Comparison of randomized controlled trials discontinued or ... · The mean number of RCTs with poor...
Transcript of Comparison of randomized controlled trials discontinued or ... · The mean number of RCTs with poor...
RESEARCH Open Access
Comparison of randomized controlled trialsdiscontinued or revised for poorrecruitment and completed trials with thesame research question: a matchedqualitative studyMatthias Briel1,2* , Benjamin Speich1, Erik von Elm3 and Viktoria Gloy1
Abstract
Background: More than a quarter of randomized controlled trials (RCTs) are prematurely discontinued, mostly dueto poor recruitment of patients. In this study, we systematically compared RCTs discontinued or revised for poorrecruitment and completed RCTs with the same underlying research question to better understand the causes ofpoor recruitment, particularly related to methodological aspects and context-specific study settings.
Methods: We compared RCTs that were discontinued or revised for poor recruitment to RCTs that were completedas planned, matching in terms of population and intervention. Based on an existing sample of RCTs discontinued orrevised due to poor recruitment, we identified matching RCTs through a literature search for systematic reviewsthat cited the discontinued or revised RCT and matching completed RCTs without poor recruitment. Based onextracted data, we explored differences in the design, conduct, and study settings between RCTs with and withoutpoor recruitment, separately for each research question using semi-structured discussions.
Results: We identified 15 separate research questions with a total of 29 RCTs discontinued or revised for poorrecruitment and 48 RCTs completed as planned. Prominent research areas in the sample were cancer and acutecare. The mean number of RCTs with poor recruitment per research question was 1.9 ranging from 1 to 4suggesting clusters of research questions or settings prone to recruitment problems. The reporting quality of therecruitment process in RCT publications was generally low. We found that RCTs with poor recruitment often hadnarrower eligibility criteria, were investigator- rather than industry-sponsored, were associated with a higher burdenfor patients and recruiters, sometimes used outdated control interventions, and were often launched later in timethan RCTs without poor recruitment compromising uncertainty about tested interventions through emergingevidence. Whether a multi- or single-center setting was advantageous for patient recruitment seemed to dependon the research context.
(Continued on next page)
© The Author(s). 2019 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
* Correspondence: [email protected] Institute for Clinical Epidemiology and Biostatistics, Department ofClinical Research, University of Basel and University Hospital Basel, Basel,Switzerland2Department of Health Research Methods, Evidence, and Impact, McMasterUniversity, Hamilton, ON, CanadaFull list of author information is available at the end of the article
Briel et al. Trials (2019) 20:800 https://doi.org/10.1186/s13063-019-3957-4
(Continued from previous page)
Conclusions: Our study confirmed previously identified causes for poor recruitment, i.e., narrow eligibility criteria,investigator sponsorship, and a reduced motivation of patients and recruiters. Newly identified aspects were thatresearchers need to be aware of all other RCTs on a research question so that compromising effects on therecruitment can be minimized and that a larger number of centers is not always advantageous.
Keywords: randomized controlled trials , early termination of clinical trials, poor recruitment, systematic reviews,qualitative analysis
BackgroundEvidence-based health care relies on high-quality clinicalresearch. Randomized controlled trials (RCTs) are themethod of choice to assess preventive and therapeuticinterventions and are a cornerstone in the final phase ofdrug development and in comparative effectiveness re-search. Conducting high-quality RCTs, however, is chal-lenging. More than a quarter of RCTs do not reach theplanned sample size, mostly due to poor recruitment ofpatients, and are prematurely discontinued [1, 2].Investigator-initiated RCTs are particularly prone [1].Implications of poor patient recruitment and prematurediscontinuation of RCTs are that up to 70% of such tri-als remain unpublished, root causes of recruitment diffi-culties are not shared with the scientific community andmay therefore be repeated in the future, research ques-tions remain unanswered, and substantial amounts ofscarce research resources are wasted [1, 3].To better understand the causes of poor recruitment,
previous quantitative approaches using RCT protocolsand registry information [1, 2] as well as qualitative ana-lyses from published reports and semi-structured inter-views with trialists and other stakeholders in clinicalresearch have already provided important insights [3, 4].That is, for instance, that investigator-initiated RCTs orRCTs in the acute care setting were found to be at muchhigher risk for discontinuation due to poor recruitmentthan industry-sponsored RCTs or RCTs in non-acutecare settings [1, 2]; or that insufficient preparation,overly narrow eligibility criteria, and prejudiced views ofrecruiters and patients on trial interventions are com-mon reasons for poor recruitment [3]. As suggested byothers previously [5], we undertook a systematic com-parison of RCTs discontinued or revised for poor re-cruitment and RCTs completed as planned with thesame underlying research question. We aimed to provideadditional evidence on potential causes for poor recruit-ment specifically related to design aspects and context-specific study settings of RCTs.
MethodsThis is a matched qualitative study comparing RCTs thatdid not reach 90% of their originally planned sample sizedue to poor recruitment (cases; discontinued or revised
RCTs with poor recruitment) to RCTs completed asplanned (controls without poor recruitment) matchingin terms of patient population and experimental inter-vention. The study is reported according to the Stan-dards for Reporting Qualitative Research (SRQR, http://www.equator-network.org/reporting-guidelines/srqr/) asdescribed in Additional file 1.
