Pediatrics 2014 Weisz e1024 46
-
Upload
karyn13692 -
Category
Documents
-
view
214 -
download
0
description
Transcript of Pediatrics 2014 Weisz e1024 46
-
PDA Ligation and Health Outcomes: A Meta-analysis
abstractBACKGROUND AND OBJECTIVE: Patent ductus arteriosus (PDA) liga-tion has been variably associated with neonatal morbidities andneurodevelopmental impairment (NDI). The objective was to system-atically review and meta-analyze the impact of PDA ligation in preterminfants at ,32 weeks gestation on the risk of mortality, severeneonatal morbidities, and NDI in early childhood.
METHODS: Medline, Embase, Cochrane Central Register of ControlledTrials, Education Resources Information Centre (ERIC), Cumulative In-dex to Nursing and Allied Health (CINAHL), PsycINFO, and the Disserta-tion database were searched (1947 through August 2013). Risk of biaswas assessed by using the Newcastle-Ottawa Scale and the CochraneRisk of Bias tool. Meta-analyses were performed by using a random-effects model. Unadjusted and adjusted odds ratios (aORs) with 95%condence intervals (CIs) were pooled when appropriate.
RESULTS: Thirty-nine cohort studies and 1 randomized controlled trialwere included. Nearly all cohort studies had at least moderate risk ofbias mainly due to failure to adjust for survival bias and importantpostnatal preligation confounders such as ventilator dependence,intraventricular hemorrhage, and sepsis. Compared with medicaltreatment, surgical ligation was associated with increases in NDI(aOR: 1.54; 95% CI: 1.012.33), chronic lung disease (aOR: 2.51; 95%CI: 1.983.18), and severe retinopathy of prematurity (aOR: 2.23; 95%CI: 1.623.08) but with a reduction in mortality (aOR: 0.54; 95% CI:0.380.77). There was no difference in the composite outcome ofdeath or NDI in early childhood (aOR: 0.95; 95% CI: 0.581.57).
CONCLUSIONS: Surgical ligation of PDA is associated with reducedmortality, but surviving infants are at increased risk of NDI. However,there is a lack of studies addressing survival bias and confounding byindication. Pediatrics 2014;133:e1024e1046
AUTHORS: Dany E. Weisz, MD,a,b Kiran More, MD,b Patrick J.McNamara, MD, MSc,b,c,d and Prakesh S. Shah, MD, MSce,f
aDepartment of Newborn and Developmental Pediatrics,Sunnybrook Health Sciences Center, Toronto, Canada;bDepartment of Pediatrics, Hospital for Sick Children, Toronto,Canada; cDepartment of Physiology and fInstitute of Health Policy,Management and Evaluation, University of Toronto, Toronto,Canada; dPhysiology and Experimental Medicine Program,Hospital for Sick Children Research Institute, Toronto, Canada;and eDepartment of Pediatrics, Mt Sinai Hospital, Toronto,Canada
KEY WORDSpatent ductus arteriosus, neurodevelopmental impairment,death, chronic lung disease, retinopathy of prematurity, cerebralpalsy, cognitive impairment, mortality
ABBREVIATIONSaORadjusted odds ratioCIcondence intervalCLDchronic lung diseaseGAgestational ageIVHintraventricular hemorrhageNDIneurodevelopmental impairmentNECnecrotizing enterocolitisNSAIDnonsteroidal antiinammatory drugPDApatent ductus arteriosusRCTrandomized controlled trialROPretinopathy of prematurityVCPvocal cord paresis
Dr Weisz conceptualized and designed the study, acquired andinterpreted the data, drafted the initial manuscript, and revisedthe manuscript; Dr More revised the protocol, acquired andinterpreted the data, and revised the manuscript; Dr McNamararevised the protocol, interpreted the data, and revised themanuscript; Dr Shah conceptualized and designed the study,revised the protocol, interpreted the data, and revised themanuscript; and all authors approved the nal manuscript.
This systematic review has been registered with PROSPERO(international database of prospectively registered systematicreviews) (identier CRD42013005390).
www.pediatrics.org/cgi/doi/10.1542/peds.2013-3431
doi:10.1542/peds.2013-3431
Accepted for publication Dec 19, 2013
Address correspondence to Dany E. Weisz, MD, Department ofNewborn and Developmental Paediatrics, Sunnybrook HealthSciences Centre, 2075 Bayview Ave, Toronto, ON, Canada M4N 3M5.E-mail: [email protected]
PEDIATRICS (ISSN Numbers: Print, 0031-4005; Online, 1098-4275).
Copyright 2014 by the American Academy of Pediatrics
FINANCIAL DISCLOSURE: The authors have indicated they haveno nancial relationships relevant to this article to disclose.
FUNDING: No external funding.
POTENTIAL CONFLICT OF INTEREST: The authors have indicatedthey have no potential conicts of interest to disclose.
e1024 WEISZ et al at Indonesia:AAP Sponsored on June 7, 2015pediatrics.aappublications.orgDownloaded from
-
Patent ductus arteriosus (PDA) occurs innearly 50% of preterm infants born at,32 weeks gestation.1 The PDA shuntsblood away from the descending aortainto the pulmonary artery, resulting insystemic hypoperfusion and pulmonaryovercirculation.2,3 It is considered a sig-nicant precursor to mortality andmorbidity in extremely preterm neo-nates, including congestive heart failure,intraventricular hemorrhage (IVH), nec-rotizing enterocolitis (NEC), prolongedventilator dependency, and chronic lungdisease (CLD).1,4 Methods to close orminimize the effects of a clinically sig-nicant PDA include conservative man-agement (eg, uid restriction, diuretics,ventilation strategies), cyclooxygenaseinhibitors (eg, indomethacin or ibupro-fen), acetaminophen, or surgical liga-tion.5,6 Surgical ligation is usually onlyconsidered whenmedical treatment haseither failed or was contraindicated.7
Several investigators have studied PDAligation and its association with neo-natal mortality, short-term morbidity,and neurodevelopmental impairment(NDI) in early childhood.Mireaetal8 andClyman et al9 found increased reti-nopathy of prematurity (ROP) and CLDin infants treated with ligation com-pared with nonsurgically treatedinfants. Two large studies found thatextremely preterm infants who weretreated with surgical ligation had sig-nicantly higher odds of NDI at 2-yearfollow-up,10,11 although a third studyshowed no increased risk.12
In light of these concerns and some-what conicting results, themerits andsafety of ligation have been questioned,with increasing uncertainty about ap-propriate patient selection and theoptimal timing for surgery to minimizemorbidity.1316 This uncertainty hasbeen associated with a secular trendtoward a permissive approach to thePDA.1719 However, few of the studiesperformed to date have controlled forconfounding by indication, that infants
referred for surgical ligation mayrepresent a more ill cohort and maybe at higher risk of NDI. As a result,there is a paucity of relevant, reliablestudies to guide the PDA ligation de-cision.
Our objective was to systematicallyreview and meta-analyze the impact ofPDA ligation in preterm infants at,32weeks gestation on the risk of mor-tality, severe neonatal morbidities, andNDI.
METHODS
We conducted and reported this reviewfollowing the Preferred Reporting Itemsfor Systematic Reviews and Meta-analyses guidelines.20 The protocol forthis review was registered with PROS-PERO, the international prospectiveregister of systematic reviews (http://www.crd.york.ac.uk/NIHR_PROSPERO,identier CRD42013005390).
Types of Studies
Randomized controlled trials (RCTs)and case-control or cohort studieswitha comparator group were included ifpublished in the form of an originalresearch manuscript in a peer-reviewed journal, an abstract in con-ference proceedings, or a dissertation.Narrative reviews, letters, editorials,and commentaries were excluded butread to identify potential studies. Du-plicate reports not providing additionalinformation were excluded. Cross-sectional studies, case reports andseries, qualitative studies, review arti-cles, and studies that did not reportmethods were excluded but read toidentify potential studies. Studies ininfantswith a PDA that did not report onoutcomes in infants who underwentsurgical ligation were excluded.
Types of Participants
Preterm infants born at a gestationalage (GA),32weekswith a clinical and/or
echocardiographic diagnosis of PDAwere included. Studies in which someinfants were born$32 weeks GA wereincluded if.80% of infants in the studypopulation had a GA ,32 weeks. Thediagnosis of PDA was made on the ba-sis of clinical suspicion and/or echo-cardiography. Echocardiography diagnosiswas preferable but not mandatory forinclusion in this review.
Exposure and Comparison
Studies must have included and com-pared, at aminimum,a surgically versusmedically treated group or subgroup.Surgical subgroups included thoseinfants treated with pharmacotherapyfollowed by surgical ligation (pharma-cotherapy and ligation subgroup, in-dicating treatment with a nonsteroidalantiinammatory drug [NSAID] or acet-aminophen, followed by surgery) orthose infants who underwent primaryligation (primary ligation subgroup,indicating surgical ligation withoutpreceding pharmacotherapy). Medicalsubgroups included infants who re-ceived pharmacotherapy only (phar-macotherapy only subgroup, indicatingtreatment with NSAIDs or acetamino-phen but not surgery) and those treatedwithout surgery or pharmacotherapy(conservative subgroup). Conservativemanagement referred only to watchfulobservation and the use of uid re-striction, diuretics, digoxin, and/ormechanical ventilation adjustment tomanage the PDA shunt but excludedNSAID/acetaminophen and surgicaltreatments. Infants in the surgical orpharmacotherapy groups may havebeen treated with conservative mea-sures initially before their respectivedenitive treatment.
Surgical ligation must have been per-formed before 40 weeks corrected GA,either at the bedside or in the operat-ing room, via a left lateral thoracotomyusing a clip or ligature and not inconjunction with another surgery.
