses for pregnancy follow up at the ANC or at outcome follow up
were captured electronically on a Netbook computer with an in‐house program for data collection
developed using visual studio 2008 (VB.net programming language) and stored on MSSQL SERVER
2005/2008 on the Netbooks and Ms Access for the central database held in the KEMRI/CDC office.
The data entry system contained in‐built skip pattern depending on the entry, data checks for
inconsistencies (i.e. data out of expected range) and certain fields which could not be left blank.
Hard copies of the questionnaire were available in case the netbook had a problem. The data was
backed up daily on SD‐cards in the field and was downloaded once a week and brought to the
central database in Kisian.
EnrolmentandpregnancydetectionA sample of eligible participants from each village was selected for re‐visit by field assistants every 3
months following VRs visit. The data collected by the VR was compared to the data collected by the
field assistant to verify that 1) the participant had been visited, 2) had been explained the study
adequately and 3) the data collected on the form was correct.
NewbornexamOn a bi‐weekly basis 2 study nurses would independently carry out a newborn and Ballard score
examinations for the same baby and compare their findings on form provided. Any discrepancy was
discussed at the time and the baby re‐examined to identify the correct value. In case were there was
ambiguity about a correct value further explanation was sought from the study paediatrician and
refresher training on Ballard assessment was organised.
UltrasoundQuality control for ultrasound scans was implemented during the pre‐implementation phase
following the ultrasound training. All ultrasound images were downloaded from the machine to a
laptop and sent to ultrasound expert at the collaborating institution at the University of Washington
(after removing all identifiers from the images). The reviewers provided feedback on a formatted
excel spreadsheet and highlighted areas needing re‐training or special attention. If required,
additional training materials were sent for review with the study nurses.
EthicalreviewandconsentThe EMEP study protocol was reviewed and approved by the institutional review boards of CDC (No.
5889), KEMRI (No. 1752) and the Liverpool School of Tropical Medicine (No. 09.70). Informed written
consent or assent was obtained from each participant. For illiterate participants, an independent
witness had to be there during the consenting process and sign the consent form while the
Association between artemisinin exposure in the first trimester and miscarriage
177
participant marked the form with a thumbprint. Minors (girls under 18 years of age) needed parental
consent before being asked for assent. Mature minor (defined as an under 18 year old who is either
married, head of a household or already a parent) provided consent for themselves.
References1. Adazu K, Lindblade KA, Rosen DH, Odhiambo F, Ofware P, et al. (2005) Health and demographic surveillance
in rural western Kenya: a platform for evaluating interventions to reduce morbidity and mortality from infectious diseases. Am J Trop Med Hyg 73: 1151‐1158.
2. Odhiambo FO, Laserson KF, Sewe M, Hamel MJ, Feikin DR, et al. (2012) Profile: the KEMRI/CDC Health and Demographic Surveillance System‐‐Western Kenya. Int J Epidemiol 41: 977‐987.
3. Feikin DR, Audi A, Olack B, Bigogo GM, Polyak C, et al. (2010) Evaluation of the optimal recall period for disease symptoms in home‐based morbidity surveillance in rural and urban Kenya. Int J Epidemiol 39: 450‐458.
4. Kenya National Bureau of Statistics (KNBS), Macro I ( 2010.) Kenya Demographic and Health Survey 2008‐09. Calverton, Maryland.
5. KEMRI/CDC (2010) Health and Demographic Surveillance System Report for 2009. Kisian: KEMRI/CDC. 6. Kenya National Bureau of Statistics (KNBS) and ICF Macro (2011) Kenya Demographic and Health Survey
2008‐09. Calverton, Maryland, USA. http://www.measuredhs.com/pubs/pdf/FR229/FR229.pdf. 7. Phillips‐Howard PA, Nahlen BL, Kolczak MS, Hightower AW, ter Kuile FO, et al. (2003) Efficacy of permethrin‐
treated bed nets in the prevention of mortality in young children in an area of high perennial malaria transmission in western Kenya. Am J Trop Med Hyg 68: 23‐29.
8. Seffah JD, Adanu RM (2009) Obstetric ultrasonography in low‐income countries. Clin Obstet Gynecol 52: 250‐255.
9. Sasidharan K, Dutta S, Narang A (2009) Validity of New Ballard Score until 7th day of postnatal life in moderately preterm neonates. Arch Dis Child Fetal Neonatal Ed 94: F39‐44.
10. White LJ, Lee SJ, Stepniewska K, Simpson JA, Dwell SL, et al. (2012) Estimation of gestational age from fundal height: a solution for resource‐poor settings. J R Soc Interface 9: 503‐510.
11. Ward SA, Sevene EJ, Hastings IM, Nosten F, McGready R (2007) Antimalarial drugs and pregnancy: safety, pharmacokinetics, and pharmacovigilance. Lancet Infect Dis 7: 136‐144.
12. Landis JR, Koch GG (1977) The measurement of observer agreement for categorical data. Biometrics 33: 159‐174.
Chapter 8 Appen
dix
178
TableS1Miscarriageweeklyrate,cumulativeprobabilitiesofsurvivalandfailureandrem
ainingriskofm
iscarriage.
Ges
tatio
nal
wee
k Pr
egna
ncie
s D
etec
ted
durin
g w
eek
Preg
nanc
y-W
eeks
at
Ris
k
Mis
carr
iage
Indu
ced
abor
tion
Loss
to
follo
w u
p &
w
ithdr
awal
s
Wee
kly
mis
carr
iage
rate
pe
r 100
0 pr
egna
ncy-
wee
ks
(95%
CI)
Cum
ulat
ive
Prob
abili
ty o
f Su
rviv
al
Cum
ulat
ive
Prob
abili
ty
of F
ailu
re
Rem
aini
ng
risk
of
mis
carr
iage
1
3 2.
7
0 0
0 0.
00
1.00
0
0.00
0
0.19
1
2 8
8.4
0
0 0
0.00
1.
000
0.
000
0.
191
3 15
18
.4
0 1
0 0.
00
1.00
0
0.00
0
0.19
1
4 22
32
.4
0 0
0 0.
00
1.00
0
0.00
0
0.19
1
5 43
68
.4
2 0
0 29
.23(
7.31
- 1
20)
0.
971
0.
029
0.
191
6 74
12
7.6
2
0 0
15.6
8(3.
92-
62.
69)
0.95
4
0.04
6
0.16
7
7 78
19
8.4
5
3 0
25.2
(10.
49-
60.
54)
0.93
0
0.07
0
0.15
3
8 69
27
3.1
2
1 1
7.32
(1.8
3- 2
9.2
8)
0.92
3
0.07
7
0.13
2
9 65
32
9.1
3
0 0
9.11
(2.9
4- 2
8.2
6)
0.91
5
0.08
5
0.12
5
10
64
388.
9
6 0
0 15
.4(6
.92-
34.
28)
0.
901
0.
099
0.
117
11
54
441.
4
7 1
1 15
.78(
7.52
- 3
3.1)
0.
886
0.
114
0.
104
12
59
490.
6
6 1
1 12
.2(5
.48-
27.
16)
0.
877
0.
123
0.
089
13
55
540.
4
12
1 0
22.2
3(12
.62-
39.
14)
0.85
8
0.14
2
0.07
8
14
42
575.
1
3 0
1 5.
21(1
.68-
16.
15)
0.
854
0.
147
0.
058
15
52
620.
3
4 0
0 6.
44(2
.42-
17.
15)
0.
848
0.
152
0.
053
16
41
661.
4
9 0
0 13
.59(
7.07
- 2
6.11
) 0.
838
0.
162
0.
046
17
43
699.
3
2 1
0 2.
86(0
.72-
11.
44)
0.
835
0.
165
0.
033
18
44
735.
3
5 0
0 6.
8(2.
83-
16.
34)
0.83
0
0.17
0
0.03
1
19
31
765.
3
5 0
0 6.
53(2
.72-
15.
7)
0.82
4
0.17
6
0.02
4
20
34
793.
6
2 0
1 2.
52(0
.63-
10.
08)
0.
822
0.
178
0.
018
21
29
823.
3
4 0
1 4.
86(1
.83-
12.
96)
0.
818
0.
182
0.
015
22
30
849.
6
1 0
1 1.
18(0
.17-
8.3
7)
0.81
6
0.18
4
0.01
0
23
24
870.
0
0 0
0 0.
00
0.81
6
0.18
4
0.00
9
24
35
903.
1
1 0
0 1.
11(0
.16-
7.8
7)
0.81
5
0.18
6
0.00
9
25
29
932.
6
0 0
0 0.
00
0.81
5
0.18
6
0.00
8
26
21
954.
0
2 0
0 2.
1(0.
52-
8.38
) 0.
813
0.
187
0.
008
27
33
984.
4
4 0
12
4.06
(1.5
3- 1
0.8
3)
0.81
0
0.19
1
0.00
6
28
29
100
0.0
2
0 5
1.99
(0.5
- 7.