Identification of discontinued and matching completedRCTsIn a previous study, we included 20 RCTs that were dis-continued or revised for poor recruitment with theplanned sample size being reported in a publication [1].For these studies, we aimed to find matching RCTsreaching at least 90% of their originally planned samplesize. We searched for systematic reviews that cited thediscontinued or revised RCT using the “times cited”function in Web of Science (times cited “view all of thearticles that cite this one” https://apps.webofknowledge.com) in January 2016. One reviewer (VG) screened thetitles, abstracts, and full texts of potentially eligible sys-tematic reviews for relevance. An eligible systematic re-view had to describe a literature search conducted in atleast one electronic database (e.g., Medline), includeRCTs, and had to have a research question similar tothat of the discontinued or revised RCT. If more thanone systematic review were eligible, we chose the mostup-to-date, comprehensive systematic review (frequentlya Cochrane review). If no systematic review was identi-fied, we searched by the same means for similar narra-tive reviews. In case VG was in doubt about theeligibility of a systematic review, she involved a secondreviewer (MB) for discussion and consensus decision.From each eligible systematic review we retrieved the
full text articles of included RCTs (i.e., potentially eli-gible matching RCTs) and collected a small set of pre-liminary data. These included whether the RCT wasdiscontinued or revised for poor recruitment, plannedand actually achieved sample size, the patient popula-tion, and the experimental and control intervention.This was done by two methodologically trained investi-gators (VG, BS) working independently and in duplicate.Any disagreements were resolved by consensus and, ifneeded, through involving another investigator (MB).
Briel et al. Trials (2019) 20:800 Page 2 of 12
The matching criteria (inclusion criteria) were the testedintervention and included patient population, whichneeded to be similar enough so that the RCTs could beincluded in the same meta-analysis. Other trial charac-teristics such as comparator interventions, outcomes ofinterest, and trial settings already qualified as factors po-tentially associated with poor recruitment.
Data collectionThe following data were collected from the eventually in-cluded RCTs: study design (e.g., superiority or non-inferiority trial, factorial or parallel design, allocation ratio),sample size calculation, allocation concealment, blinding,reporting quality of the recruitment process, eligibility cri-teria, trial sponsor, country and place of patient recruitment,recruitment period, reporting of recruitment networks, sup-port from a clinical trial unit or contract researchorganization, study population, interventions, comparators,and primary outcome. In addition, we also extracted self-reported reasons (if any) for poor recruitment of thediscontinued or revised RCTs. Data were collected by one in-vestigator (VG or BS) and checked by another (VG, BS orMB). Any disagreements were resolved by consensus.
Data analysisWe explored differences in the design, conduct. and studysettings between RCTs with and without poor recruit-ment, separately for each research question, thus context-specific, using semi-structured discussions (MB, BS, andVG). In a first round, we gathered differences betweenRCTs with and without poor recruitment, based on ourextracted data; we systematically went through a pre-specified checklist of items potentially associated withpoor recruitment, including eligibility criteria, trial spon-sor, single versus multiple centers, etc. (Additional file 2);and discussed each item in turn followed by observationsbeyond the checklist items (e.g., the chronology of RCTsbuilding up the evidence base for a specific research ques-tion). When we reached agreement among us, the identi-fied differences and potentially relevant observations werecaptured by VG in free-text form. In a second round, wereflected on the identified differences between RCTs withand without poor recruitment and other relevant observa-tions for each research question and integrated the rele-vant information in order to come up with an explanationwhy for a certain research question one or more RCTshad serious recruitment problems while others did nothave such problems. In addition, we considered the self-reported reasons for poor recruitment (if any).
Researchers’ reflexivityThe three researchers who carried out the analysis have di-verse disciplinary background and training such as medicine/clinical epidemiology (MB), nutrition/health technology
assessment (VG), and epidemiology/public health (BS). Tobetter understand specific characteristics of oncology trials,we involved a practicing oncologist in the respective discus-sions. For the other topics, the analysis team members feltthat they had sufficient knowledge to assess trial characteris-tics, and their methodological expertise with clinical trialslikely strengthened the analysis. None of us knew any of theincluded RCTs or their investigators. During analysis, all re-searchers worked together as a team and extensively dis-cussed the data interpretation to minimize bias.
ResultsStudy sampleIn a previous study, there were 20 published RCTs discon-tinued or revised for poor recruitment that reported theplanned sample size [1]. For five of these RCTs our litera-ture search could not identify a systematic or narrative re-view that cited the discontinued or revised RCT (Fig. 1).For the remaining 15 RCTs discontinued or revised due topoor recruitment representing 15 different research ques-tions, we identified 110 matching RCTs. We excluded 48RCTs because the planned sample size was not reported(n = 31); the RCT was discontinued for another reasonthan poor recruitment (n = 9); the article was not pub-lished in English (n = 3); the research question was notsimilar enough (n = 2); there was no full-text publicationavailable (n = 2); or the same results were published mul-tiple times (n = 1). Of the 62 matching RCTs, 48 werecompleted as planned and 14 were newly identified RCTsdiscontinued or revised for poor recruitment. Hence, for15 separate research questions a total of 77 RCTs (29 withand 48 without poor recruitment) were available for com-parative analyses. Twenty-five of the 29 RCTs with poorrecruitment explicitly reported that they were discontin-ued due to recruitment problems; the remaining fourRCTs [6–9] had recruitment problems but revised theiroriginal target sample sizes during the trial (reduction ofthe originally planned number of patients in each trial byabout 50%) and met the revised targets. References of allincluded RCTs are provided in Additional file 3.
Research questions, recruitment, and reporting qualityMedical areas of the 15 included research questions werecancer research (n = 5), research in acute care (n = 8; in-cluding preterm infants [n = 2] and laboring women [n =1]), surgery (n = 1), and infectious diseases (n = 1). Themean number of RCTs with poor recruitment was 1.9 perresearch question ranging from 1 to 4. Table 1 summa-rizes the recruitment characteristics for the RCTs withand without poor recruitment across all research ques-tions. In general, RCTs with poor recruitment recruitedfewer patients per year per recruiting center compared toRCTs that were completed as planned. More detailed re-cruitment characteristics for each research question and
Briel et al. Trials (2019) 20:800 Page 3 of 12
for all included RCTs, as well as general study characteris-tics are provided in Additional files 4, 5 and 6.The reporting of the recruitment process was generally
in included RCT publications with little detail. None ofthe articles reported on who actually recruited patientsor the anticipated prevalence of eligible patients. Only6% (5/77) of the RCTs reported on the anticipated re-cruitment duration, 51% (39/77) reported the locationwhere patients were recruited, 27% (21/77) provided adetailed patient flow, and 90% (69/77) reported the ac-tual recruitment period or duration (Additional file 7).