REVIEW ARTICLE
PEDIATRICS Volume 133, Number 4, April 2014 e1025 at Indonesia:AAP Sponsored on June 7, 2015pediatrics.aappublications.orgDownloaded from
-
Outcomes
Studies were included if they reportedat least 1 of the following outcomes forany treatment groups or subgroups:
1. death before discharge;
2. CLD, dened as the need for anyform of respiratory support (oxygenor positive-pressure support) at 36weeks corrected GA or at the timeof transfer to a step-down unit21;
3. severe ROP, dened as stage $3according to the international clas-sication,22 or stage 2 with plusdisease requiring treatment with la-ser therapy or vascular endothelialgrowth factor inhibitor;
4. NDI in early childhood (1548months corrected GA), dened as acomposite of at least 1 of cognitive/language impairment, cerebral palsy,severe hearing impairment, or se-vere visual impairment (cognitiveimpairment was dened as a BayleyScales of Infant Development II men-tal development index or BayleyScales of Infant Development III cog-nitive or language score [or similarstandardized examination] .2 SDsbelow the mean23,24; cerebral palsywas dened as a nonprogressivemotor impairment characterizedby abnormal muscle tone and de-creased range or control of move-ments [diagnosed on or before 24months of age]25);
5. composite outcome of death or NDIin early childhood;
6. cognitive impairment as denedabove; and
7. cerebral palsy as dened above.
Studies were included if neonatal orneurodevelopmental outcomes werecomplete for $90% and $75% of thecohort, respectively. A lower neuro-developmental follow-up rate ($70%)was accepted if infants lost to follow-upwere described in detail and compara-ble to the cohort with known outcomes.
Review Methods
Search Strategy
The search strategy was designed withthe assistance of an information scien-tist. We searched the following data-bases without language restriction:Medline (1948 through August 21, 2013),Embase Classic and Embase (1947through August 21, 2013), CochraneCentral Register of Controlled Trials(through August 21, 2013), CINAHL(through August 21, 2013), ERIC (throughAugust 21, 2013), PsycINFO (throughAugust 21, 2013), Pediatric AcademicSocieties Conference E-abstracts (20022013), and Canadian Pediatric SocietyConference E-abstracts (20102013).Detailed search terms are provided inSupplemental Information 1.
Data Extraction
Two authors (D.E.W. and K.M.) indepen-dently conducted the literature search.Information about study inclusion, studydesign, key characteristics, and out-comes was extracted independently bythe 2 reviewers using a standardizeddata collection form. Discrepancieswereresolved by consensus or by involvinga third author (P.S.S.). Authors werecontacted for clarications and/or ad-ditional data.
Assessment of Risk of Bias
For randomized studies, we used theCochrane Handbook Risk of Bias as-sessment tool.26 Studies were includedif they had a low or moderate risk ofbias. For observational studies, the riskof bias was assessed by using the in-formation in the original publications,and included evaluation for selection(representative cohort or selectedpopulation), exposure assessment,outcome assessment, attrition, andconfounding factor biases by usinga modied Newcastle-Ottawa Scale,27
which was altered to reect 2 differentfollow-up periods (neonatal and neuro-developmental follow-up) and a focus
on confounding by indication (Supple-mental Information 2). Studies wereincluded if they scored at least 6 (of 10stars) on this modied Newcastle-Ottawa Scale.
We identied, a priori, that the use ofsurgical ligationasa rescue treatmentafter failure of medical therapy poten-tially imparts 2 important sources ofbias. First, it may result in survival bias,because infants initially treated medi-cally must have survived to be eligiblefor ligation. Second, it may result inconfounding by indication, where li-gated infants may bemore likely to haveincreased pretreatment morbidity. In-fants with higher illness severity afterfailed medical treatment, characterizedby, for example, NEC, hypotension, ordependence on mechanical ventilation,may have been more likely to be re-ferred for ligation. Included studieswere therefore descriptively evaluatedon how they adjusted for postnatalconfounders at the time of the decisionto treat the PDA.
The risk of bias assessment was per-formed independently by 2 authors(D.E.W. and K.M.). Differences of opinionwere resolved by discussion and in-volvement of a third author (P.S.S.).
Data Synthesis and StatisticalAnalysis
Unadjusted and adjusted data wererespectively combined in random-effects model meta-analyses (ReviewManager 5.2; Cochrane Collaboration,Nordic Cochrane Centre, Copenhagen,Denmark).28 Meta-analyses were per-formed comparing all groups andsubgroups based on the data pre-sented in the included studies. Themeta-analyses of adjusted data onlyincluded studies that had, at a mini-mum, controlled for GA. Similar toother meta-analyses, no adjustmentfor multiple analyses was made. Studyweight in the meta-analyses was cal-culated by using the generic inverse
e1026 WEISZ et al at Indonesia:AAP Sponsored on June 7, 2015pediatrics.aappublications.orgDownloaded from
-
variance method. This method assignsgreater weight to larger studies (whichhave smaller SEs) and minimizes theimprecision of the pooled effect esti-mate. The random-effects model wasselected to account for variability be-tween and within studies because weanticipated a degree of clinical and sta-tistical heterogeneity. Because all out-comeswere binary, odds ratioswith 95%condence intervals (CIs) were used toreport differences between groups. Datafrom studies that reported both ad-justed and unadjusted risks in theirpopulation were used in the respectivemeta-analyses. It was anticipated thatnot all studies would have accounted forsimilar confounders in their adjustment,and we obtained information on thecovariates used. Adjusted odds ratios(aORs) were converted to log odds andthen meta-analyzed.
Heterogeneity and Publication BiasAssessment
Clinical heterogeneitywasassessed andreported in the table of included studies(Table 1) by describing the populationsincluded, treatment groups compared,and confounders adjusted. Statisticalheterogeneity was assessed and re-ported by using the I 2 statistic.29
Sources of heterogeneity were soughtif I2 was .50%.
RESULTS
Description of Studies
The results of the search, the studyselection log, and thenumberof studiesare reported in Fig 1. One hundredtwenty-three articles were reviewed indetail. Primary authors were contactedto obtain additional information, asneeded. Six authors provided addi-tional data or analyses not included intheir original publication.8,3034 Base-line characteristics of the 40 studiesincluded8,10,11,3066 (39 cohort studiesand 1 RCT), encompassing 32 345 pre-term infants, are reported in Table 1.
Eighty-three studies were excluded; 63studies were excluded because theydid not report sufcient data to com-pare outcomes of treatment assign-ment groups or subgroups.9,17,67127
Five studies were excluded becausethey were a review or an editorial.128132
Three studies were excluded becausethey represented cohorts alreadyaccounted for in another includedstudy.133135 Ten studies were excludedbecause.20% of infants in the cohortwere$32 weeks GA.136145 Two studieswere excluded because the cohort andanalysis included all admitted infantsrather than only infants with a PDA.12,146
NSAIDs (ibuprofen or indomethacin)were the only pharmacologic therapiesused.
Twenty-eight of the 39 included cohortstudies performed treatment groupcomparisons without adjustment forcovariates, and these data were pooledonly in the meta-analyses of univariateresults. Eleven cohort studies performedmultivariate analyses. Data from 9 ofthese studies8,10,11,30,33,51,58,63,64 and theRCT35were combined inmeta-analyses ofmultivariate results. Data from 2 cohortstudies32,34 could not be pooled becauseof insufcient information to interpretthe adjusted results. Meta-analyses wereperformed comparing all treatmentgroup and subgroup combinations whendata were provided in the includedstudies (Table 2).
Risk of Bias Among IncludedStudies
The risk of bias among cohort studies isreported in Table 3. The single includedRCT35 had a low to moderate risk of biaswith adequate allocation concealment,complete data collection, and no obvi-ous bias in reporting of outcomes.However, the study did not describe useof random sequence generation ormethods used to blind the assessmentof outcomes. Most cohort studies hada low to moderate risk of bias based onthe modied Newcastle-Ottawa Scale.Studies that performed only univariatecomparisons had a moderate to highrisk of bias because there were impor-tant perinatal differences between theligated and medically treated infants(Table 1). Ligated infants tended to havelower GA and birth weight comparedwith the medically treated infants.Among the 9 cohort studies included inmeta-analyses of multivariate results,6 studies8,10,11,33,51,64 adjusted for peri-natal covariates only. Three cohortstudies30,58,63 adjusted for postnatalPDA-related morbidities. However, 2 ofthese studies30,63 adjusted for morbid-ities that were measured or occurredafter the date of ligation (such as totalduration of mechanical ventilation, CLD,and ROP), which may have introducedbias if these morbidities are on thecausal pathway between ligation andthe outcomes of death or NDI. Only 1
FIGURE 1Study selection log.
REVIEW ARTICLE
PEDIATRICS Volume 133, Number 4, April 2014 e1027 at Indonesia:AAP Sponsored on June 7, 2015pediatrics.aappublications.orgDownloaded from
-
TABLE1
Charac
teristicsof
Stud
iesInclud
edin
theAn
alysis
FirstA
utho
r,Year
Type
ofStud
yPo
pulatio
nDenitio
nan
dDiag
nosisof
PDA
Trea
tmen
taOu
tcom
ebCo
variates
Follo
w-upCo
mplete
Cotton
,3519
78RC
TVLBW
preterm
infants,
with
HSPD
Aan
dde
pend
ent
oninvasive
mecha
nica
lven
tilationde
spite
cons
ervativ
eman
agem
ent(
n=25
)
Rand
omly
assign
edto
cons
ervativ
eman
agem
ent
orim
med
iate
surgical
clos
ure.