94)
0.80
8
0.19
2
0.00
2
179
Chapter9
SummaryandDiscussion
Chapter 9
180
Post‐marketing surveillance of drugs used in pregnancy is challenging, especially in developing
countries where resources for pharmacovigilance are rare. There is a need to establish simple but
effective systems to monitor safety of drugs given during pregnancy in resource constrained
countries. Although ACTs are not recommended in early stages of pregnancy, inadvertent exposures
to ACTs are unavoidable; this will occur through different channels including outpatient clinics, the
informal sector, clinical trials or mass drug administration programmes because either the woman or
the prescriber is unaware of her pregnancy status. Deliberate exposures will occur in situations
where the benefit outweighs the potential risk, when no other effective antimalarial drugs are
available and particularly in the case of severe malaria. There are currently no established systems to
monitor drug safety in pregnancy in malaria endemic countries. The aim of this thesis was to develop
and evaluate pharmacovigilance systems in resource constrained settings to provide a better
estimate of the risk‐benefit profile of ACTs in pregnancy.
In Chapter 2 the published evidence with regard to the safety of artemisinin compounds when
administered during pregnancy was reviewed through systematic literature searches. At the time of
the review (November 2006), fourteen relevant studies (nine descriptive/case reports and five
controlled trials) were identified. Overall there were reports on 945 women exposed to an
artemisinin during pregnancy, 123 in the first trimester and 822 in second or third trimesters. The
limited data available suggested that artemisinins are effective and safe when used in late
pregnancy, although rare adverse events could not be ruled out. There was insufficient evidence to
effectively assess the risk–benefit profile of artemisinin compounds for pregnant women,
particularly with respect to exposure in the first trimester of pregnancy. This was in line with the
WHO recommendation following informal consultations convened in 2006 to assess the safety of
artemisinin compounds in pregnancy, highlighting the need to set up pharmacovigilance systems to
document first trimester exposure to ACTs.
In Chapter 3 we provided an estimate for the number of pregnancies living in areas with malaria
transmission in 2007. Updated maps of P. falciparum and P. vivax transmission were combined with
gridded population data and growth rates to estimate total populations at risk of malaria in 2007.
The numbers of pregnancies was derived from country‐specific demographic data from the United
Nations, combined with estimates for stillbirths and induced abortions derived from contemporary
published reviews and estimates of spontaneous abortions based on an established formula (using
the number of live‐births and induced abortions). We estimated that in 2007, 125 million
pregnancies occurred in areas with P. falciparum and/or P. vivax transmission of which
approximately 60% resulted in live births. Estimates from Africa (32 million) were similar to previous
estimates by WHO (25–30 million) but estimates for non‐African regions was much higher than
previously estimated (95 million vs. 25 million). These estimates of the number of pregnancies at
risk of malaria provided a first step towards a spatial map of the burden of malaria in pregnancy.
Recently Walker et al estimated the risk of placental infection and low birthweight attributable to
Plasmodium falciparum malaria in Africa taking into account heterogeneity in transmission and
parity‐dependent effects.[1] They report that 12 million pregnancies resulting in live‐births would
have been exposed to malaria infections without malaria protection (i.e. through insecticide treated
nets or IPTp) resulting in 900,000 low birthweight babies. They projected that the majority of
placental infections (65%) would occur early in the first trimester of pregnancy. Considering that
only about two thirds of pregnancies result in live‐births the number of pregnancies potentially
exposed to infection could be up to 18 million. It is difficult to predict what proportion of such
Summary and Discussion
181
infections would lead to treatment with an antimalarial as in areas of high transmission most malaria
infections remain asymptomatic particularly in multigravidae. Nevertheless even if only a small
proportion were to seek treatment, millions of pregnancies could be exposed to antimalarials each
year and many in the first trimester of pregnancy.
In Chapter 4, we estimated that the probability an embryo will encounter artemisinins during the
critical six‐ week period (weeks 6 to 12 post‐LMP) through accidental exposure is 12% for areas
where adults receive on average one treatment with three days of artemisinin‐based combination
therapy per year. This assumed that the likelihood of infections was similar throughout pregnancy. In
light of the modelling study by Walker et al, this probability is likely an underestimate. This
information is useful to grasp the scale of the problem related to malaria in pregnancy and the
potential for exposures to antimalarial in pregnancy. In this chapter we further described the
methodological considerations for the systematic assessment of pregnancy outcomes and congenital
malformations in women exposed to antimalarials early in pregnancy, as well as approaches to
capture drug exposure information, choice of comparison groups and sample size concerns. We
proposed a targeted prospective pharmacovigilance approach enabling timely assessment of the
risk‐benefit profile of antimalarials through the establishment of an international antimalarial
pregnancy exposure registry. The WHO recently published a description of methods for pregnancy
exposure registries in resource constrained settings which provides valuable tools for standard
methods of data collection facilitating pooled data analysis.[2]
A complementary approach to pregnancy exposure registries is illustrated in Chapter 5, which
describes the findings from a pilot study to assess the feasibility of record linkage using routinely
collected healthcare data as a means of monitoring the safety of ACTs in early pregnancy. This
study used data from paper‐based registers (2004–2008) from a mission‐run dispensary in Mlomp,
south‐western Senegal. Data from the outpatient clinics and delivery registers were linked based on
a probabilistic matching approach as no unique identifier was available. The findings from this
feasibility study suggest that record linkage using routine healthcare data is feasible in resource‐
constrained settings with a relatively well defined catchment population. Tapping into readily
available data sources of sites adequate for record linkage could greatly contribute to the high
numbers needed to provide adequate reassurance for ACT use in the first trimester of pregnancy,
in terms of stillbirths and major congenital malformations. However, in these settings assessment
of the risk of miscarriage and specific birth defects, such as congenital heart defects, would require
dedicated studies. Sites amenable for record linkage studies need to have reliable and
comprehensive medical records for treatment, pregnancy and maternity services. This requires a
high proportion of health facility deliveries and limited availability of the drug of interest outside the
central health facility record system (and therefore not captured in the health records). A prime
example of such sites is the Shoklo Malaria Research Unit (SMRU) clinic which has already
contributed significantly to what is known about ACT risk‐benefit profile in the first trimester of
pregnancy.[3] The unique set‐up of the SMRU allows it to serve a stable refugee population where
many pregnant women receive weekly antenatal care, pregnancies can be identified early, the
gestational age assessed at every contact, the health‐care provided, and drug exposure and birth
outcome recorded centrally. Other potential sites with comprehensive and reliable healthcare
records include agricultural estates providing free healthcare for their workers, refugee camps or use
of health insurance data where coverage is high and homogeneous.
Chapter 9
182
In Chapter 6 we investigated contextual information on perceptions of adverse pregnancy outcomes
in an area of high malaria transmission in western Kenya. Through ten focus group discussions we
explored the perceptions, beliefs and health‐seeking behaviours of women from rural western Kenya
regarding congenital anomalies and miscarriages. We found that lack of information regarding
causes of adverse pregnancy outcomes such as miscarriages and congenital anomalies could lead to
stigmatisation of the mother and further hamper health seeking behaviour. This could have
devastating consequences for children born with a malformation who are sometimes hidden from
society and don’t have access to care that could improve their quality of life. Although women
reported that care is usually sought in case of complications following a miscarriage, delaying care
until symptoms are pronounced could adversely affect recovery and the woman’s health. These
findings highlight the need for education about the potential cause of adverse pregnancy outcomes
and better information on where care is available. The qualitative data gathered through these focus
group discussions was informative to understand the socio‐cultural context around pregnancy and
pregnancy outcome. It was also interesting to understand that although some women were aware of
the potential danger of medicines used in pregnancy, drugs considered safe in medical practice were
being reported as potentially dangerous. Mostly a risk was perceived when drugs were used without
or not following clinician recommendations. This emphasised the need for education materials for
both community and healthcare providers regarding the consideration for the choice of treatment in
pregnancy.
In Chapter 7 we described healthcare provider and drug dispenser adherence to and knowledge of
national guidelines for treatment of uncomplicated malaria in pregnancy. We conducted a cross‐
sectional study from September to November 2013, in health facilities and randomly selected drug
outlets in an area of high malaria transmission in western Kenya. This study highlighted knowledge
inadequacies and incorrect prescribing practices in the treatment of malaria in pregnancy,
particularly in the first trimester. As the first line treatment for malaria, artemether‐lumefantrine, is
not recommended in the first trimester of pregnancy, all women of childbearing age should be
assessed for potential pregnancy. Such pregnancy enquiries were only observed in 44% of cases in
health facilities and 7% in drug outlets, although 93% and 49% reported routinely assessing for
potential pregnancy respectively. A reason for this could be that prescribers and dispensers do not
feel adequately prepared to enquire about potential pregnancy and/or the lack of access to
pregnancy tests. Prescription of the correct drug at the correct dosage was observed in 32% of all
cases in health facilities and 0% in drug outlets for first trimester cases. Knowledge was moderately
higher for health facility staff (56%) and remained nonexistent in drug outlets staff. The reason for
the discrepancy between what is reported and what was observed should be further explored.