Comparison of RCTs with and without poor recruitmentTable 2 summarizes the differences observed betweenRCTs with and without poor recruitment for each researchquestion as well as our context-specific conclusions on thepossible reasons for poor recruitment. The most recurrenttheme across research questions was that, in RCTs withpoor recruitment, eligibility criteria were substantially nar-rower than in RCTs without poor recruitment (researchquestions #1, #2, #3, #4, #9, #12, #13 in Table 2).There was no consistent pattern as to whether an
international or national multicenter setting or a single-
Fig. 1 Selection of included randomized controlled trials. *Five of the original 20 discontinued RCTs for poor recruitment that were publishedand reported a sample size calculation were excluded, because we did not identify a systematic review or narrative review citing the RCT.**Including four RCTs with recruitment problems but revised original target sample sizes during the trial (reduction of the originally plannednumber of patients in each trial by about 50%) and meeting their revised targets. RCT randomized controlled trial
Table 1 Recruitment characteristics of included randomized controlled trials with and without poor recruitment across researchquestions
RCTs with poor recruitment, n = 29 RCTs without poorrecruitment, n = 48
Originally planned number of patients (median; IQR) 395 (72–600) 272 (22–500)
Number of patients randomized (median; IQR) 150 (41–329) 306 (23–559)
Duration of recruitment period in months (median; IQR) 39 (10–47) 24 (7–45)
Number of patients recruited per year and centera (median; IQR) 3 (1–5) 7 (1–25)
Trial sponsor (industry n (%) / investigator n (%)) 12 (41%) / 17 (59%) 20 (42%) / 27 (56%) /NR 1 (2%)
IQR interquartile range (25th and 75th percentile), NR not reported, RCT randomized controlled trialaRough estimate for recruitment speed based on own calculations (number of patients recruited divided by recruitment duration and number of study centers);not adjusted to the time a site was actually open for recruitment, because this was not reported in the publications of included trials)
Briel et al. Trials (2019) 20:800 Page 4 of 12
Table
2Differen
cesbe
tweenrand
omized
controlledtrialswith
andwith
outpo
orrecruitm
entandconclusion
son
thepo
ssiblereason
sforpo
orrecruitm
ent
Group
sof
RCTs
with
acommon
research
questio
nObserveddifferences
with
respectto
stud
ycharacteristics
(design,cond
uct,spon
sor)be
tweenRC
Tswith
andwith
outpo
orrecruitm
ent
Con
text-spe
cific
conclusion
son
thepo
ssiblereason
sforpo
orrecruitm
ent
#1:Surgicaltreatmen
tcomparedto
proton
-pum
pinhibitorsin
gastroesop
hage
alrefluxdisease
RCTs
with
poor
recruitm
entn=2
RCTs
with
outpo
orrecruitm
entn=2
1)Stud
yspon
sor:RC
Tswith
poor
recruitm
entwereinvestigator-
spon
soredwhe
reas
RCTs
with
outpo
orrecruitm
entwere
indu
stry-spo
nsored
Absen
ceof
indu
stry
spon
sorin
gcouldmeanless
profession
altrial
organizatio
nandlim
itedfund
ingin
gene
ral;narrow
ereligibility
criteria
may
correlatewith
overestim
ationof
recruitm
entrates
andwith
slow
recruitm
ent;andahigh
erbu
rden
forpatients
durin
gfollow-upmay
have
lowered
themotivationof
reflux
patientsto
participate
2)Eligibility
criteria:m
orerestrictivein
theRC
Tswith
poor
recruitm
entwith
respectto
diseasedu
ratio
nanddu
ratio
nof
priorproton
-pum
pinhibitorusage
3)Patient
burden
durin
gfollow-up:
patient
assessmen
tdu
ring
follow-upmorefre
quen
tin
oneof
theRC
Tswith
poor
recruit-
men
t(Anvarietal.2011)
than
intheRC
Tswith
outpo
orrecruit-
men
t(e.g.,assessmen
tof
quality
oflifeevery3mon
thsvs.every
12mon
ths)
#2:A
nthracyclines
with
orwith
outanytaxane
scomparedto
anthracyclines
plus
anyanticancertreatm
entor
comparedto
sing
le-age
nttaxane
sin
metastatic
breastcancer
RCTs
with
poor
recruitm
entn=3
RCTs
with
outpo
orrecruitm
entn=3
1)National/internatio
nal:RC
Tswith
poor
recruitm
entwere
cond
uctedin
onecoun
try(national)whe
reas
theRC
Tswith
out
poor
recruitm
entwereinternational(1no
trepo
rted
)
Absen
ceof
aninternationaln
etworkof
recruitin
gcentersmay
beassociated
with
limitedrecruitm
entcapacitiesforpatients
with
metastatic
breastcancer;narrower
eligibility
criteria
may
correlatewith
overestim
ated
recruitm
entratesandwith
slow
recruitm
ent;andthefact
that
eviden
ceon
theeffectiven
ess
ofanthracyclinecombinatio
ntherapyfro
mprevious
RCTs
was
alreadyavailablemay
have
comprom
ised
themotivation
ofrecruitersin
theRC
Tswith
poor
recruitm
ent
2)Eligibility
criteria:m
orerestrictivein
2of
3RC
Tswith
poor
recruitm
ent(Bon
neterreet
al.2004andBo
nten
baletal.2005)
with
respectto
adjuvant
therapypriorto
enrolm
ent,i.e.,adjuvant
therapyhadto
bestop
pedat
least12
mon
thsbe
fore
3)Pu
blicationchrono
logy/availableeviden
ce:RCTs
with
poor
recruitm
entwerepu
blishe
d(years:2004,2005,and
2010)after2
ofthe3RC
Tswith
outpo
orrecruitm
ent(Biganzoliet
al.2002and
Jassem
etal.2001);the