Days
toextuba
tion,
death,
retrolen
tal
brop
lasia,
CLD
None
100%
neon
atal
outcom
es
Trea
tmen
tgroup
sco
mpa
red:
Cons
ervativ
eman
agem
ent(
n=15
):GA
,28.36
0.3wk;
BW,1
0066
49g
Surgical
clos
ure(n
=10
):GA
,29.06
0.5wk;
BW,1
0886
46g
Merritt,36
1978
Retros
pective
coho
rt19
7419
77Co
mmon
PDAtrea
tmen
tap
proa
chDe
ath
None
100%
neon
atal
outcom
esPreterm
infantswith
anHS
PDA(clin
ical
andecho
diag
nosis)
desp
iteco
nservativ
eman
agem
ent
(n=59
).Co
mpa
riso
n:Indo
only
(n=31
),Indo
+Ligation(n
=4),P
rimaryLiga
tion(n
=24
)
Moratorium
onindo
use
during
thestud
yresu
ltedin
16infants
unde
rgoing
prim
ary
ligationrather
than
prim
aryindo
trea
tmen
t.Merritt,37
1979
Retros
pective
coho
rtAu
gust
1975Oc
tobe
r197
7Co
mmon
PDAtrea
tmen
tap
proa
chDe
ath,
cogn
itive
delay
(BSIDMDI/PDI),
neurom
otor
abno
rmality
None
Indo
grou
p:95
%;
Liga
tiongrou
p:80
%85
%Preterm
infantswith
anHS
PDA(clin
ical6
echo
diag
nosis)
desp
iteco
nservativ
eman
agem
ent
(n=52
).Co
mpa
riso
n:Indo
only
(n=26
):BW
,14
336
405g;
Prim
aryLiga
tion(n
=26
):BW
,13
136
330g
Cats,38
1980
Retros
pective
coho
rt19
7519
79Co
mmon
PDAtrea
tmen
tap
proa
chin
197619
77,
usingPO
/PR
indo
metha
cin.
In19
78-
1979
,moved
toprim
ary
ligationifco
nservativ
etrea
tmen
tfailed
(noindo
).
Death
None
100%
neon
atal
outcom
esPreterm
infantswith
BW,2kg
with
HSPD
A(clin
ical
diag
nosis)
(n=28
).Co
mpa
riso
n:Indo
only(n
=9):
GA,2
9(26
33.5)wk;
BW,1
260(77019
00)g;
Indo
+Ligation(n
=3):G
A,30
(28
34)wk;
BW,
1565
(138
519
00)g;
Prim
aryLiga
tion(n
=13
):GA
,27.4(25
30)wk;
BW,1
090(93515
00)g
Mikha
il,39
1982
Retros
pective
coho
rtJa
nuary19
76Ja
nuary19
81Trea
tmen
tofH
SPDA
was
initially
cons
ervativ
e.Liga
tionpe
rformed
afterfailu
reof
cons
ervativ
etherap
y.
Death
None
100%
neon
atal
outcom
esVLBW
subs
et(n
=41
3)of
preterm
infantsdiag
nosed
with
HSPD
A.Co
mpa
riso
n:Co
nservativ
e(n
=16
1),
Prim
aryLiga
tion(n
=25
2)
Zerella
,4019
83Re
tros
pective
coho
rt19
7719
83Ep
ochA:
PDAtrea
tmen
tinitially
with
indo
with
ligationba
ckup
Death
None
100%
hosp
ital
discha
rge
VLBW
infantswith
severe
RDSan
dPD
A(clin
ical
and
echo
diag
nosis)
(n=38
).Co
mpa
riso
n:Indo
only
(n=9),Ind
o+Liga
tion(n
=11
),Prim
aryLiga
tion
(n=18
)
EpochB:
prim
aryPD
Alig
ation
e1028 WEISZ et al at Indonesia:AAP Sponsored on June 7, 2015pediatrics.aappublications.orgDownloaded from
-
TABLE1
Continue
d
FirstA
utho
r,Year
Type
ofStud
yPo
pulatio
nDenitio
nan
dDiag
nosisof
PDA
Trea
tmen
taOu
tcom
ebCo
variates
Follo
w-upCo
mplete
Wag
ner,4
119
84Pros
pective
coho
rtAp
ril1
979
March
1981
Trea
tmen
tallo
catio
nwas
rand
omized
and
decision
toproc
eed
with
ligation
stan
dardized
bystud
yprotoc
ol.
Retrolen
tal
brop
lasia,
death
None
100%
neon
atal
outcom
esSu
bset
ofpreterm
infants,17
50gwith
HSPD
A(clin
ical
andecho
diag
nosis)
enrolle
din
RCT
(Gerso
nyet
al84)trea
tedwith
surgical
ligation
(n=10
2).C
ompa
riso
n:Indo
+Ligation(n
=28
),Prim
aryLiga
tion(n
=74
)Ca
stan
on,42
1988
Retros
pective
coho
rt19
8219
86Co
mmon
PDAtrea
tmen
tap
proa
chDe
ath
None
100%
neon
atal
outcom
esPreterm
infantswith
HSPD
A(clin
ical
andecho
diag
nosis)
(n=48
).Co
mpa
riso
n:Co
nservativ
e(n
=21
),Indo
only
(n=9),Ind
o+Liga
tion
(n=3),P
rimaryLiga
tion(n
=15
)Trus
,4319
93Re
tros
pective
coho
rtJa
nuary19
88De
cembe
r19
90Co
mmon
PDAtrea
tmen
tap
proa
chBP
D(not
dene
d)No
ne10
0%BP
Dou
tcom
esPreterm
infantswith
BW,80
0gwith
HSPD
A(clin
ical
6echo
diag
nosis)
andtrea
tedwith
NSAIDs6
Liga
tion(n
=40
).Co
mpa
riso
n:Indo
only(n
=23
):GA
,25.76
2.0wk;
BW,6
396
97g;
Indo
+Ligation
(n=17
):GA
,24.66
1.3wk;
BW,6
626
85g
Perez,4
419
98Re
tros
pective
coho
rt19
9319
97Co
mmon
PDAtrea
tmen
tap
proa
chDe
ath
None
100%
neon
atal
outcom
esVLBW
infantswith
HSPD
A(clin
ical6
echo
diag
nosis)
(n=76
).Co
mpa
riso
n:AllL
igation(n
=40
):GA
,26
wk;
BW,8
47g;
Med
ical
Man
agem
ento
nly
(n=36
):GA
,28wk;
BW,9
97g
Koeh
ne,45
2001
Retros
pective
coho
rtJa
nuary19
87De
cembe
r19
98Co
mmon
PDAtrea
tmen
tap
proa
ch,e
xcep
tind
oco
urse
was
initially
0.2mg/kg
q12h3
3do
sesfollo
wed
by0.1mg/kg
q24
hfor
upto
6d.
Death,
CLD
None
100%
neon
atal
outcom
esVLBW
infantstrea
tedforHS
PDA(clin
ical
andecho
diag
nosis)
(n=15
6).C
ompa
riso
n:Indo
only
(n=67
):GA
,26.4wk;
BW,9
10(52514
80)g;
Indo
+Ligation(n
=34
):GA
,26.1wk;
BW,8
85(57814
50)g;
Prim
aryLiga
tion(n
=55
):GA
,26
.0wk;
BW,7
60(54012
50)g
Niinikos
ki,46
2001
Retros
pective
coho
rt19
8819
98Infants.1kg
trea
tedwith
indo
(0.2
mg/kg
q12
h3
3do
ses);infan
ts,10
00gtrea
tedwith
prim
arylig
ation
Death
None
100%
neon
atal
outcom
esVLBW
infantswith
HSPD
A(clin
ical
andecho
diag
nosis)
trea
tedwith
surgicallig
ation(n
=10
1).
Compa
riso
n:Indo
+Ligation(n
=25
),Prim
ary
Liga
tion(n
=76
)Little,47
2003
Retros
pective
coho
rtJu
ne19
98March
2001
Common
PDAtrea
tmen
tap
proa
chCLD(not
dene
d)No
ne10
0%ne
onatal
outcom
esNe
onates
with
HSPD
A(clin
ical
andecho
diag
nosis)
(n=21
2).C
ompa
riso
n:Indo
alon
e(n
=12
5),
Indo
+Ligation(n
=42
),Prim
aryLiga
tion(n
=30
)
Death
REVIEW ARTICLE
PEDIATRICS Volume 133, Number 4, April 2014 e1029 at Indonesia:AAP Sponsored on June 7, 2015pediatrics.aappublications.orgDownloaded from
-
TABLE1
Continue
d
FirstA
utho
r,Year
Type
ofStud
yPo
pulatio
nDenitio
nan
dDiag
nosisof
PDA
Trea
tmen
taOu
tcom
ebCo
variates
Follo
w-upCo
mplete
ODo
novan,
4820
03Re
tros
pective
coho
rt19
9519
98Co
mmon
PDAtrea
tmen
tap
proa
chDe
ath
None
100%
neon
atal
outcom
esInfantstrea
tedforHS
PDA(clin
ical
andecho
diag
nosis)
(n=23
0).Exclude
dinfantswith
NEC
before
anytrea
tmen
t(n=6).C
ompa
riso
n:Indo
only(n
=10
8):G
A,26
.56
2wk;
BW,9
166
282g,
Indo
+Ligation(n
=66
):GA
,25.66
2wk;BW
,8266
233g;
Prim
aryLiga
tion(n
=50
):GA
,26.06
2wk;
BW,8
496
219g
Hwan
g,49
2005
Retros
pective
coho
rtJa
nuary19
99Ja
nuary20
02Trea
tmen
tallo
catio
nno
tde
scribe
d;indo
trea
tmen
t:0.2mg/kg
q12
h3
3do
ses
Death
None
100%
neon
atal
outcom
esELBW
infantswith
HSPD
A(clin
ical
andecho
diag
nosis)
(n=57
).Co
mpa
riso
n:Indo
alon
e(n
=19
):GA
,26.46
1.6wk;
BW,8
736
70g;
Indo
+Ligation(n
=8):G
A,25
.06
1.2wk;
BW,7
496
104g;
Prim
aryLiga
tion(n
=14
):GA
,26.56
2.3wk;
BW,7
846
145g;
Cons
ervativ
e(n
=16
):GA
,26.36
2.6wk;
BW,7
966
160g
Lee,
5020
06Re
tros
pective
coho
rt19
9520
00Trea
tmen
tallo
catio
nno
tde
scribe
d;us
e/no
nuse
ofNS
AIDS
know
nin
84%
Death,
CLD
(not
dene
d)No
neCLD:
86%;d
eath:
89%;R
OP:8
1%Preterm
VLBW
infantstrea
tedwith
surgical
ligation
who
sene
onatal
outcom
esarekn
own(n
=82
).Co
mpa
riso
n:Liga
tionon
ly(n
=17
),Indo
+Ligation
(n=65
)Pe
laus
a,51
2006
Retros
pective
coho
rt19
9420
02Trea
tmen
tallo
catio
nno
tdescribed
CLD,
ROP
(thresho
ld)
Match
edforGA
,BW
,gen
der
100%
neon
atal
outcom
esInfantswith
GA,30
wkwith
HSPD
Awho
faile
dmed
ical
trea
tmen
tand
unde
rwen
tsurgica
llig
ationwerematch
ed(byGA
,BW,g
ende
r)with
infantswho
received
only
med
ical
trea
tmen
t(n
=88
).Co
mpa
riso
n:NS
AID+
Liga
tion(n
=44
):GA
,25.66