Exposure to artemether‐lumefantrine in first trimester occurred in 16% and 51% of cases in health
facilities and drug outlets, respectively; none were a result of quinine stock‐out. Overall, SP was
prescribed as treatment in 11% of all cases and the vast majority of all women prescribed quinine
were given an insufficient supply. Provider knowledge in both settings was poor and was reflective
of the low levels of correct case management observed in practice. Such incorrect prescribing
practice can have serious consequence for the pregnant patients not only by the risk posed by using
a drug of undetermined safety for the unborn baby (ACTs) but also in terms of adverse
consequences due to the inability to clear a malaria infection when treated with an ineffective drug
(SP) or incomplete drug regimen (as seen with quinine). The latter also contributes to emerging
parasite resistance to these drugs. This study provided insights on the high potential for ACT
Summary and Discussion
183
exposure in pregnancy in this setting but more broadly revealed a need for better dissemination of
guidelines for case management of malaria in women of childbearing potential with the need to
include both formal and informal sectors.
In Chapter 8 the findings from a prospective cohort study on the risk of inadvertent exposure to
ACTs in the first trimester of pregnancy and its association with miscarriage in Western Kenya were
presented. Community‐based surveillance was used to identify early pregnancies among women of
childbearing age under health and demographic surveillance in western Kenya. Multiple data
sources (including outpatient records, longitudinal household level morbidity surveillance data and
study specific pregnancy follow up questionnaires) were combined to ascertain ACT exposures in
pregnancy. Out of the 1,126 pregnancies included in the analysis, there were 42 (4%) confirmed ACT
exposures in the suggested artemisinin embryo‐sensitive period (6‐12 weeks post LMP) and 75 (7%)
confirmed exposures in the first trimester. This study showed that pregnancy detection in the first
trimester (before most women present for ANC) was feasible through community based testing in
this setting. We provided the first estimates for the rates of miscarriage stratified by gestational
week in this area and found that the cumulative probability of miscarriage was 19% in this
population during the period that women were under observation (up to 28 weeks of gestation).
Compared to pregnancies without reported malaria and antimalarial treatment, those with a
confirmed ACT exposure in the suggested artemisinin embryo‐sensitive period did not have a
statistically significant different hazard of miscarriage [HR=0.85 95% CI (0.22‐3.33)]. There was also
no statistical difference for the confirmed exposures in the first trimester as a whole, although the
effect estimate suggested a potential increase in risk [HR=1.60 95% CI (0.70‐3.68)]. As noted from
the upper bound of the confidence interval, we could not rule out a 3.7 fold increase in miscarriages
linked to ACT exposure in early pregnancy. In line with what was observed in the cross‐sectional
survey on prescribing behaviours (chapter 7), very few women in their first trimester of pregnancy
were treated with the recommended treatment, quinine (n=34 with 22 exposed to both quinine and
ACTs). This hampered our ability to compare pregnancies exposed to ACTs in the first trimester to
those exposed to a safe drug. Although an increased risk of miscarriage could not be discarded, the
results are consistent with recent findings reported from the Thai‐Burmese border and provide
further reassurance regarding the use of ACTs in early pregnancy.[3] Findings from pooled data
analysis combining data from this study and 2 other sites that were part of the Malaria in Pregnancy
Consortium, assessing risk of miscarriage, stillbirth and major congenital malformations are
forthcoming and will provide further insights on the risk‐benefit profile of ACTs.
ACT risk‐benefit profile in first trimester of pregnancy: where are we to date? Since the review presented in chapter 2, more data has been published, increasing the number of
documented first trimester exposures to 757 (see table 1). The evidence so far is reassuring and
rules out artemisinin derivatives as major teratogens (defined as a drug that would increase the risk
of major congenital malformations by 5 to 10 fold). The 757 documented first trimester exposures
confer enough statistical power to detect relative risk of 2.3 or greater for overall major
malformations and a RR of 1.3 for miscarriages (assuming the background rate of 0.9% for major
malformations detectable by surface examination at birth and 12% for miscarriage with ratio of
exposed to unexposed of 1:4 at 80% power). However for specific congenital anomalies which occur
at frequencies of 1/1000 for the most common, the 757 documented exposures could only detect a
relative risk of 6.6.
Chapter 9
184
Table 13. Description of documented ACT first trimester exposures.
Author Country Publication Year
Number of first trimester exposures
Reference
McGready TBB 2012 64 [3]
Deen Gambia 2001 77 [4]
Adam Sudan 2009 62 [5]
Mayando Zambia 2010 156 [6]
Rulisa* Rwanda 2012 96 [7]
Mosha Tanzania 2014 172 [8]
Poespoprodjo Indonesia 2014 11 [9]
Dellicour Kenya Not yet published 80 MiPc
Sevene Mozambique Not yet published 3 MiPc
Tinto Burkina Faso Not yet published 36 MiPc
Total 757
*Rulisa et al published results from an observational study following pregnant women exposed to artemether‐lumefantrine however pregnancy outcomes were not reported separately for first vs second/third trimester exposures
Interestingly, few studies reported the effect of exposures during the suggested artemisinin embryo‐
sensitive period (only McGready et al 2012 and the study presented in chapter 8). The suggested
period (6‐12 weeks post LMP) was derived from the observed causal mechanism in animal models
indicating that primitive erythroblasts are the primary target for embryotoxicity and these are the
weeks when primitive erythroblasts are the predominant form of erythrocyte in circulation in human
embryos.[10] As considering the exposure for the whole first trimester could bias effect estimates
towards the null, it is important to carry out analysis to explore the associations between ACT
exposures and outcomes during this period of maximum sensitivity in humans. This was performed
and presented in chapter 8 where we found no trend of increase in risk to exposure in the sensitive
period compared to the whole first trimester. Another point for consideration is the need to
investigate congenital cardiovascular defects as this was one of the safety signals identified from
preclinical reprotoxicology studies in rodents.[11,12] None of the studies were designed to assess
cardiovascular defects, which require review by a cardiologist and access to specialised equipment
(ultrasound for echocardiography, electrocardiogram and x‐ray). Furthermore, most of the studies in
table 1 were not designed to capture early pregnancy losses and adjustment for potential survival
bias will need to be considered when assessing miscarriages as an end point.
Meta‐analysis with individual patient data would be a useful next step for better evaluation of the
safety profile of ACTs in the first trimester of pregnancy. This will need to take into account
heterogeneity in data collection methods and different levels of certainty regarding potential
exposure misclassifications. Review of current evidence by the WHO Technical Expert Group (TEG)
on malaria chemotherapy will be a first step in deciding if more and what type of data is needed to
inform the Malaria Policy Advisory Committee (MPAC) which provides independent advice to the
WHO for the development of policy recommendations.
Safety is not an absolute property of a drug and it will depend on the appropriate use of the drug
considering indication, dosage, efficacy and suitability for the patient. A balance between risk and
benefit needs to be determined. In the case of ACTs, the deleterious effect of malaria infection in the
first trimester of pregnancy and the benefit of rapid and effective treatment needs to be taken into
consideration. Recent publications highlight first trimester malaria infections are more of a problem
Summary and Discussion
185
than previously thought, with two‐thirds of all malaria placental infections occurring in the first 12
weeks of pregnancy (without preventive interventions), the high associated risk of maternal anaemia
and impact on birthweight as well as a 3‐4 fold increased odds of miscarriage with malaria infections
in early pregnancy in an area of low and seasonal malaria transmission.[1,3,13,14] We found that
treatment with ACTs was much more common than the recommended first line treatment quinine in
the first trimester of pregnancy in the study area in western Kenya. It is unclear whether this was
due to inadvertent exposures because of unknown or undisclosed pregnancies, because of lack of
knowledge regarding the national treatment guidelines or whether this was a deliberate choice
because quinine was out of stock or that an ACT was considered a better option for these women.
Poor tolerability to quinine and common low compliance to the long 7‐day regimen are known
problems.[15,16] There is a clear need for improved recommendations regarding treatment of
malaria in early pregnancy but also a need to investigate suitability of quinine for case management
of uncomplicated malaria in pregnancy.
Before a change in the treatment policy for malaria in the first trimester of pregnancy is
recommended, there will be a need to monitor and address poor compliance with current malaria
treatment guidelines for early pregnancy. We found poor awareness of contra‐indication of ACTs in
early pregnancy among healthcare providers, drug outlet dispensers and women in the community
(chapters 6‐7). Clearer guidelines emphasising the need for pregnancy assessments of all women of
childbearing age seeking treatment for malaria should be developed. Pregnancy testing in the
community was found acceptable in our study which suggests that pregnancy tests offered by
trained healthcare professionals could be feasible. There are at least 2 RDTs which combine malaria
and pregnancy in one test, and one has an acceptable performance profile as reviewed in the last
round of malaria diagnostic product testing reported by WHO.[17] These could be considered for use
in female patients of childbearing potential. Implementability (including training, costs and
availability) of pregnancy tests at point of care would need to be determined. In all situations a
woman’s privacy and the confidentiality of her pregnancy status should be preserved by handling
records carefully and choosing a suitable place for pregnancy assessments. This is particularly
delicate when dealing with minors and adolescent girls who might be reluctant to disclose their
pregnancy. This requires an objective, accepting and professional attitude from the healthcare
provider.[18] The purpose of the pregnancy assessment should be made clear to the client such that
it is to ensure the adequacy and safety of the treatment. Pregnancy assessments should be
considered in all antimalarial dispensing scenarios with women of childbearing potential. This
includes drug outlets, both formal and informal, and community based strategies for case
management of malaria. Furthermore, there is renewed interest in mass drug administration for
malaria elimination which will need to consider strategies for management of potential pregnancies.