othe
rRC
Twith
outpo
orrecruitm
entwas
publishe
din
2005
(Nabho
ltzet
al.2005)
#3:First-linetreatm
entwith
thearom
ataseinhibitorexem
estane
comparedto
anastrozolein
postmen
opausalw
omen
with
ad-
vanced
breastcancer
RCTs
with
poor
recruitm
entn=1
RCTs
with
outpo
orrecruitm
entn=1
Eligibility
criteria:m
orerestrictivein
theRC
Twith
poor
recruitm
entwith
respectto
metastasesanddiseasestage
(visceralm
etastaseson
lyvs.advancedstageor
anymetastasesin
theRC
Twith
outpo
orrecruitm
ent)
Narrower
eligibility
criteria
forpo
stmen
opausalw
omen
with
advanced
breastcancer
may
correlatewith
overestim
ated
recruitm
entratesandwith
slow
recruitm
ent
#4:A
ntiarrhythmicagen
tscomparedto
placeb
oor
backgrou
ndtherapywith
beta-blocker
inventricular
arrhythm
iaRC
Tswith
poor
recruitm
entn=1
RCTs
with
outpo
orrecruitm
ent
n=2
1)Interven
tiontested
:RCTs
with
outpo
orrecruitm
enttested
ane
winterven
tion,i.e.,celivaron
e(Kow
eyet
al.2011)
orazim
ilide
(Dorianet
al.2004)
whe
reas
theRC
Twith
poor
recruitm
ent[8]
tested
availabledrug
s(beta-blocker,am
iodarone
,and
sotalol)
Thefact
that
alreadyavailableantiarrhythmicdrug
sandno
new
drug
sweretested
may
have
comprom
ised
themotivation
ofrecruitersor
patients;recruitm
entcapacitiesmight
have
been
insufficien
tdu
eto
toofew
stud
ycenters;andnarrow
ereligibility
criteria
forpatientswith
ventricular
arrhythm
iamay
correlate
with
overestim
ated
recruitm
entratesandwith
slow
recruitm
ent
2)Num
berof
stud
ycenters:3to
4tim
eslower
inRC
Twith
poor
recruitm
ent
3)Eligibility
criteria:m
orerestrictivein
theRC
Twith
poor
recruitm
ent[8]with
respectto
usageof
antiarrhythmicagen
tspriorto
enrolm
entandtheim
plantablecardioverter
defib
rillator
device
(onlyon
especifictype
allowed
)
#5:Proph
ylactic
antib
ioticscomparedto
placeb
oor
usualcare
inacutene
crotizingpancreatitis
RCTs
with
poor
recruitm
entn=3
RCTs
with
outpo
orrecruitm
entn=0
Nodifferences
observed
becauseon
lyRC
Tswith
poor
recruitm
entwereiden
tified
Thefact
that
3RC
Tswerediscon
tinueddu
eto
poor
recruitm
ent
inthisresearch
area
sugg
eststhat
itseem
sge
nerally
prob
lematic
torecruitpatientswith
acutene
crotizingpancreatitis,which
isassociated
with
ahigh
mortalityrate
(vulne
rablepatients)
Briel et al. Trials (2019) 20:800 Page 5 of 12
Table
2Differen
cesbe
tweenrand
omized
controlledtrialswith
andwith
outpo
orrecruitm
entandconclusion
son
thepo
ssiblereason
sforpo
orrecruitm
ent(Con
tinued)
Group
sof
RCTs
with
acommon
research
questio
nObserveddifferences
with
respectto
stud
ycharacteristics
(design,cond
uct,spon
sor)be
tweenRC
Tswith
andwith
outpo
orrecruitm
ent
Con
text-spe
cific
conclusion
son
thepo
ssiblereason
sforpo
orrecruitm
ent
#6:Latepo
stnatal(>7days)corticosteroid
treatm
entcompared
toplaceb
oor
usualcarein
preterm
neon
ates
RCTs
with
poor
recruitm
entn=3
RCTs
with
outpo
orrecruitm
entn=7
1)Num
berof
stud
ycenters:allR
CTs
with
outpo
orrecruitm
ent
weresing
le-cen
terstud
ies,whe
reas
2(Karietal.1993andDoyle
etal.2006)
ofthe3RC
Tswith
poor
recruitm
entweremulticen
ter
stud
ies
Thefact
that
3RC
Tsrecruitedpo
orlyin
thisresearch
area
sugg
eststhat
itseem
sge
nerally
prob
lematicto
recruitpreterm
neon
ates
(vulne
rablepatients).Inasettingwith
vulnerable
patients,asing
le-cen
terRC
Tmay
proveadvantageo
usdu
eto
moremotivated
andbe
tter
prep
ared
/trained
recruitersand
closer
mon
itorin
gof
recruitm
ent).
Particularlyprob
lematic
appe
ared
thechoice
ofprim
aryou
tcom
ein
Doyleet
al.2006;
tofocuson
apo
tentialsideeffect
oftheinterven
tionmay
have
comprom
ised
themotivationof
recruitersandparentsto
includ
echildrenin
such
anRC
T.Equipo
isemay
becomprom
ised
forRC
Tscond
uctedlaterin
agrou
pof
RCTs
addressing
thesameresearch
questio
n(leadingto
lower
motivationof
recruitersandparents),b
ecause
eviden
ceon
the
effectiven
essof
aninterven
tionfro
mprevious
RCTs
was
already
available
2)Prim
aryou
tcom
e:theprim
aryou
tcom
echosen
inon
eRC
Twith
poor
recruitm
ent(Doyleet
al.2006)
focusedon
apo
tential
side
effect
oftheinterven
tion(m
ajor
neurosen
sory
disability),
whe
reas
theothe
rtrialshadan
effectiven
essou
tcom
e
3)Pu
blicationchrono
logy/availableeviden
ce:a
largeRC
Twith
poor
recruitm
ent(Doyleet
al.2006)
was
thelastpu
blishe
dRC
Tin
thisarea
ofresearch
whe
reas
allo
ther
RCTs
(discontinuedor
completed
)werepu
blishe
dearlier
andalreadyshow
edbe
nefits
interm
sof
better
respiratio
nof
infants
#7:Ven
tilatorygaswith
nitricoxidecomparedto
ventilatorgas
with
outnitricoxidein
preterm
neon
ates
RCTs
with
poor
recruitm
entn=2
RCTs
with
outpo
orrecruitm
entn=5
1)Num
berof
stud
ycenters:Thenu
mbe
rof
stud
ycenterswas
consistentlysm
alleram
ongtheRC
Tswith
outpo
orrecruitm
ent
Inasettingwith
vulnerablepatients(preterm
neon
ates)a
smallernu
mbe
rof
stud
ycenterswith
in1coun
trymay
prove
advantageo
usdu
eto
less
complex
trialo
rganizationand
recruitm
entmon
itorin
gandpo
tentially
moremotivated
and
better
prep
ared
/trained
recruiters.