1.6wk;
BW,7
766
218g;
NSAIDon
ly(n
=44
):GA
,25.76
1.5wk;
BW,7
966
205g
Kabra,
1120
07Pros
pective
coho
rt(sub
setof
RCT)
Janu
ary19
96March
1998
Subs
etof
RCTwhe
reinfantswererand
omly
assign
edto
PIor
plac
ebo;
trea
tmen
tde
cision
forlig
ation:
assign
edby
attend
ing
physician
Severe
ROP:
laseror
cryotherap
y$1eye
or stag
e$4;
CLD;
death;
CP;N
DI:com
posite
ofhe
aringor
vision
loss,
cogn
itive
orCP
ACS,
GA,g
ende
r,tw
ins,materna
led
ucation,
indo
dose
100%
neon
atal
outcom
es;
95%
18-to
21-m
ofollo
w-up
Infantswith
BW50
099
9gwith
HSPD
A(clin
ical
and
echo
diag
nosis)
(n=42
6).C
ompa
riso
n:Med
ical
Man
agem
ento
nly(n
=31
6):G
A,25
.66
1.8wk;BW
,77
16
126g;
AllL
igation(n
=11
0):G
A,25
.16
1.4wk;
BW,7
426
133g
Laug
hon,
5220
07Re
tros
pective
coho
rtJa
nuary19
97De
cembe
r20
04Trea
tmen
tallo
catio
nno
tdescribed
Death
None
100%
neon
atal
outcom
esAll2
3-to
30-wkGA
infantswith
ada
taba
sediag
nosis
ofPD
A.Co
mpa
riso
n:Co
nservativ
e(n
=38
86):GA
,27
wk(IQ
R:2629
);BW
,970
g(IQ
R:75
012
20);
Prim
aryLiga
tion(n
=70
1):G
A,25
wk(IQ
R:2427
);BW
,730
g(IQ
R:62
489
8)
CLD,
ROP(stage$3)
e1030 WEISZ et al at Indonesia:AAP Sponsored on June 7, 2015pediatrics.aappublications.orgDownloaded from
-
TABLE1
Continue
d
FirstA
utho
r,Year
Type
ofStud
yPo
pulatio
nDenitio
nan
dDiag
nosisof
PDA
Trea
tmen
taOu
tcom
ebCo
variates
Follo
w-upCo
mplete
Quresh
i,5320
08Re
tros
pective
coho
rt20
0020
05Trea
tmen
talloca
tion
notde
scribe
dCLD
BW,G
A,trea
tmen
ttype
Nots
pecied
Infantswith
BW,12
50gwith
PDA(clin
ical
andecho
diag
nosis)
(n=19
5).C
ompa
riso
n:Co
nservativ
e(n
=34
);Indo
only(n
=90
):GA
,27.4wk;BW
,9746
19g;
Prim
aryLiga
tion(n
=24
):GA
,26.36
0.2wk;
BW,8946
30g;Indo
+Ligation(n
=47
):GA
,26.36
0.5wk;
BW,8
456
43g
Only
univariate
resu
ltsavailable
Sato,54
2008
Retros
pective
coho
rtELBW
infantswho
unde
rwen
tPDA
ligationover
a3-ype
riod
(n=90
).Co
mpa
riso
n:Indo
+Ligation
(n=37
),Prim
aryLiga
tion(n
=43
).
Common
PDAtrea
tmen
tap
proa
chDe
ath,
ROP(stage
3/4/5),C
LD(not
dene
d)
None
Nots
pecied
Tsuc
hupp
ert,5
5
2008
Retros
pective
coho
rtJa
nuary19
85De
cembe
r20
05IfBW,12
50g,
received
indo
ifPD
Aseen
onecho
.IfB
W.12
50g,
received
indo
ifPD
Aclinically
sign
ica
nt.
Indo
0.2mg/kg
q12
h3
3do
ses(in
1980
s)an
d0.1mg/kg
q24
h3
6do
ses(afte
r19
80s).P
rimary
ligationpe
rformed
ifco
ntraindica
tion
toindo
.
Death,
CLD
None
100%
neon
atal
outcom
esInfants,35
wkGA
trea
tedforPD
A(ech
o6
clinical
diag
nosis)
(n=21
0).S
uccessfulM
edical
Clos
ure
grou
p(n
=15
4):B
W,1
.16
0.3kg
;versu
sFaile
dMed
ical
Clos
uregrou
p(n
=47
):BW
,1.16
0.3kg
.Co
mpa
riso
ns:PrimaryLiga
tion(n
=9),Ind
oalon
e(n
=16
8),Ind
o+Liga
tion(n
=33
)
Alexan
der,5
620
09Re
tros
pective
coho
rtJu
ne19
96Ap
ril2
005
Common
PDAtrea
tmen
tap
proa
chDe
ath
None
100%
neon
atal
outcom
esELBW
infantswith
HSPD
A(clin
ical
andecho
diag
nosis)
(n=25
8)Co
mpa
riso
n:Co
nservativ
e(n
=54
),Indo
only
(n=14
0),Ind
o+Liga
tion(n
=58
),Prim
ary
Liga
tion(n
=46
)Ko
,5720
09Re
tros
pective
coho
rtAp
ril1
992
March
2006
Trea
tmen
talloca
tion
notde
scribe
dDe
ath
None
100%
neon
atal
outcom
esAllV
LBW
infantswho
unde
rwen
tPD
Alig
ation.
Compa
riso
n:Indo
+Ligation(n
=37
),Prim
ary
Liga
tion(n
=4)
Mad
an,10
2009
Pros
pective
coho
rtJa
nuary20
00De
cembe
r20
0429
%received
PI.Treatmen
talloca
tionas
sign
edby
attend
ingph
ysician.
Severe
ROP:
laser/
cryotherap
y$1eye
orstag
e$4;CLD;
NDI
(18
21mo):
compo
site
ofhe
aringaids
,blindn
ess
$1eye,
severe
cogn
itive
delay,or
mod
erate
severe
CP
Center,G
A,BW
,ge
nder,P
I,labo
r,Ap
gar,
RDS,
IUGR
,AC
S,TO
RCH,
seps
is,m
arita
lstatus
,materna
lag
e
100%
(byinclus
ion
crite
ria)
Infants2328
wkan
d40
110
00gwho
survived
.72
hwith
HSPD
A(clin
ical6
echo
diag
nosis)
andha
d1821
mofollo
w-upas
sessmen
t(n
=28
38).Co
mpa
riso
n:Co
nservativ
e(n
=40
3):
GA,2
5.66
1.5wk;
BW,7
586
148g;
Indo
only
(n=15
25):GA
,25.46
1.3wk;
BW,7
456
139g;
Indo
+Ligation(n
=77
5):G
A,24
.86
1.3wk;
BW,
7196
134g;
Prim
aryLiga
tion(n
=13
5):G
A,25
.16
1.4wk;
BW,7
266
133g
REVIEW ARTICLE
PEDIATRICS Volume 133, Number 4, April 2014 e1031 at Indonesia:AAP Sponsored on June 7, 2015pediatrics.aappublications.orgDownloaded from
-
TABLE1
Continue
d
FirstA
utho
r,Year
Type
ofStud
yPo
pulatio
nDenitio
nan
dDiag
nosisof
PDA
Trea
tmen
taOu
tcom
ebCo
variates
Follo
w-upCo
mplete
Quresh
i,5820
09Re
tros
pective
coho
rtInfantswith
BW,12
50gwith
PDA(ech
odiag
nosis)
(n=17
6).C
ompa
riso
n:Med
ical
Man
agem
ento
nly
(n=11
0):G
A,27
.4wk;
BW,9
036
36g,
includ
ing
Cons
ervativ
eon
ly(n
=26
),Indo
only
(n=84
),versus
AllL
igation(n
=66
),includ
ingPrim
ary
Liga
tion(n
=19
):GA
,25.56
0.3wk;
BW,8
086
47g;Indo
+Ligation(n
=47
):GA
,26.36
0.2wk;BW
,89
46
30g
Trea
tmen
tallo
catio
nno
tdescribed
Deathor
NDI
GA,B
W,P
DAscore,
IVH,
hypo
tens
ion
inrstwee
k
100%
neon
atal
mor
tality;85
%ne
urod
evelop
men
tal
follo
w-up
Vida
,5920
09Re
tros
pective
coho
rt20
0120
07Ibup
rofenwas
NSAID
used
;com
mon
PDA
trea
tmen
tapp
roac
h
Death,
CLD,
ROP
($stag
e3)
None
100%
neon
atal
outcom
esInfants,32
wkGA
with
HSPD
A(clin
ical
andecho
diag
nosis)
trea
tedwith
prim
aryNS
AIDtherap
y(n
=20
1).C
ompa
riso
n:NS
AIDon
ly(n
=14
9):G
A,27
wk(IQ
R:2528
);BW
,840
g(IQ
R:67
010
16);
NSAID+
Liga
tion(n
=52
):GA
,25wk(IQ
R:2426
.5);
BW,7
30g(IQ
R:59
591
5)Ch
iruvolu,
6020
10Re
tros
pective
coho
rtJa
nuary20
06De
cembe
r20
08Trea
tmen
tallo
catio
nno
tdescribed
Death;
CLD;
severe
ROP:.stag
e2
None
100%
neon
atal
outcom
esELBW
infantswith
PDA(clin
ical
andecho
diag
nosis)
(n=19
0).C
ompa
riso
n:Co
nservativ
e(n
=16
),NS
AIDon
ly(n
=55
),Prim
aryLiga
tion(n
=61
),NS
AID+
Liga
tion(n
=58
)Lope
s,61
2010
Retros
pective
coho
rt19
9920
04Trea
tmen
tallo
catio
nno
tdescribed
CP,cog
nitiv
ede
lay(M
DI,70
),de
afne
ss,
Psycho
motor
developm
ent
None
100%
neon
atal
outcom
es;
81%
18-m
ofollo
w-up
(ofc
ohor
t)
Coho
rtof
143infantswith
BW#80
0g(n
=93
).Co
mpa
riso
n:Co
nservativ
e(n
=12
),Indo
only
(n=31
),Indo
+Ligation(n
=42
),Prim
ary
Liga
tion(n
=8)
Natarajan,
6220
10Re
tros
pective
coho
rtJa
nuary20
04De
cembe
r20
06Co
mmon
PDAtrea
tmen
tap
proa
ch;ibu
profen
andindo
used
CLD,
ROP(req
uiring
trea
tmen
t),d
eath
None
100%
neon
atal
outcom
esCo
hort
ofVLBW
infantswith
anHS
PDA(clin
ical
and
echo
diag
nosis)
trea
tedwith
surgical
ligation
(n=82
).Co
mpa
riso
n:Prim
aryLiga
tion(n
=28
),NS
AID+
Liga
tion(n
=54
)Rh
einlae
nder,63
2010
Retros
pective
coho
rtJa
nuary19
98De
cembe
r20
03Co
mmon
PDAtrea
tmen
tap
proa
ch;ind
o0.2mg/kg
q12
h3
3do
ses,then
0.1mg/kg
q24
hup
to6d;
ibup
rofen10
/5/5
mg/kg
q24
h
Death,
mor
talityat
2y,
CLD,
ROP.stag
e2,
hearingaids
,blindn
ess,
CP,cog
nitiv
ede
lay,
compo
site
poor
outcom
e,de
athor
poor
outcom
e
Compo
site
NDI:
CRIB,B
W,G
A,BP
D,ge
nder,
O 2da
ys,intub
ation
days,R
OP,IVH
,PVL,
surfac
tant
89.6%
2-you
tcom
e(dea
thor
NDI)
know
nVLBW
infantswith
PDA(clin
ical
andecho
diag
nosis)
trea
tedwith
NSAID6
Liga
tion(n
=18
2).