The lack of pregnancy assessment in women of childbearing potential has important implications for
risk management programmes for other medications contraindicated in pregnancy in this setting.
What are the prospects for sustainable pharmacovigilance systems for pregnant women in resource constrained settings? No single method can capture all desired data needed to make appropriate risk benefit assessments
of a drug used in pregnancy. A combination of different methods and data sources is the only
feasible approach to gather the most complete picture of potential developmental toxicity of a drug.
The different approaches presented in this thesis have their own strengths and limitations but
provide complimentary information. Common challenges include the need to obtain reliable and
Chapter 9
186
accurate exposure data. In their retrospective analysis, McGready et al included fewer than 100 well‐
documented early exposures to artemisinins after review of 25 years of data including over 48,000
pregnancies.[3] We found only 25% of first trimester ACT exposures could be confirmed (chapter 8)
and the sensitivity of women’s self‐report at the time of pregnancy follow up interviews was
estimated at 37% (based on confirmed exposures by the 2 other data sources, see chapter 8
Appendix).This indicates that studies or surveillance systems cannot solely rely on women self‐report
for antimalarial drugs, particularly if the recall period stretches over several months. Secondly, it is
very challenging to account for confounding by indication (i.e. the fact that the disease for which the
drug is being indicated, itself causes adverse pregnancy outcomes) in settings where self‐medication
of febrile episodes is common and where few women are treated with quinine.[19,20] Furthermore,
due to the nature of these observational studies, it is not possible to account for factors influencing
antimalarial treatment choice by healthcare providers and patients. If more severe cases are usually
treated with one type of drug rather than other, this could confound the association between
exposure and adverse pregnancy outcome. Inclusion of internal unexposed controls is necessary (as
opposed to using background rates which is commonly done with pregnancy exposure registries in
high income countries) as there is a lack of data on background rates of most adverse pregnancy
outcomes (particularly miscarriages and birth defects) in most malaria endemic areas. Nevertheless,
the use of internal unexposed controls allows a more appropriate comparison and provides the
potential to control for the underlying disease provided there is enough variability in the different
class of antimalarials used. Randomised controlled trials are considered the “gold standard” for
generating reliable evidence in clinical research and could address these methodological challenges.
Depending on the forthcoming findings from pooled data analysis, a randomised controlled trial
comparing ACTs to quinine for confirmed malaria in the first trimester of pregnancy may be
warranted to provide the level of reassurance needed for policy change. Whereas, a clinical trial was
considered unsuitable during the last WHO consultations to review evidence on the safety of
artemisinin derivatives in pregnancy in 2006 due to the limited data on first trimester exposure in
humans, we may have reached the point of equipoise. Equipoise is defined as a state of genuine
uncertainty regarding the comparative merits of different therapeutic choices and considered
essential for the ethical acceptability of clinical trials.[21] WHO guidelines for the treatment of
uncomplicated malaria in the first trimester of pregnancy state that an ACT is indicated if there is
uncertainty of compliance with a 7 day quinine regimen.[22] In areas where treatment with ACTs
already predominates quinine in the first trimester, a randomised controlled trial should be
considered ethical.
Most teratogenic safety signals have been identified by astute clinicians through reporting of case
series of abnormal pregnancy outcomes to exposed mothers.[23] Although this is less than ideal and
often takes several years for the signal to be detected, the potential of such reports shouldn’t be
ignored. However this requires awareness of pharmacovigilance and special considerations for
treatment of women of childbearing potential. We found this to be low among healthcare providers
and drug dispensers in our study area and this probably is not unique to rural western Kenya.
Integration of pharmacovigilance topics into university and professional curricula for all health
disciplines has been proposed.[24] The recent establishment of UMC Africa (a WHO Collaborating
Centre for Advocacy and Training in Pharmacovigilance for Africa) in Ghana provides an opportunity
to share specific tools and education materials for pharmacovigilance methods for pregnant
women.[25]
Summary and Discussion
187
Identification of sentinel sites able to capture reliable data on drug exposure and pregnancy
outcomes through a standard protocol as proposed in chapter 4 and recently by WHO would be
essential for safety signal detection and characterisation.[2] Such an approach could be used for the evaluation of a wide variety of medications that are used during pregnancy and estimation of the
risk of the spectrum of adverse pregnancy outcomes. However depending on the recruitment
strategy, assessment of miscarriages might not be feasible without introducing dedicated efforts to
detect these. Enrolling pregnant women attending for antenatal care is an interesting option, as
close to 90% of pregnant women in sub‐Saharan Africa attend antenatal clinics at least once.[26]
However as many women present for their first ANC visit after completion of the first trimester (the
average is around 22 weeks gestation), such an approach limits the possibility of studying early
miscarriages as an outcome. For this reason, prospective community‐based studies (as presented in
chapter 8) are needed for detection of early pregnancies or systems to incentivise women and/or
community health workers to refer pregnant women to present to ANC early should be explored.
Health and demographic surveillance sites make attractive candidates as they provide a framework
to identify women of childbearing age and usually have some level of health facility surveillance set
up. This is exemplified through the implementation of prospective observational cohorts to field‐test
active surveillance systems for identifying exposures to antimalarials during early pregnancy and for
monitoring pregnancy outcomes in three health and demographic surveillance sites in Africa
(Burkina Faso, Mozambique and Kenya as presented in chapter 8) as part of the Malaria in
Pregnancy consortium (MiPc).
Major limitations of traditional pregnancy exposure registries and prospective cohort studies include
costs and length of time required to achieve a desirable sample size. As an alternative, record
linkage studies using routine healthcare data provides a cost‐effective and relatively rapid means of
assessing the embryotoxic effect of a drug. Record linkage studies require minimal staff and
resources particularly in settings with electronic medical records. However, this necessitates sites
with comprehensive data on drug exposure and pregnancy outcomes. To minimize exposure
misclassification, it is essential to select sites within a relatively stable population where healthcare
is provided in a central location with a high proportion of health facility deliveries and limited
availability of the drug of interest outside the central health facility system. Sites where this
approach could be implemented effectively are few, and further study should assess suitability of
private agricultural and other industrial estates, mines and refugee camps as well as data from health
insurance schemes in malaria endemic countries.
How sustainable are these approaches? There are no obvious funding mechanisms for
pharmacovigilance activities. In resource constrained settings, priorities often lie with the delivery of
care and pharmacovigilance programmes currently have to compete for very scarce resources.[27]
However confidence in a public health programme can be seriously affected by adverse drug
reactions, particularly serious adverse events affecting pregnancy outcomes (such as pregnancy loss
or congenital anomalies).[28] Evidence on the safety of drugs implemented by public health
programmes has the potential to increase programme success by providing guidance on reducing
the risks of adverse drug reactions thereby increasing overall public trust.[29,30] The Roll Back
Malaria partnership issued guidance to countries applying for malaria funding for the inclusion of
pharmacovigilance component within donor‐supported programmes.[31] The Global Fund to fight
AIDS, tuberculosis and malaria has made provision for specific support towards pharmacovigilance
which could be used to fund such initiatives for antimalarials, antiretrovirals and tuberculosis
Chapter 9
188
treatments.[32] The United States' President's Malaria Initiative (PMI), is one of the major donors for
malaria control programmes in target countries in sub‐Saharan Africa and plays a key role in the
procurement of antimalarial treatment and assessment of drug safety. However, few proposals to
either funding bodies included a request for funding for pharmacovigilance activities.[33] In 2011,
the European Medicines Agency (EMA) approved Eurartesim (dihydroartemisinin‐piperaquine) for
marketing with the condition that a pregnancy registry be set up as part of the post‐marketing risk
management plan (among other recommendations).[34,35] This so far has resulted in the creation
of a European based pregnancy registry to monitor inadvertent exposures to Eurartesim in European
travellers, who rarely have access to Eurartesim, which is a treatment drug provided only to those
with clinical malaria. Little progress has been made with developing a platform to monitor the safety
of Eurartesim in pregnancy in malaria endemic countries where most of the exposures are likely to
occur.[36] Sustainability of the proposed approaches will need political commitment and key
stakeholders (including industry, governments and funders of public health programmes) buy‐in to
prioritise support for pharmacovigilance systems.
In conclusion, the reports presented in this thesis support the need and feasibility of implementing
various pharmacovigilance approaches for monitoring safety of medicines in pregnancy in resource
constrained settings. In light of the soaring number of first trimester pregnancies being exposed
inadvertently and deliberately to artemisinin derivatives, an update of the last 2006 WHO review of
the existing evidence is needed promptly to inform treatment guidelines for malaria in pregnancy.