2)National/internatio
nal:whe
reas
allR
CTs
with
poor
recruitm
ent
wereinternationalstudies,allbu
ton
eRC
Twith
outpo
orrecruitm
ent(Hascoet
etal.2005)
hadanatio
nalsettin
g
#8:Prim
aryangiop
lastycomparedto
on-site
thrombo
lytic
ther-
apyin
acutemyocardialinfarction(with
in12
h)RC
Tswith
poor
recruitm
entn=3
RCTs
with
outpo
orrecruitm
entn=3
Target
samplesize:the
numbe
rof
patientsplanne
dto
been
rolledwas
consistentlyhigh
erin
RCTs
with
poor
recruitm
ent
Thefact
that
3RC
Tsrecruitedpo
orlyin
thisresearch
area
sugg
eststhat
itseem
sge
nerally
prob
lematicto
recruitacute
care
patients.Thecomparison
ofangiop
lasty(rapidpatient
transfer
tocenter
necessary)to
on-site
thrombo
lytic
therapy
was
logisticallyvery
challeng
ingat
thetim
e.AllRC
Tswith
poor
recruitm
entweremeant
toinclud
emore
patientsthan
thecompleted
RCTs,con
sequ
ently
requ
iring
morecentersandofteninternationalcollabo
ratio
n,which
additio
nally
increasedchalleng
eswith
triallog
istics
#9:M
oxifloxacin
comparedto
othe
rantib
ioticsin
patientswith
pneumon
iaRC
Tswith
poor
recruitm
entn=2
RCTs
with
outpo
orrecruitm
entn=4
1)Patient
popu
latio
n:bo
thRC
Tswith
poor
recruitm
entinclud
edpatientswith
hospital-acquiredpn
eumon
ia,atleastin
partmech-
anicallyventilated(poten
tially
sicker/m
orevulnerablepatients),
whileRC
Tswith
outpo
orrecruitm
entinclud
edpatientswith
commun
ity-acquiredpn
eumon
iawith
outmechanicalven
tilation
Recruitm
entof
sicker/m
echanically
ventilatedpatients
appe
ared
morechalleng
ing(inform
edconsen
tetc.);
patientswho
acqu
iredpn
eumon
iain
aho
spitalm
ayhave
been
less
motivated
toparticipatein
anRC
Tcond
uctedat
that
hospital,particularlyifseverelyaffected
(mechanically
ventilated)
Inadditio
n,all6
RCTs
werede
sign
edas
non-inferio
ritytrials,i.e.,thetested
antib
ioticswereno
thypo
thesized
tohave
supe
rioreffectsbu
tmay
have
been
cheape
ror
easier
toapply,so
patientswith
hospital-acquired
pneumon
iacouldno
texpe
ctamoreefficacious
treatm
ent
andmight
have
thereforebe
enless
motivated
toparticipate.
Forthesamereason
s(sickerpatients,ho
spital-acquired
infection,no
n-inferio
rityde
sign
)recruitersmay
have
been
less
motivated
,too
.Theeligibility
criteria
requ
iredtheapplicationof
acomplex
scoringsystem
(e.g.,APA
CHEscore)
tode
term
inepatient
eligibility,thismay
have
increasedthebu
rden
forrecruiters
2)Eligibility
criteria:o
neof
theRC
Tswith
poor
recruitm
entused
theAPA
CHE(Acute
Physiology
And
Chron
icHealth
Evaluatio
n)scoreto
defineeligiblepatients(Höffken
etal.the
RCTs
with
out
poor
recruitm
entdidno
t
Briel et al. Trials (2019) 20:800 Page 6 of 12
Table
2Differen
cesbe
tweenrand
omized
controlledtrialswith
andwith
outpo
orrecruitm
entandconclusion
son
thepo
ssiblereason
sforpo
orrecruitm
ent(Con
tinued)
Group
sof
RCTs
with
acommon
research
questio
nObserveddifferences
with
respectto
stud
ycharacteristics
(design,cond
uct,spon
sor)be
tweenRC
Tswith
andwith
outpo
orrecruitm
ent
Con
text-spe
cific
conclusion
son
thepo
ssiblereason
sforpo
orrecruitm
ent
andredu
cedtheirmotivation
#10:Temozolom
ide(che
mothe
rape
uticagen
t)alon
eor
incombinatio
nwith
radiothe
rapy
comparedto
nochem
othe
rapy
(e.g.,radiothe
rapy),no
n-temozolom
ide-basedchem
othe
rapy
ortemozolom
ideat
different
dosesin
gliomapatients
RCTs
with
poor
recruitm
entn=1
RCTs
with
outpo
orrecruitm
entn=3
Inthisgrou
pof
RCTs
with
gliomapatients,alltrialswere
publishe
dafterthecompleted
andpu
blishe
dRC
Tcond
ucted
byStup
pet
al.2005.Subseq
uent
tothispu
blication,itwas
prob
ablydifficultto
recruitpatientsin
RCTs
that
didno
tinclud
etheinterven
tionthat
was
show
nto
besupe
riorin
Stup
pet
al.2005(com
binatio
nof
radiothe
rapy
and
chem
othe
rapy).How
ever,neither
theRC
Twith
poor
recruitm
ent(M
almström
etal.2012)
noranothe
rRC
Twith
outpo
orrecruitm
ent(W
icket
al.2012)
includ
edsuch
acombinatio
ntherapy,bo
thRC
Tswerevery
similar
with
respectto
patient
popu
latio
n(elderly)andcompared
interven
tions
(tem
ozolom
idevs.stand
ardradiothe
rapy).