Compa
riso
n:NS
AIDon
ly(n
=13
0),N
SAID+L
igation
(n=52
).Infantswith
ductal
patenc
yafterNS
AID
therap
y(52of
54werethen
ligated
)had
lower
BW,
GA,h
ighe
rCR
IBscore,an
dmoreRD
S,mecha
nica
lventila
tion,
ventila
torda
ys.
Pelaus
a,64
2011
Retros
pective
coho
rtInfants,28
wkGA
with
anHS
PDAwho
survived
.7dan
dtrea
tedwith
surgical
ligationwere
match
edforGA
(61wk)
with
aninfant
with
PDA
who
received
only
med
ical
trea
tmen
t(n=72
).Co
mpa
riso
n:AllLigation(n
=36
):GA
,256
1.1wk;
BW,7
096
20g;
Med
ical
Man
agem
ento
nly
(n=36
):GA
,24.96
1.1wk;
BW,7
266
159g
Trea
tmen
tallo
catio
nno
tdescribed
Death,
developm
ental
issu
esat
4yof
age(spe
ech,
developm
ental
delays,C
P,AD
HD,a
utism,
deafne
ss,
blindn
ess)
Match
edforGA
(61wk)
Notsp
ecied
e1032 WEISZ et al at Indonesia:AAP Sponsored on June 7, 2015pediatrics.aappublications.orgDownloaded from
-
TABLE1
Continue
d
FirstA
utho
r,Year
Type
ofStud
yPo
pulatio
nDenitio
nan
dDiag
nosisof
PDA
Trea
tmen
taOu
tcom
ebCo
variates
Follo
w-upCo
mplete
Adrouc
he-Amrani,65
2012
Retros
pective
coho
rtAu
gust
2004Ju
ly20
09Co
mmon
PDAtrea
tmen
tap
proa
ch;sed
ibup
rofen
andindo
CLD,
ROP(the
seda
tano
tinc
lude
din
analysis)
None
100%
neon
atal
outcom
esELBW
infantswith
HSPD
A(clin
ical
andecho
diag
nosis)
(n=80
).Co
mpa
riso
n:NS
AIDon
ly(n
=48
):GA
,25.46
0.2wk;
BW,7
726
16g;
NSAID+
Liga
tion(n
=32
):GA
,24.96
0.2wk;
BW,
7266
24g
Heuc
han,
6620
12c
Retros
pective
coho
rt20
0120
07Trea
tmen
tallo
catio
nno
tdescribed
Mor
talityat
1y
None
100%
1-ymor
tality
outcom
ePreterm
infantswith
HSPD
A(clin
ical
andecho
diag
nosis)
with
persistent
need
forresp
iratory
supp
ort,trea
tedwith
surgical
ligation.
Trea
tmen
tgrou
ps(m
edian[IQ
R]):NS
AID+
Liga
tion(n
=71):
GA,26(25
27)wk;BW
,840
(73310
00)g;
Prim
ary
Liga
tion(n
=54
):GA
,26(25
27)wk;
BW,8
30(72710
69)g
Hsu,
3420
12c
Retros
pective
coho
rt19
9720
07Trea
tmen
tallo
catio
nno
tdescribed
30-d
postop
erative
mortality,ho
spita
lmortality
NICU
levela
ndun
specied
comorbiditie
s
100%
neon
atal
outcom
esVLBW
infantswith
aPD
A(ech
odiag
nosis)
(n=
1078
0).C
ompa
riso
n:AllL
igation(n
=24
97),
Med
ical
Man
agem
enton
ly(n
=82
83)
Mirea
,820
12c
Retros
pective
coho
rt20
0420
08Co
mmon
PDAtrea
tmen
tap
proa
chDe
ath,
ROP$stag
e3,
CLD;
compo
site
outcom
eno
tinc
lude
din
this
review
GA,A
CS,m
ultip
lebirths
,gen
der,
SGA,
SNAP
IIscore
Nots
pecied
Preterm
infantswith
GA#32
wk,
who
survived
$72
h,with
adiag
nosisof
PDA(diagn
osed
clinically6
echo
)(n
=35
56).Co
mpa
riso
n:Co
nservativ
e(n
=57
7):G
A,28
.36
2.3wk;
Indo
only
(n=20
26):GA
,27.06
2.1wk;
Indo
+Ligation
(n=62
6):G
A,25
.56
1.7wk;
Prim
aryLiga
tion
(n=32
7):G
A,26
.06
2.3wk
Moo
re,33
2012
cRe
tros
pective
coho
rtJa
nuary19
94De
cembe
r20
05Co
mmon
PDAtrea
tmen
tap
proa
chCLD;
severe
ROP:
anyne
edfor
lasertherap
y,de
ath
Death:
GA,A
CS,
SGA,
gend
erNo
tspe
cied
Infants2326
wkGA
with
HSPD
A(clin
ical
andecho
diag
nosis)
that
didno
tclose
after1co
urse
ofindo
(n=13
3).Ligated
infantsha
dlower
BWan
dmore
inotrope
usebu
tsim
ilarGA
tono
nligated
infants.
Compa
riso
n:Indo
only
(n=58
),Indo
+Ligation
(n=75
)Ja
wa,
3020
13c
Retros
pective
coho
rt20
0520
10Trea
tmen
tallo
catio
nno
tdescribed
CLD,
ROP.stag
e2,
Death,
CPat
2y
CLD;
ROP;
death:
GA;C
P:CLD,
IVH,
ROP,seps
is
100%
neon
atal
outcom
es;
2-yfollo
w-upno
tde
scribe
d
VLBW
infantswith
HSPD
A(clin
ical
andecho
diag
nosis)
(n=36
1).C
ompa
riso
n:Co
nservativ
e(n
=85
);Med
ical
Man
agem
ento
nly(n
=17
8):G
A,27
.16
0.7wk;
AllL
igation(n
=98
):GA
,25
.86
0.8wk
Tsui
3120
13c
Retros
pective
coho
rtJa
nuary20
06De
cembe
r20
10Trea
tmen
tallo
catio
nno
tdescribed
Severe
ROPrequ
iring
lasertrea
tmen
tNo
ne;u
nivariate
resu
ltsus
edin
this
review
100%
ROP
outcom
esVLBW
infantswith
PDA(n
=20
4).C
ompa
riso
n:Co
nservativ
eon
ly(n
=88
),Indo
only
(n=49
),Indo
+Ligation(n
=20
),Prim
aryLiga
tion(n
=47
)
REVIEW ARTICLE
PEDIATRICS Volume 133, Number 4, April 2014 e1033 at Indonesia:AAP Sponsored on June 7, 2015pediatrics.aappublications.orgDownloaded from
-
study adjusted for postnatal pre-ligation confounders, by controllingfor IVH, hypotension, and a PDA-relatedillness severity score measured at thetime of the decision to treat the PDA.58
The possibility of survival bias was notaddressed by any cohort study.
Main Results
Meta-analyses of univariate and mul-tivariate results for all treatmentgroup and subgroup comparisons arepresented in Table 2. The outcomesfor ligated compared with medicallytreated infants are as follows:
1. Death before discharge from theNICU: a meta-analysis of studiesthat reported an adjusted risk ofdeath revealed that infants whounderwent surgical ligation hadlower odds of death comparedwith those infants who were trea-ted medically (5 studies, 7159 par-ticipants; pooled aOR: 0.54; 95% CI:0.380.77; I 2 = 39%) (Fig 2A, Ta-ble 2). The association between li-gation and decreased mortalitywas seen across all subgroupcomparisons (Fig 2 BD, Table 2).
2. CLD: meta-analysis revealed thatsurgical ligation was associatedwith higher odds of CLD comparedwith medical management alone(4 studies, 6703 participants; pooledaOR: 2.51; 95% CI: 1.983.18; I 2 = 44%)(Fig 3A, Table 2).