Observational studies on ACT exposures in the first trimester of pregnancy have provided a degree
of reassurance and have not detected major teratogenicity signals in humans so far. However
numbers are still limited to less than a 1000 well documented first trimester exposures, and the rule
of 3 suggests that this can demonstrate a less than two‐fold increase in the overall incidence of
malformations.[37] Thus more research is needed, also to assess the risk of cardiovascular defects
associated with ACT exposures in early pregnancy as well as studies looking at congenital anomalies
detectable later in life including developmental delays. Furthermore, once enough data has accrued
to provide sufficient statistical power, it will be important to carry out a sensitivity analysis of the
suggested period of maximum artemisinin sensitivity in human embryos. With the accrued evidence
to date and the low adherence to the recommended treatment with quinine in first trimester of
pregnancy, a randomised controlled trial would be the fastest way to provide the level of evidence
needed. Until adequate evidence is available to comprehensively assess the risk‐benefit profile of
ACTs for early pregnancy, it will be important to minimise first trimester exposures. Current options
for treatment of pregnant women with malaria are few and appropriate prescribing and dispensing
is critical for rational drug use. As such healthcare providers and drug dispensers should have
relevant knowledge and skills regarding antimalarial drug use in women of childbearing potential.
Further research is needed to investigate the best approach to implement risk management to
minimise first trimester exposures including ways for appropriate dissemination of malaria
treatment guidelines, as well as pregnancy assessment strategies for healthcare professionals in
health facilities, drug dispensers working in the informal sector, and community health workers in
areas where their mandate includes treatment of malaria in women of childbearing age. This thesis
focused on treatment of malaria in pregnancy; however most of the findings are applicable to other
tropical diseases in pregnancy. The case of ACTs is not unique and most of the new medicines for
tropical diseases in pregnancy are not recommended for pregnancy due to lack of information.[38]
Given the high burden of disease in the tropics and the extensive overlap between malaria and HIV,
Summary and Discussion
189
it will be important to consider all concomitant drug exposures and potential for drug‐drug
interactions. Joint efforts are needed to tackle this paradoxical situation to enable safe and effective
treatment of tropical diseases in pregnancy.
References
1. Walker PG, Ter Kuile FO, Garske T, Menendez C, Ghani AC (2014) Estimated risk of placental infection and low birthweight attributable to Plasmodium falciparum malaria in Africa in 2010: a modelling study. Lancet Glob Health 2: e460‐e467.
2. Mehta U, Clerk C, Allen E, Yore M, Sevene E, et al. (2012) Protocol for a drugs exposure pregnancy registry for implementation in resource‐limited settings. BMC Pregnancy Childbirth 12: 89.
3. McGready R, Lee SJ, Wiladphaingern J, Ashley EA, Rijken MJ, et al. (2012) Adverse effects of falciparum and vivax malaria and the safety of antimalarial treatment in early pregnancy: a population‐based study. Lancet Infect Dis 12: 388‐396.
4. Deen JL, von Seidlein L, Pinder M, Walraven GE, Greenwood BM (2001) The safety of the combination artesunate and pyrimethamine‐sulfadoxine given during pregnancy. Trans R Soc Trop Med Hyg 95: 424‐428.
5. Adam I, Elhassan EM, Omer EM, Abdulla MA, Mahgoub HM, et al. (2009) Safety of artemisinins during early pregnancy, assessed in 62 Sudanese women. Ann Trop Med Parasitol 103: 205‐210.
6. Manyando C, Mkandawire R, Puma L, Sinkala M, Mpabalwani E, et al. (2010) Safety of artemether‐lumefantrine in pregnant women with malaria: results of a prospective cohort study in Zambia. Malar J 9: 249.
7. Rulisa S, Kaligirwa N, Agaba S, Karema C, Mens PF, et al. (2012) Pharmacovigilance of artemether‐lumefantrine in pregnant women followed until delivery in Rwanda. Malar J 11: 225.
8. Mosha D, Mazuguni F, Mrema S, Sevene E, Abdulla S, et al. (2014) Safety of artemether‐lumefantrine exposure in first trimester of pregnancy: an observational cohort. Malar J 13: 197.
9. Poespoprodjo JR, Fobia W, Kenangalem E, Lampah DA, Sugiarto P, et al. (2014) Dihydroartemisinin‐piperaquine treatment of multidrug resistant falciparum and vivax malaria in pregnancy. PLoS One 9: e84976.
10. Clark RL (2009) Embryotoxicity of the artemisinin antimalarials and potential consequences for use in women in the first trimester. Reprod Toxicol 28: 285‐296.
11. WHO, TDR (2007) Assessment of the safety of artemisinin compounds in pregnancy: report of two joint informal consultations convened in 2006. Geneva: WHO. WHO/GMP/TDR/Artemisinin/07.1. WHO/GMP/TDR/Artemisinin/07.1.
12. Clark RL, White TE, S AC, Gaunt I, Winstanley P, et al. (2004) Developmental toxicity of artesunate and an artesunate combination in the rat and rabbit. Birth Defects Res B Dev Reprod Toxicol 71: 380‐394.
13. Cottrell G, Mary JY, Barro D, Cot M (2007) The importance of the period of malarial infection during pregnancy on birth weight in tropical Africa. Am J Trop Med Hyg 76: 849‐854.
14. Huynh BT, Fievet N, Gbaguidi G, Dechavanne S, Borgella S, et al. (2011) Influence of the timing of malaria infection during pregnancy on birth weight and on maternal anemia in Benin. Am J Trop Med Hyg 85: 214‐220.
15. Achan J, Talisuna AO, Erhart A, Yeka A, Tibenderana JK, et al. (2011) Quinine, an old anti‐malarial drug in a modern world: role in the treatment of malaria. Malar J 10: 144.
16. Taylor WR, White NJ (2004) Antimalarial drug toxicity: a review. Drug Saf 27: 25‐61. 17. WHO (2013) Malaria Rapid Diagnostic Test Performance – Summary results of WHO product testing of
malaria RDTs: Round 1‐5 (2008‐2013). Geneva: WHO. 18. Perlman SE, Postlethwaite D, Stump S, Bielan B, Rudy SJ (2001) Taking a sexual history from and counseling
women on teratogenic drugs. J Reprod Med 46: 163‐168. 19. Chipwaza B, Mugasa JP, Mayumana I, Amuri M, Makungu C, et al. (2014) Self‐medication with anti‐
malarials is a common practice in rural communities of Kilosa district in Tanzania despite the reported decline of malaria. Malar J 13: 252.
20. Geissler PW, Nokes K, Prince RJ, Odhiambo RA, Aagaard‐Hansen J, et al. (2000) Children and medicines: self‐treatment of common illnesses among Luo schoolchildren in western Kenya. Soc Sci Med 50: 1771‐1783.
21. Freedman B (1987) Equipoise and the ethics of clinical research. N Engl J Med 317: 141‐145. 22. WHO (2010) Guidelines for the treatment of malaria. Second Edition. Geneva.
Chapter 9
190
23. Carey JC, Martinez L, Balken E, Leen‐Mitchell M, Robertson J (2009) Determination of human teratogenicity by the astute clinician method: review of illustrative agents and a proposal of guidelines. Birth Defects Res A Clin Mol Teratol 85: 63‐68.
24. Isah AO, Pal SN, Olsson S, Dodoo A, Bencheikh RS (2012) Specific features of medicines safety and pharmacovigilance in Africa. Ther Adv Drug Saf 3: 25‐34.
25. UMC‐A the Uppsala Monitoring Centre Africa. 26. van Eijk AM, Hill J, Alegana VA, Kirui V, Gething PW, et al. (2011) Coverage of malaria protection in
pregnant women in sub‐Saharan Africa: a synthesis and analysis of national survey data. Lancet Infect Dis 11: 190‐207.
27. Bakare N, Edwards IR, Stergachis A, Pal S, Holmes CB, et al. (2011) Global pharmacovigilance for antiretroviral drugs: overcoming contrasting priorities. PLoS Med 8: e1001054.
28. Brabin BJ, Warsame M, Uddenfeldt‐Wort U, Dellicour S, Hill J, et al. (2008) Monitoring and evaluation of malaria in pregnancy ‐ developing a rational basis for control. Malar J 7 Suppl 1: S6.
29. Dodoo A, Adjei S, Couper M, Hugman B, Edwards R (2007) When rumours derail a mass deworming exercise. Lancet 370: 465‐466.
30. Pal SN, Duncombe C, Falzon D, Olsson S (2013) WHO strategy for collecting safety data in public health programmes: complementing spontaneous reporting systems. Drug Saf 36: 75‐81.
31. Roll Back Malaria (2008) Steps for Completing Global Fund’s Round 8 Proposal for Requesting funds for Pharmacovigilance Geneva.
32. Xueref S, Daviaud J, Pal S (2013) The Global Fund and pharmacovigilance systems in resource‐limited settings. Lancet 381: 1360.
33. Stergachis A, Bartlein RJ, Dodoo A, Nwokike J, Kachur SP (2010) A situational analysis of pharmacovigilance plans in the Global Fund Malaria and U.S. President's Malaria Initiative proposals. Malar J 9: 148.
34. European Medicines Agency (2011) Assessment Report Eurartesim. London. 35. Ubben D, Poll EM (2013) MMV in partnership: the Eurartesim(R) experience. Malar J 12: 211. 36. Sigma‐Tau (2014) Eurartesim Pregnancy Registry. 37. European Medicines Agency (2008) Guidelines on risk assessment of medicinal products on human
reproduction and lactation: from data to labelling. London: EMA. 38. White NJ, McGready RM, Nosten FH (2008) New medicines for tropical diseases in pregnancy: catch‐22.