Theon
lyothe
rRC
Twith
outpo
orrecruitm
entcompared
temozolom
ideto
anothe
rchem
othe
rape
utictreatm
entin
patientsno
tsuitableforradiothe
rapy
(Brada
etal.2010)
#11:Capecitabine
-based
chem
othe
rapy
comparedto
non-
cape
citabine
chem
othe
rapy
inmetastatic
breastcancer
RCTs
with
poor
recruitm
entn=4
RCTs
with
outpo
orrecruitm
entn=5
Interven
tiontested
:excep
tfor1RC
Twith
poor
recruitm
ent
(Bache
lotet
al.2011)
theRC
Tswith
poor
recruitm
entallocated
patientsto
either
anoralmon
othe
rapy
with
cape
citabine
oran
intraven
ousapplicationof
acape
citabine
-free
chem
othe
rapy
ei-
ther
asmon
othe
rapy
oras
combinatio
ntherapy.In
contrast,inall
but1RC
Twith
outpo
orrecruitm
ent(O’Shaug
hnessy
etal.2001),
therouteof
administrationwas
iden
ticalforbo
thtreatm
entarms
(alwaysinclud
edan
intraven
ousapplication)
andcape
citabine
was
(with
thesameexceptionof
O′Shaug
hnessy
etal.2001)
givenas
combinatio
ntherapy.TheRC
Twith
outpo
orrecruitm
ent
O′Shaug
hnessy
etal.2001,similarto
theRC
Tswith
poor
recruit-
men
talso
allocatedpatientseither
tointraven
ousor
oraltreat-
men
t,bu
tat
thetim
eof
trialcon
duct,oralm
onothe
rapy
with
cape
citabine
was
relativelyne
wanda2:1rand
omizationwas
done
tomakeitmorelikelyforpatientsto
getallocatedto
the
anticipated
patients’preferredtreatm
entop
tion
Patients’preferen
cesregardingtheform
ofapplication
might
have
played
arolein
thisresearch
area,i.e.,patient
recruitm
entin
RCTs
comparin
gdifferent
interven
tions
that,
inadditio
n,usedifferent
routes
ofapplicationishampe
red
ifpatientshave
preferen
cesabou
ttherouteof
application
andthereforeareless
willingto
berand
omized
.Tw
oothe
rreason
sforpo
orrecruitm
entrelatedto
patients’
orrecruiterpreferen
cesarepo
ssible,b
utcouldbe
specific
fortherespectiveRC
T.(1)In
1RC
Twith
poor
recruitm
ent
(Talbo
tet
al.2002)
thetested
drug
was
appliedcontinuo
usly
whe
reas
interm
itten
tlymight
have
been
preferredby
patients.
(2)In
anothe
rRC
Twith
poor
recruitm
ent(Stockleret
al.2011)
thecontroltreatmen
t(cycloph
osph
amide,metho
trexate,and
fluorou
racil,CMF)
was,atthetim
e,ou
tdated
andtherefore
themotivationof
recruitersandpatientsmight
have
been
redu
ced.
Inthediscussion
authorsstated
:“Themain
limitatio
nsof
ourtrialare
theam
bigu
ityof
broad,
pragmatic
eligibility
criteria
andtheuseof
anun
fashionablecontrol
regimen
.”(Stockler
etal.2011)
#12:Prim
arythrombo
prop
hylaxis(hep
arin)comparedto
placeb
oor
usualcarein
ambu
latory
cancer
patientsreceiving
chem
othe
rapy
RCTs
with
poor
recruitm
entn=2
RCTs
with
outpo
orrecruitm
entn=2
1)Stud
yspon
sor:bo
thRC
Tswith
poor
recruitm
entwere
investigator
initiated
whe
reas
both
RCTs
with
outpo
orrecruitm
entwereindu
stry-spo
nsored
Testingthrombo
prop
hylaxiswith
heparin
incancer
patientsmay
bechalleng
ingin
gene
ral,be
causetheinterven
tionwas
not
focusedon
combatin
gthecancer,and
thereforerecruiters’and
patients’interestin
such
anRC
Tmay
beredu
cedandseen
asbu
rden
;com
petin
gRC
Tstestingantitum
oragen
tsmight
bepreferred.
Inadditio
n,theabsenceof
indu
stry
spon
sorin
gcouldmeanless
profession
altrialo
rganizationandlim
ited
fund
ingin
gene
ral,andafocuson
only1type
ofcancer
(narrow
eligibility
criteria,Perry
etal.2002)
may
have
contrib
uted
topo
orrecruitm
ent
2)Eligibility
criteria:o
nlygliomapatientswereeligiblein
1of
the
RCTs
with
poor
recruitm
ent(Perry
etal.2002)
whilebo
thRC
Tswith
outpo
orrecruitm
entinclud
edpatientswith
vario
ustype
sof
cancer
Briel et al. Trials (2019) 20:800 Page 7 of 12
Table
2Differen
cesbe
tweenrand
omized
controlledtrialswith
andwith
outpo
orrecruitm
entandconclusion
son
thepo
ssiblereason
sforpo
orrecruitm
ent(Con
tinued)
Group
sof
RCTs
with
acommon
research
questio
nObserveddifferences
with
respectto
stud
ycharacteristics
(design,cond
uct,spon
sor)be
tweenRC
Tswith
andwith
outpo
orrecruitm
ent
Con
text-spe
cific
conclusion
son
thepo
ssiblereason
sforpo
orrecruitm
ent
#13:Recombinant
tissueplasminog
enactivator
(rtPA
)compared
toplaceb
oin
acuteischem
icstroke
RCTs
with
poor
recruitm
entn=1
RCTs
with
outpo
orrecruitm
entn=5
1)Patient
popu
latio
nandeligibility
criteria:the
RCTwith
poor
recruitm
entfocusedon
patientswho
didno
tfulfilthe
licen
secriteria
ofrtPA
(i.e.,elderlypatients>80
years),w
hereas
RCTs
with
outpo
orrecruitm
entdid(fo
cuson
patients<80
years)
Thefact
that
theRC
Twith
poor
recruitm
entfocusedon
elde
rlypatients(>
80years;i.e.,narrow
eligibility
criteria,vulne
rable
patients)to
berecruitedin
anacutecare
settingmay
have
contrib
uted
topo
orrecruitm
ent.Themorecomplex
subg
roup
hypo
theses
intheRC
Twith
poor
recruitm
entcouldhave
made
thetrialm
oredifficultto
explainto
colleaguesandpatients.