3. Severe ROP: meta-analysis re-vealed that surgical ligation wasassociated with higher odds of se-vere ROP compared with medicalmanagement alone (3 studies, 3122participants; pooled aOR: 2.23; 95%CI: 1.623.08; I 2 =37%) (Fig 4A, Table 2).
4. NDI in early childhood: meta-analysis revealed that NDI washigher among infants who under-went surgical ligation comparedwith those treated medically (3studies, 3250 participants; pooledTA
BLE1
Continue
d
FirstAu
thor,Yea
rType
ofStud
yPo
pulatio
nDenitio
nan
dDiag
nosisof
PDA
Trea
tmen
taOu
tcom
ebCo
variates
Follo
w-upCo
mplete
Youn
,3220
13c
Retros
pective
coho
rtNo
vembe
r20
09No
vembe
r20
11Initial
trea
tmen
twith
ibup
rofen3on
ce-daily
doses(10/5/5mg/kg
);prim
arylig
ationifinfant
hemod
ynam
ically
unstab
le(exclude
dfrom
analysis)
ROP:
Anyplus
diseas
eor
need
forlaser
therap
y;CLD;
death
BW,G
A,initial
PDA
shun
tsize;
only
univariate
data
used
inmeta-an
alyses
100%
neon
atal
outcom
esVLBW
infantswith
HSPD
A(clin
ical
andecho
diag
nosis)
(n=11
5).C
ompa
riso
n:Ibup
rofen+
Liga
tion(n
=29
):GA
,27.86
2.8wk;
BW,1
.16
0.3kg
;PDA
diam
eter,3.16
1.0mm;Ibu
profen
only
(n=75
):GA
,30.46
3.14
wk;
BW,1
.56
0.6kg
;PD
Adiam
eter,2
.16
0.9mm
Alls
tudies
performed
surgical
ligationviaapo
sterolateral
thorac
otom
ywith
useof
aclip
orlig
ature.BW
andGA
areprovided
whe
navailable.
BW(ing)
ispresen
tedas
mea
ns6
SDsor
med
ians
(ran
ge)an
dGA
(inwk)
ispresen
tedas
mea
ns6
SDs
ormed
ians
(ran
ge)u
nlessothe
rwisesp
ecied
.ACS
,anten
atal
corticos
teroids;BW
,birth
weigh
t;BS
ID,B
ayleySc
ales
ofInfant
Developm
entII;CP,cereb
ralp
alsy;ech
o,echo
cardiogram
;ELB
W,extremelylowbirthweigh
t;HS
PDA,he
mod
ynam
icallysign
ica
ntPD
A;indo
,ind
ometha
cin;
IQR,
interqua
rtile
rang
e;MDI,m
entald
evelop
men
tinde
x;PD
I,ps
ycho
motor
developm
entinde
x;PI,p
roph
ylac
ticindo
metha
cin;
q,qu
aque
(Latin)SG
A,sm
allfor
gestationa
lage
;SNA
PII,
ScoreforNe
onatal
AcutePh
ysiology
II;VLBW
,verylow
birthweigh
t.aTh
eco
mmon
PDAtrea
tmen
tapp
roac
hindica
tesinitial
man
agem
ento
fHSP
DAwith
indo
metha
cin(w
ithor
with
outc
onservativeman
agem
ent).Ind
ometha
cinwas
administeredintraven
ously(0.2mg/kg3
3do
sesevery1224
h).S
urgica
lligationwas
performed
ifindo
metha
cintherap
yfaile
dor
was
contraindica
ted.
Trea
tmen
talloca
tionwas
assign
edby
theattend
ingph
ysician.
bCLDwas
dene
das
thene
edforoxygen
orpo
sitiv
e-pressu
reventila
tionat
36wkpo
stmen
strual
ageun
less
othe
rwisesp
ecied
.cAd
ditio
nald
ataor
analyses
wereprovided
bytheau
thor.
e1034 WEISZ et al at Indonesia:AAP Sponsored on June 7, 2015pediatrics.aappublications.orgDownloaded from
-
TABLE2
Meta-an
alyses
oftheEffect
ofPD
ATrea
tmen
tAs
sign
men
tSu
bgroup
onNe
onatal
andNe
urod
evelop
men
talOu
tcom
es
Outcom
eCo
mpa
riso
nNo
.ofS
tudies
(Univariate)
Pooled
OR(95%
CI)
No.o
fPa
rticipan
tsI2 ,
%No
.ofS
tudies
(Multiv
ariate)
Pooled
aOR
(95%
CI)
No.o
fPa
rticipan
tsI2 ,
%
Death
AllL
igationversus
AllM
edical
Man
agem
enton
ly16
0.71
(0.54
0.94
)86
6756
50.54
(0.38
0.77
)71
5939
NSAIDan
dLiga
tionversus
Prim
aryLiga
tion
210.70
(0.51
0.96
)31
7025
0
NSAIDan
dLiga
tionversus
NSAIDon
ly18
0.77
(0.57
1.04
)66
8840
30.44
(0.37
0.53
)51
110
NSAIDan
dLiga
tionversus
Cons
ervativ
e7
0.35
(0.19
0.64
)26
5072
10.20
(0.13
0.31
)12
01
Prim
aryLiga
tionversus
NSAIDon
ly14
1.02
(0.71
1.46
)51
2337
20.65
(0.48
0.89
)40
060
Prim
aryLiga
tionversus
Cons
ervativ
e10
0.58
(0.41
0.83
)67
2257
20.29
(0.18
0.47
)92
80
CLD
AllL
igationversus
AllM
edical
Man
agem
enton
ly7
2.19
(1.43
3.34
)46
0382
42.51
(1.98
3.18
)67
0344
NSAIDan
dLiga
tionversus
Prim
aryLiga
tion
81.06
(0.75
1.50
)14
8930
0
NSAIDan
dLiga
tionversus
NSAIDon
ly12
2.49
(1.81
3.43
)38
4560
32.76
(2.10
3.62
)48
1753
NSAIDan
dLiga
tionversus
Cons
ervativ
e3
4.98
(2.57
9.67
)12
5850
13.12
(2.32
4.20
)11
03
Prim
aryLiga
tionversus
NSAIDon
ly5
1.60
(0.93
2.77
)26
4269
22.03
(1.57
2.61
)37
950
Prim
aryLiga
tionversus
Cons
ervativ
e4
4.13
(2.56
6.68
)52
7977
12.67
(1.89
3.77
)80
9
Severe
ROP
AllL
igationversus
AllM
edical
Man
agem
enton
ly5
3.97
(3.33
4.73
)35
1622
32.23
(1.62
3.08
)31
2237
NSAIDan
dLiga
tionversus
Prim
aryLiga
tion
61.04
(0.79
1.38
)12
080
0
NSAIDan
dLiga
tionversus
NSAIDon
ly7
3.72
(3.04
4.57
)27
520
21.82
(1.41
2.33
)20
800
NSAIDan
dLiga
tionversus
Cons
ervativ
e2
9.26
(4.85
17.70)
961
301
3.02
(1.73
5.27
)85
3
Prim
aryLiga
tionversus
NSAIDon
ly3
3.51
(2.61
4.72
)18
540
11.80
(1.25
2.59
)16
42
Prim
aryLiga
tionversus
Cons
ervativ
e4
5.06
(4.23
6.04
)50
500
12.98
(1.61
5.52
)50
3
NDI
AllL
igationversus
AllM
edical
Man
agem
enton
ly1
2.13
(1.32
3.45
)34
0
31.54
(1.01
2.33
)32
5048
NSAIDan
dLiga
tionversus
NSAIDon
ly1
0.65
(0.27
1.58
)14
1
21.39
(0.97
1.98
)24
4129
Prim
aryLiga
tionversus
NSAIDon
ly0
1
1.79
(1.11
2.89
)16
60
Deathor
NDI
AllL
igationversus
AllM
edical
Man
agem
enton
ly1
1.50
(0.97
2.33
)42
6
40.95
(0.58
1.57
)35
1271
NSAIDan
dLiga
tionversus
NSAIDon
ly1
1.02
(0.52
1.98
)18
2
11.03
(0.82
1.30
)23
00
Prim
aryLiga
tionversus
NSAIDon
ly0
1
1.54
(1.00
2.37
)16
60
Cogn
itive
impa
irmen
tAllL
igationversus
AllM
edical
Man
agem
enton
ly2
2.15
(1.34
3.44
)42
40
11.96
(1.14
3.37
)33
1
NSAIDan
dLiga
tionversus
NSAIDon
ly1
0.70
(0.28
1.80
)14
1
0
Prim
aryLiga
tionversus
NSAIDon
ly1
3.33
(0.32
34.99)
42
0
Cerebral
palsy
AllL
igationversus
AllM
edical
Man
agem
enton
ly3
2.65
(1.14
6.18
)70
964
21.51
(0.86
2.63
)61
60
NSAIDan
dLiga
tionversus
NSAIDon
ly1
2.06
(0.64
6.65
)14
1
0
Prim
aryLiga
tionversus
NSAIDon
ly1
4.27
(1.13
16.05)
42
0
AOR,
adjusted
odds
ratio
;OR,
odds
ratio
;NSA
ID,n
on-steroidal
anti-inam
matorydrug
;ROP
,retinop
athy
ofprem
aturity.