PLoS Med 5: e133.
191
Samenvatting
Samenvatting
192
Nauwkeurige informatie over de veiligheid van geneesmiddelen in de zwangerschap na de afronding
van de formele klinische studies is moeilijk te verkrijgen, vooral in ontwikkelingslanden waar de
infrastructuur voor pharmacovigilantie niet goed ontwikkeld is. In deze landen is er behoefte aan
een simpel maar effectief systeem om de veiligheid van geneesmiddelen gedurende de
zwangerschap te volgen. Hoewel artemisinine gebaseerde combinatietherapie (ACT) niet is
aanbevolen voor gebruik tijdens het eerste trimester van de zwangerschap, is onbedoelde
blootstelling onvermijdelijk. Er zijn op dit moment geen gevestigde methodes om de veiligheid van
deze geneesmiddelen tijdens de zwangerschap te volgen in landen waar malaria endemisch is. Het
onderzoek beschreven in dit proefschrift evalueert surveillance systemen voor medicijnen in landen
met beperkte middelen om een beter beeld te krijgen van het risico‐baten profiel van het gebruik
van artemisinine derivaten tijdens de vroege zwangerschap.
In hoofdstuk 2 wordt het gepubliceerde bewijs met betrekking tot de veiligheid van artemisinine
derivaten gebruik tijdens de zwangerschap geëvalueerd door middel van systematisch literatuur
onderzoek. Ten tijde van de evaluatie (november 2006) werden 14 relevante studies gevonden met
in totaal 945 vrouwen die waren blootgesteld aan artemisinine derivaten gedurende de
zwangerschap, waarvan 123 in het eerste trimester. De beperkte hoeveelheid informatie
suggereerde dat artemisinine derivaten in het tweede en derde trimester effectief en waarschijnlijk
veilig waren. Zeldzame bijwerkingen konden echter niet worden uitgesloten. Er was onvoldoende
informatie voor een afweging van het risicoprofiel tijdens het eerste trimester, de meest gevoelige
periode voor eventuele ongewenste bijwerkingen op het ongeboren kind. De bevindingen van het
literatuur onderzoek waren overeenkomstig een aanbeveling van de Wereldgezondheidsorganisatie
(WHO) in 2006. De noodzaak om een systeem op te zetten om blootstelling aan, en gevolgen van
artemisinine derivaten tijdens het eerste trimester te documenteren werd benadrukt door de WHO
in 2006.
In hoofdstuk 3 geven wij een schatting van het jaarlijks aantal zwangerschappen in gebieden met
malaria transmissie. Geografische gegevens van de transmissie van Plasmodium falciparum en
Plasmodium vivax werden gecombineerd met demografische gegevens van de Verenigde Naties om
de totale populatie in risico gebieden voor malaria in 2007 te inventariseren. Wij schatten dat er in
2007 125 miljoen zwangerschappen waren in malaria gebieden, waarvan ongeveer 60%
resulteerden in een levend geboren kind. Schattingen voor Afrika (32 miljoen) waren vergelijkbaar
met eerdere berekeningen van de Wereldgezondheidsorganisatie (25‐30 miljoen), maar schattingen
voor regio’s buiten Afrika waren veel hoger dan voorheen. Deze berekening van het aantal
zwangerschappen in malaria endemische gebieden was een eerste stap om het aantal
zwangerschappen in te kunnen schatten dat jaarlijks wordt blootgesteld aan malaria en dus het
aantal eventuele behandelingen met antimalariamiddelen.
In hoofdstuk 4 berekenen we dat een embryo een kans heeft van 12 procent om onbedoeld aan
artemisinine derivaten blootgesteld te worden gedurende de 6 weken die het meest gevoelig zijn
voor mogelijke schadelijke effecten van artemisinine derivaten (6‐12 weken na de laatste
menstruatie). Deze informatie is nuttig om de omvang van de kans op blootstelling aan
antimalariamiddelen gedurende de zwangerschap te begrijpen. In dit hoofdstuk worden verder de
methodologische overwegingen beschreven voor een systematische beoordeling van de uitkomst
van zwangerschap en aangeboren afwijkingen bij vrouwen die vroeg tijdens de zwangerschap zijn
bloot gesteld aan antimalariamiddelen. Tevens worden verschillende methoden besproken hoe men
193
blootstelling aan medicijnen kan vast stellen, de keuze van mogelijke controle groepen en de
steekproefgrootte. Voor tijdige beoordeling van het risicoprofiel van antimalariamiddelen wordt een
doelgerichte prospectieve aanpak voorgesteld door middel van de oprichting van een internationaal
register voor blootstelling aan antimalariamiddelen. De Wereldgezondheidsorganisatie publiceerde
recentelijk een beschrijving voor het opzetten van landelijke registers om blootstelling aan
medicijnen tijdens de zwangerschap vast te leggen in gebieden met beperkte middelen. Dit biedt
mogelijkheden om gestandaardiseerde methodes te gebruiken bij het verzamelen van gegevens. [2]
In hoofdstuk 5, worden de bevindingen beschreven van een verkennende studie naar de
haalbaarheid van het gebruik van routinematig verzamelde gegevens over antimalariamiddelen in
registerboeken van algehele en verloskunde poliklinieken en over de uitkomst van bevallingen in
registers van verloskamers om de veiligheid van artemisinine derivaten tijdens het begin van de
zwangerschap te beoordelen. Deze studie werd verricht in een medische missie‐post in Mlomp, in
het zuidwesten van Senegal. De informatie van de verschillende registers werden gekoppeld op basis
van een statistische methode die gebruikt maakt van ‘probabilistische matching’ om
corresponderende gegevens van dezelfde patiënt te vinden in de verscheidene registers op basis van
een naam, leeftijd, woonplaats en dergelijke. De bevindingen lieten zien dat het koppelen van
routinematig verzamelde gezondheidszorggegevens via registers haalbaar is. Dit type gegevens zou
een belangrijke bijdrage kunnen leveren aan de aantallen zwangerschappen die nodig zijn om
zekerheid te geven dat het gebruik van artemisinine derivaten gedurende het eerste trimester van
de zwangerschap veilig is.
In hoofdstuk 6 worden de resultaten beschreven van een studie naar de belevingen, overtuigingen
en het gedrag met betrekking tot aangeboren afwijkingen en miskramen in het westen van Kenia. De
studie maakte gebruik van tien focusgroepdiscussies. Het gebrek aan informatie over de oorzaken
van miskramen en aangeboren afwijkingen kon leiden tot het stigmatiseren van de moeder en kon
verder het zoeken naar hulp en betere zorg belemmeren. Dit kon wrede gevolgen hebben voor
kinderen met aangeboren afwijkingen die vaak verborgen werden gehouden voor de gemeenschap
en geen toegang hadden tot zorg die de kwaliteit van hun leven zou kunnen verbeteren. Sommige
vrouwen die zich bewust waren van de potentiële gevaren van geneesmiddelen gebruik tijdens de
zwangerschap waren zich niet bewust van welke medicijnen veilig waren. Meestal werd een risico
gezien als een medicijn zonder voorschrift werd gebruikt of waarbij de aanbeveling van de clinicus
niet werd gevolgd. Dit benadrukt het belang van educatie materiaal voor zowel de gemeenschap als
zorgverleners met betrekking tot de keuze van medicijnen voor behandeling van aandoeningen
tijdens de zwangerschap. Deze bevindingen benadrukken tevens de noodzaak van voorlichting over
de potentiële oorzaak van slechte zwangerschapsuitkomsten en betere informatie over de
beschikbaarheid van zorg.
In hoofdstuk 7 beschrijven we de resultaten van een studie in west Kenia naar de kennis en naleving
van de nationale behandelingsrichtlijnen voor ongecompliceerde malaria tijdens de zwangerschap.
De studie werd uitgevoerd in klinieken en een willekeurige selectie van winkeltjes (‘Dukas’) die
geneesmiddelen verkochten. Vrouwen in de vruchtbare periode (15 tot 49 jaar) werden
geïnterviewd wanneer ze de kliniek verlieten. Informatie van de winkeltjes werd verzameld door
gebruik te maken van gesimuleerde patiënten die zich voordeden als vrouwen met een vroege
zwangerschap of als familie van vrouwen die in hun derde trimester waren. De studie bracht
onvoldoende kennis en onjuiste voorschrijfpraktijken aan het licht voor de behandeling van malaria
Samenvatting
194
tijdens de zwangerschap, vooral in het eerste trimester en in de informele sector. Bijvoorbeeld,
slechts 44% van de vrouwen in de klinieken en 7% van de simulatie patiënten werd gevraagd of ze
eventueel zwanger zouden kunnen zijn alvorens een geneesmiddel voor te schrijven. Dit was
tegenstrijdig met de bevinding dat 93% en 49% van de personen in de klinieken en winkeltjes
rapporteerden dat zij routinematig vroegen of een vrouw in de vruchtbare periode mogelijk zwanger
zou kunnen zijn. Het voorschrijven van het juiste medicijn, in de correcte dosering voor de
behandeling van malaria tijdens het eerste trimester werd waargenomen bij 32% van alle gevallen in
klinieken en bij 0% in de winkeltjes. Kennis was matig bij het personeel in de kliniek (56%) en afwezig
in de winkeltjes. Voorschrijven van artemether‐lumefantrine (een ACT) in het eerste trimester
gebeurde in 16% en 51% van de patiënten in de klinieken en winkels en dit was niet een gevolg van
een voorraadtekort aan kinine. Dergelijk onjuist voorschrijfgedrag kan ernstige gevolgen hebben
voor de zwangere patiënt, niet alleen omdat medicijnen werden gebruikt met een onduidelijk
veiligheidsprofiel (artemisinine derivaten), maar ook omdat ineffectieve medicijnen werden
voorgeschreven die niet geschikt zijn voor de behandeling van malaria of effectieve medicijnen,
maar in een te lage of te korte dosis (zoals bij kinine).