Finally,itispo
ssiblethat
themotivationof
recruitersand
patientswas
redu
ced,
becauseeviden
ceon
theeffectiven
ess
from
previous
RCTs
was
alreadyavailable(com
prom
ised
equipo
ise)
2)Target
samplesize:the
numbe
rof
patientsplanne
dto
been
rolledwas
high
erin
theRC
Twith
poor
recruitm
entdu
eto
planne
dsubg
roup
analyses
(morecomplex
design
/hypothe
ses)
3)Pu
blicationchrono
logy/availableeviden
ce:the
RCTwith
poor
recruitm
ent(Sande
rcocket
al.2012)
was
thelastRC
Tcond
ucted
forthisresearch
questio
nto
date
#14:Transdermalnitrog
lycerin
comparedto
placeb
oor
usual
care
inlabo
ringwom
en(gestatio
nalage
betw
een24
and32
weeks)
RCTs
with
poor
recruitm
entn=1
RCTs
with
outpo
orrecruitm
entn=1
1)Patient
burden
durin
gfollow-up:
long
erfollow-upwith
more
complex
assessmen
tsof
thene
onates
intheRC
Twith
poor
recruitm
ent
Labo
ringwom
enin
both
RCTs
hadto
beat
high
risk
ofpreterm
deliverybu
talso
nottooadvanced
sothat
they
couldstillbe
rand
omized
;how
ever,the
RCTwith
poor
recruitm
entim
posedahigh
erbu
rden
onpatientswith
long
erfollow-upandamorecomplex
assessmen
tof
thene
onates,
prob
ablyredu
cing
themotivationof
patientsandrecruitersto
participate;in
contrast,the
prim
aryou
tcom
e(i.e.,num
berof
days
from
rand
omizationto
delivery)in
theRC
Twith
outpo
orrecruitm
entdidno
trequ
ireanyeffortfro
mpatients.Afew
well-p
ickedstud
ycentersmay
have
been
advantageo
usin
thisacutecare
setting.
Themotivationof
recruitersand
patientswas
prob
ablyfurthe
rredu
cedin
thetrialw
ithpo
orrecruitm
ent,whe
neviden
ceon
theeffectiven
essof
the
tested
interven
tion(nitrog
lycerin
)fro
motherRC
Tsbe
came
available(com
prom
ised
equipo
ise,Sm
ithet
al.2007).
Only2RC
Tswereconsidered
inthisgrou
pof
RCTs,b
ecause
themajority
(n=7)
ofRC
Tsinclud
edin
thesamesystem
atic
review
didno
trepo
rtanysamplesize
calculation/target
samplesize
potentially
hiding
recruitm
entprob
lemsin
thesetrials
2)Prim
aryou
tcom
e:theRC
Twith
poor
recruitm
enthadamore
complex
prim
aryou
tcom
e(neo
natalm
orbidity
associated
with
long
-term
morbidity
andpe
rinatalmortalityvs.num
berof
days
from
rand
omizationto
delivery)
3)Num
berof
stud
ycenters:theRC
Twith
poor
recruitm
enthad
over
3tim
esmorestud
ycentersthan
theRC
Twith
outpo
orrecruitm
ent
4)Pu
blicationchrono
logy/availableeviden
ce:the
RCTwith
poor
recruitm
entwas
stillon
goingwhe
ntheRC
Twith
outpo
orrecruitm
entwas
publishe
d
#15:Vasopressin-containing
regimen
comparedto
epinep
hrine
incardiacarrest
RCTs
with
poor
recruitm
entn=1
RCTs
with
outpo
orrecruitm
entn=5
Num
berof
stud
ycenters:theRC
Twith
poor
recruitm
entwas
amulticen
terRC
Twhe
reas
allb
uton
eRC
Twith
outpo
orrecruitm
ent(Gueug
niaudet
al.2008)
weresing
le-center
RCTs.
Inthisacutecare
settingwith
vulnerablepatientsasing
lecenter
RCTmay
beadvantageo
usdu
eto
high
ermotivation
andbe
tter
prep
ared
/trained
recruitersandtrialteam
and
closer
mon
itorin
gof
recruitm
ent).
TheRC
Twith
poor
recruitm
entwas
cond
uctedin
44commun
ity-based
emerge
ncyroom
s.One
othe
rRC
Twith
outpo
orrecruitm
ent(Gueug
niaudet
al.2008)
hadasimilar
numbe
rof
stud
ycentersas
thediscon
tinuedRC
T,bu
tauthorsrepo
rted
that
thene
tworkof
centerswas
well
stablishe
dandexpe
rienced
inrunn
ingacutecare
RCTs.