REVIEW ARTICLE
PEDIATRICS Volume 133, Number 4, April 2014 e1035 at Indonesia:AAP Sponsored on June 7, 2015pediatrics.aappublications.orgDownloaded from
-
TABLE3
Risk
ofBias
Assessmen
t:Co
hort
Stud
iesa
FirstA
utho
r,Year
Selection
Compa
rability
Outcom
eOv
erallb
Represen
tativ
eness
ofExpo
sedCo
hort
SelectionofNo
nexpos
edCo
hort
Ascertainm
ent
ofExpo
sure
Demon
stratio
nTh
atOu
tcom
eof
Interest
Not
Presen
tatS
tart
Compa
rability
ofCo
hortson
Basisof
Design
/An
alysis
(Out
ofPo
ssible
2)
Metho
dology
Addresses
Confou
nding
byIndica
tion
Assessmen
tof
Outcom
eFollo
w-up
Long
Enou
gh
Adeq
uacy
ofFollo
w-upof
Coho
rts(toDischa
rge)
Adeq
uacy
ofFollo
w-up
ofCo
horts
(Lon
g-term
)
Merritt,1
9783
6*
**
*
No*
**
N/A
7/9
Merritt,1
9793
7*
**
*
No*
**
*8/10
Cats,1
9803
8*
**
*
No*
**
N/A
7/9
Mikha
il,19
8239
**
**
No
**
*N/A
7/9
Zerella
,198
340
**
**
No
**
*N/A
7/9
Wag
ner,19
8441
**
**
No
**
*N/A
7/9
Castan
on,1
9884
2*
**
*
No*
**
N/A
7/9
Trus
,199
343
**
**
No
**
*N/A
7/9
Perez,19
9844
**
**
No
**
*N/A
7/9
Koeh
ne,2
0014
5*
**
*
No*
**
N/A
7/9
Niinikos
ki,2
0014
6*
**
*
No*
**
N/A
7/9
Little,2
0034
7*
**
*
No*
**
N/A
7/9
ODo
novan,
2003
48*
**
*
No*
**
N/A
7/9
Hwan
g,20
0549
**
**
No
**
*N/A
7/9
Lee,
2006
50*
**
*
No*
*
N/A
6/9
Pelaus
a,20
0651
**
**
**No
**
*N/A
9/9
Kabra,
2007
11*
**
***
No*
**
*10
/10
Laug
hon,
2007
52*
**
*
No*
**
N/A
7/9
Quresh
i,20
0853
**
**
No
**
*N/A
7/9
Sato,2
0085
4*
**
*
No*
**
N/A
7/9
Tsch
uppe
rt,2
0085
5*
**
*
No*
**
N/A
7/9
Alexan
der,20
0956
**
**
No
**
*N/A
7/9
Ko,2
0095
7*
**
*
No*
**
N/A
7/9
Mad
an,2
0091
0*
**
***
No*
**
*10
/10
Quresh
i,20
0958
**
**
**Yes
**
**
10/10
Vida
,200
959
**
**
No
**
*N/A
7/9
Chiruvolu,
2010
60*
**
*
No*
**
N/A
7/9
Lope
s,20
1061
**
**
No
**
*
7/10
Natarajan,
2010
62*
**
*
No*
**
N/A
7/9
Rheinlae
nder,20106
3*
**
***
Yes
**
**
10/10
Pelaus
a,20
1164
**
**
*No
**
**
9/10
Adrouc
he-Amrani,
2012
65*
**
*
No*
**
N/A
7/9
Heuc
han,
2012
66*
**
*
No*
**
N/A
7/9
Hsu,
2012
34*
**
***
No*
**
N/A
9/9
Mirea
,201
28*
**
***
No*
**
N/A
9/9
Moo
re,2
0123
3*
**
***
No*
**
N/A
9/9
Jawa,
2013
30*
**
**
Yes
**
*
8/10
Tsui,2
0133
1*
**
*
No*
**
N/A
7/9
Youn
,201
332
**
**
**No
**
*N/A
9/9
N/A,
notap
plicab
le.
aMod
ied
Newca
stle-Ottaw
ascale(Sup
plem
entalInformation2).
bMaxim
um9(for
stud
iesrepo
rtingne
onatal
outcom
es)or
10(for
stud
iesrepo
rtingne
urod
evelop
men
talo
utco
mes).
e1036 WEISZ et al at Indonesia:AAP Sponsored on June 7, 2015pediatrics.aappublications.orgDownloaded from
-
aOR: 1.54; 95% CI: 1.012.33; I 2 =48%) (Fig 5A, Table 2). There wasa trend toward higher NDI amonginfants treated with NSAIDs andligation compared with infantstreated with NSAIDs alone (2 stud-ies, 2441 participants; aOR: 1.39;95% CI: 0.971.98; I 2 = 29%) (Fig5B, Table 2).
5. Death or NDI in early childhood: 4studies reported the composite out-come comparing all surgically andmedically treated infants. Madanet al10 reported on the neurodevel-opmental outcomes of 2838 infantswith birth weights of 400 to 1000 gwho survived .72 hours and had
a symptomatic PDA and follow-up at18 to 21 months corrected GA.Kabra et al11 performed a secondaryanalysis of the Trial of IndomethacinProphylaxis in Preterm Infants.Pelausa et al64 performed a retro-spective cohort study comparinga small group of ligated preterminfants with a group of medicallytreated infants matched for GA,body weight, and gender. Qureshiet al58 reported a multivariate anal-ysis of a retrospective cohort ofpreterm infants with PDA, adjustingfor PDA-related illness severity atthe time of the decision to treatthe PDA. Meta-analysis of these
studies revealed no difference inthe composite outcome of deathor NDI between surgically and med-ically treated infants (4 studies,3512 participants; pooled aOR: 0.95;95% CI: 0.581.57; I 2 = 71%) (Fig 6,Table 2). Most of the statistical hetero-geneity between these 4 studies isaccounted for by 1 study,58 whichwas the only study to report a signif-icant protective effect of surgical liga-tion. Importantly, this study was alsothe only one in this review to controlfor confounding by indication.
6. Cognitive impairment: only 1 study11
reported an adjusted estimate ofcognitive impairment and found
FIGURE 2Adjusted estimates of death. A, All Ligation versusAllMedicalManagement only. B, NSAIDand Ligation versusNSAID only. C, Primary Ligation versusNSAIDonly. D,Primary Ligation versus Conservative. df, degrees of freedom; IV, inverse variance; Mgmt, management.
REVIEW ARTICLE
PEDIATRICS Volume 133, Number 4, April 2014 e1037 at Indonesia:AAP Sponsored on June 7, 2015pediatrics.aappublications.orgDownloaded from
-
increased odds associated with liga-
tion compared with medical manage-
ment alone (331 participants; aOR:
1.96; 95% CI: 1.143.37) (Table 2).
7. Cerebral palsy: meta-analysis re-vealed that there was no difference
in the odds of cerebral palsy11,63
between those who underwent sur-
gical ligation and those treated
medically (2 studies, 616 partici-
pants; pooled aOR: 1.51; 95% CI:
0.862.63; I 2 = 0%) (Fig 5C, Table 2).
DISCUSSION
In this systematic review and meta-analysis of the effects of PDA treatmentassignment on neonatal and neuro-developmental outcomes, we identiedthat surgical ligation is associated with
FIGURE 3Adjusted estimates of CLD. A, All Ligation versus All MedicalManagement only. B, NSAID and Ligation versus NSAID only. C, Primary Ligation versus NSAID only. df,degrees of freedom; IV, inverse variance; Mgmt, management.
FIGURE 4Adjusted estimates of severe ROP. A, All Ligation versus All MedicalManagement only. B, NSAID and Ligation versus NSAID only. df, degrees of freedom; IV, inversevariance; Mgmt, management.
e1038 WEISZ et al at Indonesia:AAP Sponsored on June 7, 2015pediatrics.aappublications.orgDownloaded from
-
decreasedmortality but increased NDI inearly childhood when compared withmedical treatment alone. There are sev-eral possible explanations for the di-vergence of these competing outcomes:rst, surgical ligation may improve thesurvival of infants but be simultaneouslyneurologically detrimental; second, sur-gical ligationmay improve the survival ofinfants, but infants referred for ligationmay be at higher preligation risk of NDI(confounding by indication); and nally,the decrease in mortality may be inu-enced by survival bias (where medicallytreated infants with a PDA die before
being referred for ligation), with theincrease in NDI explained by either con-founding by indication or a true detri-mental effect of ligation.
Several aspects of treatmentwith ligationhave been proposed as contributing to anincreased risk of NDI, such as surgical oranesthesia effects or postoperative he-modynamic compromise. Early surgicalmortality associated with PDA ligation isreported to be low.49,56,66 Direct surgicalmorbidities include bleeding, pneumo-thorax, and left vocal cord paresis (VCP).The incidence of left VCP is variable butmay complicate 5% to 50% of PDA liga-
tions and is associated with an increasedrisk of death, extubation failure, CLD, needfor gastrostomy tube, and gastroesoph-ageal reux disease.66,147149 Recentstudies have reported an association be-tween the use of halothane gases for an-esthesia in young children and NDI.150,151
Preterm infants are at risk of postligationhemodynamic instability, which may re-sult in cerebral hypoperfusion, neuronalinjury, and subsequent NDI.152157 In ourreview, only 1 of the included studies30
reported on an association of VCP andNDI, and no study described an associa-tion between postligation hemodynamic
FIGURE 5Adjusted estimates of NDI in early childhood (A, B) and cerebral palsy (C). A, All Ligation versus All MedicalManagement only. B, NSAID and Ligation versus NSAIDonly. C, All Ligation versus All Medical Management only. df, degrees of freedom; IV, inverse variance; Mgmt, management.
FIGURE 6Adjusted estimate of the composite outcome of death or NDI in early childhood: All Ligation versus All Medical Management only. df, degrees of freedom; IV,inverse variance; Mgmt, management.
REVIEW ARTICLE
PEDIATRICS Volume 133, Number 4, April 2014 e1039 at Indonesia:AAP Sponsored on June 7, 2015pediatrics.aappublications.orgDownloaded from
-
instability and NDI. Infants with lower GAand weight,1 kg at the time of ligationare at highest risk of postligation hemo-dynamic instability, possibly due todecreased myocardial adaptability to al-tered loading conditions.158 This ndingraises the possibility that age at ligationmay contribute to the risk of NDI; how-ever, a post hoc analysis of the Trial ofIndomethacin Prophylaxis in PretermInfants found no association between thetiming of PDA ligation and the risk ofNDI.11
Despite the reported association be-tween ligation and NDI, observationalstudies to date have not adequately dif-ferentiated theeffect of biasversusa truedetrimental effect of ligation. In general,observational studies are subject to biaswhen treatment assignment is not in-dependentofbaselineprognostic factors.Many of the included studies did notperformanyadjustment forconfounders.Infants who underwent surgical ligationof their PDA had signicantly differentbaseline characteristics compared withtheir medically treated peers, with gen-erally lowerGAandbirthweight (Table 1).The presence of these major con-founders limits the validity of unadjustedcomparisons. In addition, most of themultivariate analyses included in thisreview only adjusted for antenatal orperinatal covariates (Table 1). This set ofcovariates, if complete, would be suf-cient to balance baseline prognosticfactors for interventions that occurshortly after birth. However, PDA ligationoften occurs several weeks after birth,and the interval accumulation of PDA-associated comorbidities inuences bothtreatment assignment and outcomes.