In hoofdstuk 8 worden de bevindingen van een prospectieve studie in het westen van Kenia
gepresenteerd om de associatie te bestuderen tussen miskramen en artemisinine derivaten gebruik
tijdens de vroege zwangerschap. De studie werd verricht in een gebied met een goed ontwikkeld
demografische surveillance informatiesysteem waar de gezondheidsstatus van de bevolking werd
gevolgd door wekelijkse (of twee keer per maand) huisbezoeken te doen en door surveillance in de
kliniek. Surveillance in de gemeenschap werd gebruikt om vroege zwangerschappen bij vrouwen in
de vruchtbare leeftijd te identificeren. Meerdere informatie bronnen werden gecombineerd om
blootstelling aan artemisinine derivaten tijdens de zwangerschap te identificeren en te verifiëren.
Deze studie liet zien dat vaststelling van zwangerschap in het eerste trimester (voordat de meeste
vrouwen een verloskundige bezoeken) met het gebruik van zwangerschapstesten in de
gemeenschap haalbaar was. Hieruit kon het percentage miskramen per zwangerschapsweek bepaald
worden. Het totale risico op een miskraam werd ingeschat op 19%. Van de 1,126 zwangerschappen
waren er 42 (4%) met bevestigde blootstelling aan artemisinine derivaten tijdens de embryo‐
gevoelige periode (6‐12 weken na de laatste menstruatie) en 75 (7%) tijdens het eerste trimester (0‐
13 weken). Zwangerschappen met malaria en blootstelling aan artemisinine derivaten in de embryo‐
gevoelige periode hadden geen statistisch significant verschillend risico op miskramen vergeleken
met zwangerschappen zonder blootstelling (risico ratio 0.85, 95% betrouwbaarheidsinterval 0.22‐
3.33). Een verhoogd risico door blootstelling in het eerste trimester kon niet worden uitgesloten
(risico ratio 1.60, 95% betrouwbaarheidsinterval 0.70‐3.68), maar de bovengrens van het 95%
betrouwbaarheidsinterval suggereert dat een meer dan 3,7‐voudige verhoogd risico uiterst
onwaarschijnlijk is. In lijn met hoofdstuk 7 werd gevonden dat kinine, het aanbevolen middel voor
behandeling tijdens het eerste trimester, weinig werd gebruikt (n=34 met 22 blootgesteld aan zowel
kinine als artemisinine derivaten). Dit belemmerde ons vermogen om het eventuele effect van
artemisinine derivaten in het eerste trimester te vergelijken met kinine. De resultaten waren echter
in overeenstemming met recente bevindingen van studies aan de Thai‐Birmese grens die ook geen
verhoogd risico vonden.[3] Binnenkort zullen tevens de bevindingen beschikbaar zijn van een
gemeenschappelijke data‐analyse waarbij de informatie uit deze en twee andere studies worden
gecombineerd. Dit zal verder inzicht geven in de risico’s en voordelen van het gebruik van
artemisinine derivaten in het eerste trimester.
195
Concluderend kan worden gesteld dat het onderzoek beschreven in dit proefschrift de noodzaak
benadrukt voor betere pharmacovigilantie van antimalariamiddelen tijdens de zwangerschap in
gebieden waar beperkte middelen voorhanden zijn zoals in Afrika. Geen enkele methode alleen is
toereikend om de gewenste informatie vast te leggen die nodig is voor een passende analyse van het
risicoprofiel van antimalariamiddelen. Een combinatie van verschillende methodes en bronnen van
informatie is nodig om een compleet beeld te krijgen voor het opsporen, evalueren en voorkomen
van eventuele ongewenste bijwerkingen van deze geneesmiddelen tijdens de zwangerschap. Een
geschikte studielocatie is essentieel waar betrouwbare gegevens zouden kunnen worden verkregen
over zowel de blootstelling aan geneesmiddelen als de uitkomst van de zwangerschap. Mogelijke
opties zijn gebieden waar gezondheidsstatus en demografische gegevens routinematig worden
bijgehouden zoals onderzoek locaties met demografische surveillancesystemen, agrarische of
industriële complexen, vluchtelingen kampen of via ziektekostenverzekering bedrijven die werkzaam
zijn in gebieden met malaria. Wij vonden een onverwacht hoog aantal zwangerschappen die
onbedoeld of opzettelijk aan artemisinine derivaten waren blootgesteld tijdens de vroege
zwangerschap in west Kenia. Conform het werk van andere groepen hebben deze en andere studies,
tot nu toe geen ernstige teratogene afwijkingen aan het licht gebracht. De aantallen zijn echter nog
beperkt, en in totaal zijn minder dan 1000 goed gedocumenteerde blootstellingen in het eerste
trimester beschreven wereldwijd. Dit geeft een zekere mate van vertrouwen in de veiligheid van de
artemisinine derivaten voor gebruik tijdens de vroege zwangerschap, maar de regel‐van‐drie
suggereert dat dit nog steeds een minder dan tweevoudige toename in the incidentie van
aangeboren afwijkingen kan inhouden. Meer onderzoek is dus nodig, ook om het risico op
cardiovasculaire afwijkingen in associatie met blootstelling aan artemisinine derivaten in de vroege
zwangerschap vast te stellen en om te kijken naar aangeboren afwijkingen die pas later in het leven
duidelijk zijn, zoals een ontwikkelingsachterstand. Ons onderzoek laat ook zien dat de aanbieders
van zorg en medicijnen in het westen van Kenia beperkte kennis hebben met betrekking tot het
veilig gebruik van antimalariamiddelen door vrouwen in de vruchtbare leeftijd. Met het tot nu toe
opgebouwde bewijs en de slechte naleving van de huidige aanbevolen behandeling met kinine in het
eerste trimester van de zwangerschap zou een gerandomiseerd klinisch onderzoek de snelste weg
zijn om het bewijs te verkrijgen dat noodzakelijk is. Deze studie was onderdeel van een multi‐landen
initiatief en het is belangrijk dat alle nieuwe gegevens nu zo snel mogelijk gedeeld kunnen worden
met de Wereldgezondheidsorganisatie voor het bijwerken van het risico‐baten profile van de
artemisinine derivaten tijdens de zwangerschap. Totdat er voldoende gegevens beschikbaar zijn om
een volledige kosten‐batenanalyse te maken van het gebruik van de artemisinine derivaten tijdens
de vroege zwangerschap, zal het belangrijk zijn om eerste trimester blootstellingen te minimaliseren.
Daarvoor is verder onderzoek nodig om de beste methode te bepalen ter verbetering van het
uitvoeren van de nationale malaria behandelingsrichtlijnen, vooral onder de
geneesmiddelenverkopers in de informele sector. De bevindingen in dit proefschrift zijn tevens van
toepassing voor de behandeling van andere tropische ziekten. De situatie met artemisinine
derivaten is niet uniek en het merendeel van de nieuwe medicijnen voor tropische ziekten die op de
markt komen worden niet aanbevolen voor gebruik tijdens de zwangerschap. Vaak is dit niet
gebaseerd op gegevens over onveiligheid, maar eerder over het gebrek aan informatie over
veiligheid in de zwangerschap, met als gevolg dat zwangere vrouwen vaak jarenlang moeten
toedoen met inferieure geneesmiddelen zoals kinine. Gezamenlijke inspanningen zijn nodig om deze
situatie aan te pakken zodat veilige en effectieve behandeling van tropische ziektes gedurende de
zwangerschap mogelijk is.
196
197
AcknowledgementsThe papers presented in this thesis are the result of joint efforts involving many people across
multiple institutions without whom the work could not have been successfully completed. I would
like to thank all study participants for their time and cooperation and all the staff involved in the
studies presented in this thesis.
Foremost, I would like to express my sincere gratitude to Feiko ter Kuile who has been an amazing
supervisor and played a pivotal role in the work that went into this thesis. I have learned a great deal
through his rigour, high scientific standards and integrity. He was always there if I needed him and all
his advices have been invaluable. I thank him for always driving me to go the extra mile, for making
time when he had very little, for considering “happiness levels” every step of the way and for making
work fun. To work with him over the past 7 years has been a real pleasure and a privilege.
I am grateful to Michael Boele van Hensbroek for accepting to be my second promotor and for his
help and guidance in the last few months of the production of this thesis.