Themost importantsourceofbias in theincluded multivariate studies is con-founding by indication, in which infantswith higher illness severity, whomay beat higher risk of NDI,159,160 may be morelikely to be assigned to ligation. Only 1study58 adjusted for preligation post-natal confounders, such as IVH or the
duration/intensity of mechanical ven-tilation at the time of the decision totreat the PDA. In the setting of surgicalligation, severe IVH is a potential con-founder. It has been associated withPDA ligation45,52,62 and NDI161165 and isnot typically on the causal pathwaybetween PDA ligation and NDI. Most IVH(90%) occurs in the rst week of life,166
preceding the timing of surgical liga-tion reported inmost studies.33,45,48,50,59
Other potential key confounders includeprolonged duration of mechanical ven-tilation, severe hypotension, postnatalsepsis, and NEC, which increase illnessseverity but are also associatedwith PDAligation, death, NDI, CLD, and ROP.167175
These morbidities, however, can occurbefore, during, or after PDA treatmentsand thus may be considered con-founders in some infants and outcomesin others. Therefore, data on the timingof these confounders relative to ligationshould be incorporated into multivari-able models examining the impact ofsurgical treatment.
The lowermortality in the ligated groupmay be attributable, in part, to survivalbias. The PDA treatment algorithms inmany of the included studies (Table 1)cited that infants were treated with li-gation when indomethacin failed orwas contraindicated. Ligation was of-ten undertaken later in life relative tomedical therapy, meaning that ligatedinfants were more likely to have al-ready survived the period of high earlyneonatal mortality. This implies thatsome of the sickest infants, treatedinitially with conservativemanagementand/or indomethacin, may have diedbefore becoming eligible for ligation,resulting in selection bias in assem-bling the cohort of ligated infants. Thisissue was present inmost of the cohortstudies, and thus our ndings of de-creased mortality should be inter-pretedwith caution. Future studies thatcompare surgical ligation with medicaltherapy should ensure careful match-
ing of the control population with con-sideration of age at ligation.
The possibility of survival bias is sup-ported by ndings of the subgroupmeta-analyses. Survival bias associ-ated with ligation would be expected tomanifest as an apparent survival ad-vantage when ligation is performedlater in life, but that ligation performedat a similar day of life as medicaltherapy would have similar mortalityrates.We found that infants treatedwithNSAIDs and ligation had lower adjustedodds of death compared with infantswhounderwent treatmentwithprimaryligation, NSAIDs alone, or conservativemanagement (Table 3). In studies inwhich ligation was performed early inlife (either as primary therapy or im-mediately after failure of indomethacinin the rst week of life), there was nodifference in mortality compared withmedically treated infants.12,84
If ligation truly improved survival (andsurvival bias was not a factor), thena reduction in rates of surgical treat-ment might be expected to increasemortality, assuming that the clinicalcharacteristics of infants remained thesame. Two studies14,17 reported nochange in mortality across epochs af-ter moving to a delayed selective liga-tion strategy from an early routineligation strategy after indomethacinfailure. However, other studies havereported an increase inmortality whensurgical ligation was no longer avail-able176 and when ligation was notperformed in infants with a persistentPDA after failure of medical therapy.177
This nding suggests that both survivalbias and a true survival benet ofsurgical ligation may be present.
We found that infants exposed to PDAligation also had increased CLD andsevere ROP. This is biologically sup-ported by the potential cardiorespira-tory instability and inammatory effectsassociated with surgical ligation178180
and was shown in a secondary analysis
e1040 WEISZ et al at Indonesia:AAP Sponsored on June 7, 2015pediatrics.aappublications.orgDownloaded from
-
of anRCTof early prophylactic ligation inextremely low birth weight infants.9,77
Given its association with NDI, this riskof CLD may represent a plausible path-way for the increased NDI associatedwith ligation.181183 Nonetheless, con-founding by indication remains an im-portant source of bias in these studiesas well. Only 1 study controlled forpostnatal sepsis,10 a known risk factorfor CLD,184 and this study did not dif-ferentiate between sepsis that occurredbefore or after surgical ligation. In ad-dition, ventilator dependence is com-monly considered an indication forligation in infants in whom medicaltherapy had failed or was contra-indicated,33,44,47,48,62 and yet it is alsoa risk factor for CLD.185 Given the asso-ciation between CLD and ROP and deathor NDI,181,182,186 it is important, whenanalyzing these outcomes, to adjust forpreligation respiratory morbidity.
Conclusions and Implications
This systematic reviewandmeta-analysisidentied an association between PDAligation and decreased odds of death;
increased odds of CLD, ROP, and NDI inearly childhood; and no difference in thecomposite outcome of death or NDI.However, this association comes pre-dominantly from observational studiesthat inadequatelyaddressedsurvivalbiasand confounding by indication. Many ofthese studies also lacked standardizedechocardiographicandclinical criteria todene a hemodynamically signicantPDA. Our review highlights the difcultyfaced by clinicians considering surgicalligation. The clinician must navigate lit-erature that reports an association withsignicant morbidity, albeit fraught withmethodologic biases and clinical un-certainty regarding patient selection andthe optimal timing for surgery.
This study provides direction to improvetheavailable evidence toguidecliniciansabout thePDA ligationdecision.Whereasan RCT examining 2 different PDA treat-ment protocolswould be instructive, thevariability in practice among centersmay reduce the external validity of sucha trial. Observational studies are there-fore needed that adjust for preligation
time-dependent covariates to fully elu-cidate the effects of PDA ligation.
Strengths and Limitations
This review encompasses a comprehen-sive search and explicit inclusion andexclusion criteria and poses clinicallyimportant questions. There was low tomoderate clinical and statistical hetero-geneity in most studies included in thisreview. This meta-analysis is limited by apaucity of studies that performed multi-variate analyses and the insufciency ofthese studies in addressing survival biasand confounding by indication for post-natal morbidities, such as ventilator depen-dence, IVH, sepsis, or NEC, that occurredbefore treatment with surgical ligation.
ACKNOWLEDGMENTSWe thank Ms Elizabeth Uleryk, BA MLS, Di-rectorHospital Library&Archives, TheHos-pital for Sick Children, for her assistancewith conducting the literature search andDrLuciaMireaMt. SinaiHospital, DrGregoryMoore (University of Ottawa), andDr Nicholas Barrowman (University of Ot-tawa) for providing additional study data.
REFERENCES
1. Sellmer A, Bjerre JV, Schmidt MR, et al.Morbidity and mortality in preterm neo-nates with patent ductus arteriosus onday 3. Arch Dis Child Fetal Neonatal Ed.2013;98(6):F505F510.
2. Giliberti P, De Leonibus C, Giordano L,Giliberti P. The physiopathology of thepatent ductus arteriosus. J Matern FetalNeonatal Med. 2009;22(suppl 3):69
3. Hamrick SE, Hansmann G. Patent ductusarteriosus of the preterm infant. Pediat-rics. 2010;125(5):10201030
4. Shimada S, Kasai T, Hoshi A, Murata A,Chida S. Cardiocirculatory effects ofpatent ductus arteriosus in extremelylow-birth-weight infants with respiratorydistress syndrome. Pediatr Int. 2003;45(3):255262
5. Sehgal A, Coombs P, Tan K, McNamara PJ.Spectral Doppler waveforms in systemicarteries and physiological signicance of
a patent ductus arteriosus. J Perinatol.2011;31(3):150156
6. Hammerman C, Bin-Nun A, Markovitch E,Schimmel MS, Kaplan M, Fink D. Ductalclosure with paracetamol: a surprisingnew approach to patent ductus arterio-sus treatment. Pediatrics. 2011;128(6).Available at: www.pediatrics.org/cgi/con-tent/full/128/6/e1618
7. Malviya MN, Ohlsson A, Shah SS. Surgicalversus medical treatment with cyclo-oxygenase inhibitors for symptomaticpatent ductus arteriosus in preterminfants. Cochrane Database Syst Rev.2013;3:CD003951
8. Mirea L, Sankaran K, Seshia M, et al; Ca-nadian Neonatal Network. Treatment ofpatent ductus arteriosus and neonatalmortality/morbidities: adjustment fortreatment selection bias. J Pediatr. 2012;161(4):689694, e1
9. Clyman R, Cassady G, Kirklin JK, Collins M,Philips JB III. The role of patent ductusarteriosus ligation in bronchopulmonarydysplasia: reexamining a randomizedcontrolled trial. J Pediatr. 2009;154(6):873876
10. Madan JC, Kendrick D, Hagadorn JI, FrantzID III; National Institute of Child Health andHuman Development Neonatal ResearchNetwork. Patent ductus arteriosus ther-apy: impact on neonatal and 18-monthoutcome. Pediatrics. 2009;123(2):674681
11. Kabra NS, Schmidt B, Roberts RS, DoyleLW, Papile L, Fanaroff A; Trial of In-domethacin Prophylaxis in PretermsInvestigators. Neurosensory impairmentafter surgical closure of patent ductusarteriosus in extremely low birth weightinfants: results from the Trial of In-domethacin Prophylaxis in Preterms. JPediatr. 2007;150(3):229234, e1
REVIEW ARTICLE
PEDIATRICS Volume 133, Number 4, April 2014 e1041 at Indonesia:AAP Sponsored on June 7, 2015pediatrics.aappublications.orgDownloaded from
-
12. Chorne N, Leonard C, Piecu