I am most grateful to the Malaria in Pregnancy Consortium (MiPc) secretariat at the Liverpool School
of Tropical Medicine (LSTM), particularly Alison Reynolds for her friendship, great logistical support
and for proof reading part of this thesis. My deepest thanks to Jenny Hill and Jayne Webster for
creating the opportunity for me to work with them on the interesting MiPc implementation research
studies in Kenya and for the good times in Kweisos. Jenny also for the inspiring discussions while we
shared an office in Liverpool (which played an important role in my move to Kenya) and for keeping
me on track to meet PhD deadlines. Andy Stergachis and the MiPc ASAP team (Esperanca Sevene,
Tinto Halidou, Umberto d’Alessandro, Laura Sangare, Becky Bartlein, Gillian Levine and Greg Calip)
made important contributions to the work presented in chapter 8, I have greatly enjoyed our
collaboration in this stimulating work. My sincere thanks also to Andy for his support throughout the
MiPc years and the valuable input to study protocols and co‐authored papers. Special thanks to Rob
Nathan for his assistance with ultrasound quality control and for reviewing the numerous scans,
Stephen Rulisa and Mark Londema for training study nurses to do obstetric ultrasounds. At LSTM: I
thank Cheryl Pace and James Smedley for taking over the MiPc Central Safety Database and doing an
excellent job, and Helen Wong for her efficient administrative support. I am grateful to Bernard
Brabin for his encouragement and for the captivating malaria discussions.
I would like to thank the Malaria Branch, Centers of Disease Control and Prevention, Atlanta, for
financial and technical support for the studies in Kenya. Particularly, I am extremely grateful to
Meghna Desai, my co‐supervisor, who gave tremendous helpful advice on the fieldwork in Kenya
and for giving me the opportunity to be involved and learn from various, very interesting, projects
through the KEMRI/CDC collaboration. Her sense of humour and composure when things didn’t go
to plan were highly appreciated. I would like to thank Mary Hamel, without whom the EMEP study
(chapter 8) might not have seen the light of day. Her interest in this project, her hospitality during
my first visits in Kenya and her input to protocol development were invaluable. I thank Danny Feikin
and Deron Burton for making the collaboration with the IEIP program not only possible but
effortless. I also thank Meredith McMorrow for training the study nurses on Ballard score. My
sincere thanks go to Kayla Laserson, the KEMI/CDC director at the time of EMEP study (chapters 6 &
8), for her support throughout my time in Kisumu. Her enthusiasm, energy and dedication have been
198
truly inspiring. I am very grateful to Larry Slutsker for his support and constructive feedback on study
proposals, protocols and manuscripts.
I will be eternally grateful to Professors Brian Greenwood and Daniel Chandramohan at the London
School of Hygiene and Tropical Medicine and Jamie Robinson and Susan Hall at GlaxoSmithKline for
enabling a fellowship year whilst I was very new to malaria research and grant proposals.
To all my colleagues at KEMRI/CDC, Kisumu, Kenya: Peter Ouma, Abraham Katana, Florence Achieng,
Roseline Odera, George Aol, Godfrey Bigogo, Allan Audi, Frank Odhiambo and others from the
malaria and IEIP branches which are too many to cite ‐ thanks for making me feel welcome from the
day I arrived in Kisumu and for guiding me through local systems and situations. Special thanks go to
Dr Simon Kariuki, head of malaria branch, and Dr Vulule, director of the KEMRI research station in
Kisumu at the time, for their support throughout my time in Kisumu.
The work presented in chapters 6 & 8 was possible because of the dedication and hard work of the
EMEP study staff. I feel truly lucky to have worked with such a remarkable team, and I am
particularly grateful to Emily Ayanga, Everlyne Oteyo, Jane Oiro, Theresa Aluoch, Elizabeth Aballa,
Faith Samo, Eric Onyango and Joshua Auko as well as the 40 village reporters from Asembo and to
Tina Oneko for her careful review of newborns with suspected congenital anomalies‐ each team
member was invaluable. My deepest gratitude to George Aol for his guidance and wisdom
throughout the study which were instrumental to understand what would be culturally acceptable
and feasible. This study could not have happened without the support and collaboration of the IEIP,
MNHCU and STOPMIP teams‐ thank you all.
Friends and colleagues who have contributed positively to my life in Kisumu: Nelli and Matt
Westercamp, Alkesh Patel, Maria Oziemkowska, Tina Oneko, Penny Phillips‐Howard, Gemma Aellah,
Kurt and Amy Herman‐Roloff, Martin Mwangi, Ellen van Puffelen, Bobby Ochieng, Christina Polyak,
Titus Kwambai, Kat Tumlinson, Marieke Sassen, Erwan Piriou, Anja van’t Hoog, Josephine Walker,
Kelly Alexander, Jo Halliday, Daryn Knobbel and Nat Robinson‐ thank you for all the good memories.
My heartfelt thanks go to Professor Philippe Brasseur and Soeur Marie Joelle for their generous
support and for hosting us during our trip to Dakar and Mlomp, in Senegal (chapter 5).
Julie Gutman, Christina Riley and Ann Buff were critical to the successful completion of the cross‐
sectional survey presented in chapter 7. This has been a very conducive collaboration and I thank
them for putting up with sudden deadlines over the last few months.
Huge thanks to Annemieke van Eijk for doing the Dutch translation of the summary and to Olaya and
Eloisa Astudillo for designing the cover page of this thesis.
Last but not least, I would like to thank my family for their love and encouragement. My parents
Oswald and Dominique, who have supported me enthusiastically in all my undertakings and gave me
the thirst of travel. My sisters Savina and Ondine, for always being there for me. My daughter Lea,
for putting things in perspective and never failing to put a smile on my face. And especially to my
considerate and patient partner Per who has been an amazing support through all stages of this
PhD. I thank him for not only putting up with my moves and travels but also for joining me for part of
the adventures. Words cannot express how grateful I am for his help and encouragement,
particularly in the last few months, which carried me through to the end.
199
AbouttheauthorStephanie Dellicour was born in Belgium on December 4, 1980. In 1998, she finished her secondary
education at the European School in Brussels where she obtained a European Baccalaureate. She
subsequently moved to South Africa where she completed a BSc in Biochemistry and Genetics in
2001 and a Bsc Honours in Pharmacology in 2002 from the University of Cape Town. She obtained an
MSc in Epidemiology from the London School of Hygiene and Tropical Medicine (LSHTM) in 2004.
Her Msc dissertation was in pharmacoepidemiology and involved using the UK General Practice
Research Database (GPRD) to assess drug‐induced autoimmune haemolytic anaemia. She worked for
GlaxoSmithKline (GSK) from 2004 to 2005 in their pharmacoepidemiology department where she
first became interested in the issues surrounding the safety concerns for antimalarial drugs used in
pregnancy. At the time GSK was developing a risk management plan for potential exposures in early
pregnancy to CDA (chlorproguanil‐dapsone‐artesunate combination) which was in the pipeline. In
2005, she received a fellowship through GSK to conduct research at LSHTM with Professors
Greenwood and Chandramohan to investigate options for setting up studies assessing the safety of
antimalarials in the first trimester of pregnancy. In 2007, she joined the Liverpool School of Tropical
Medicine to work with the Malaria in Pregnancy consortium (MiPc) under Professor ter Kuile. There
she worked in the Safety Working Group coordinating pharmacovigilance cross‐cutting activities
within the MiPc and developing grant proposals for the work presented in this thesis. In 2009 she
moved to Kenya to coordinate studies for the MiPc at the Kenya Medical Research Institute (KEMRI)/
Centers for Disease Control and Prevention (CDC) Research and Public Health Collaboration. There
she coordinated implementation research studies focused on interventions for the control of malaria
in pregnancy. Concurrently she developed, implemented and coordinated the research project on
the pharmacovigilance system for monitoring the safety of antimalarial used in early pregnancy
throughout the 3.5 years of fieldwork. She moved back to the UK in 2013 to write up the work
coordinated in Kenya.
Additionalpublicationsbytheauthor:1. Dellicour S, Greenwood B (2007) Systematic review: Impact of meningococcal vaccination on pharyngeal carriage of meningococci. Trop Med Int Health 12: 1409‐1421.
2. Brabin BJ, Warsame M, Uddenfeldt‐Wort U, Dellicour S, Hill J, et al. (2008) Monitoring and evaluation of malaria in pregnancy ‐ developing a rational basis for control. Malar J 7 Suppl 1: S6.
3. Desai M, Dellicour S (2012) Effects of malaria and its treatment in early pregnancy. Lancet Infect Dis 12: 359‐360.
4. Hill J, Dellicour S, Bruce J, Ouma P, Smedley J, et al. (2013) Effectiveness of antenatal clinics to deliver intermittent preventive treatment and insecticide treated nets for the control of malaria in pregnancy in Kenya. PLoS One 8: e64913.
5. Kwambai TK, Dellicour S, Desai M, Ameh CA, Person B, et al. (2013) Perspectives of men on antenatal and delivery care service utilisation in rural western Kenya: a qualitative study. BMC Pregnancy Childbirth 13: 134.