RCTrand
omized
controlledtrial;Allof
thearticlescitedin
Table2arereferenced
inAdd
ition
alfile3
Briel et al. Trials (2019) 20:800 Page 8 of 12
center setting was advantageous for patient recruitmentin RCTs. In research question #4, for instance, investi-gating antiarrhythmic drugs, the RCT with poor recruit-ment had three to four times fewer study centers thanRCTs without poor recruitment; or in research area #2(metastatic breast cancer therapy) the RCTs with poorrecruitment were all restricted to a national setting,while the RCTs without poor recruitment were all donein large international collaborations. On the other hand,single-center RCTs or settings with only a few, carefullychosen study centers may have worked better in settingswith particular logistical challenges (e.g., question #8 onprimary angioplasty versus onsite thrombolysis andquestion #15 testing therapies for resuscitation) or theinclusion of particularly vulnerable patients (e.g., re-search questions #6 and #7 focusing on the recruitmentof preterm neonates) in the absence of well-establishedand experienced trial networks.We observed that investigator-sponsored RCTs are as-
sociated with a higher risk for poor recruitment thanindustry-sponsored RCTs; in research questions #1 and#12 all RCTs with poor recruitment were investigator-sponsored, while all RCTs without poor recruitmentwere industry-sponsored.The chronology of RCTs, e.g., when a RCT is launched
while other RCTs on the same research question havealready been completed, appears to impact on recruit-ment. RCTs with poor recruitment were often initiatedand published later, after other RCTs had already beensuccessfully completed (research questions #2, #6, #10,#13, #14); i.e., evidence on the potential benefits andharms of a certain intervention was already available atsome point during the conduct of such RCTs. This mayhave compromised the uncertainty about the testedtreatments as an ethical precondition (equipoise), andthe motivation of patients and recruiters for furtherrandomization. In research questions #10 and #11 thecontrol intervention(s) were already outdated, when theRCTs with poor recruitment were launched. In addition,in some RCTs with poor recruitment the burden for pa-tients, such as numerous or invasive assessments duringfollow-up (research questions #1, #12, #14), or the burdenfor recruiters, such as the need to apply a complex scoringsystem in order to include patients (research question #9),was higher than in corresponding RCTs without poor re-cruitment. Furthermore, it happened that the tested inter-ventions (already available drugs in RCTs with poorrecruitment versus new drugs in RCTs without poor re-cruitment; research question #4) or the study design withside effects of the experimental drug being the primaryoutcome (research question #6) were less attractive inRCTs with poor recruitment than in corresponding RCTswithout poor recruitment. If interventions to be comparedin an RCT not only differed in the administered substance
or drug but, in addition, in the route (e.g., intravenous ororal administration) or timing of application, then pa-tients’ preferences could have compromised their willing-ness to be randomized (research question #11 in Table 2).
DiscussionMain findingsIn this qualitative comparison between RCTs that didnot achieve their originally planned sample size due torecruitment problems and RCTs that were completed asplanned, we identified several reasons for poor recruit-ment. We found that RCTs with poor recruitment oftenhad narrower eligibility criteria than RCTs without poorrecruitment, were investigator-sponsored rather thanindustry-sponsored, and were less attractive for patientsand recruiters due to higher burden, outdated controlinterventions, or already existing or accumulating evi-dence on benefits and harms of interventions from otherRCTs. An existing network of study centers experiencedin the conduct of RCTs is instrumental for successful re-cruitment, but whether one or a few study centers or aninternational multicenter design was more advantageousseemed to depend on the research context. With chal-lenging settings such as acute care, vulnerable patients,or complex logistics, one or a few carefully chosen cen-ters may be preferable due to closer monitoring of re-cruitment, potentially better prepared and motivatedstudy staff, established procedures and, perhaps mostimportant, efficient communication among the trialteam. Prominent research areas in our sample were can-cer research and research in acute care with, on average,more than one RCT with poor recruitment per researchquestion suggesting that there are clusters of researchareas typically prone to recruitment problems.
Strengths and limitationsThe strengths of our study include a new systematic ap-proach to qualitatively compare RCTs discontinued orrevised due to poor recruitment and RCTs completed asplanned, that had previously been suggested [5] but, toour knowledge, never been used to date. We included 77RCTs from a broad range of settings and research topicsand analyzed them specifically in their context, therebystrengthening the applicability of our results.Our study has several limitations. First, the present
qualitative analysis was limited to the information pro-vided in publications of RCTs and did not include infor-mation from other sources such as study protocols orinterviews with trialists. This might have constituted a se-lection because the majority of RCTs discontinued forpoor recruitment were not published in a peer-reviewedjournal [1]. Second, we were not able to assess reasons forpoor recruitment that were not described (e.g., lack offunding, a theme recurrently coming up in an interview
Briel et al. Trials (2019) 20:800 Page 9 of 12
study on the topic [4]). Third, we did not comprehensivelysearch the literature for reports of RCTs discontinued orrevised due to poor recruitment, but pragmatically startedout with a sample of discontinued or revised RCTs identi-fied in a previous study [1], and we used existing system-atic and narrative reviews to find matching RCTs. Inaddition, we excluded 31 RCTs from our analysis becausearticles did not report a planned sample size, and there-fore we were unable to judge whether the originallyplanned sample size was achieved or not. Fourth, thereporting of the patient recruitment process in includedRCT publications provided little detail (irrespective ofwhether the RCT struggled with recruitment or not) com-promising the effectiveness of our qualitative analysis. Par-ticularly the fact that many articles did not report thenumber of patients screened for eligibility, the numbernot meeting eligibility criteria, and the number of patientsdeclining to participate, often limited a better understand-ing of potential recruitment problems. Fifth, one re-searcher extracted all relevant information from includedRCT publications and another checked this informationrather than two researchers extracting relevant data inde-pendently and in duplicate. We chose this approach forfeasibility reasons risking a higher rate of extraction errors.However, information directly relevant for our interpret-ation of reasons underlying recruitment problems were ac-tually verified by two methodologically trained researchers.Sixth, although our study captured a broad range of clinicalareas, 12 of the 15 research questions were related to drugtherapy, leaving uncertainty whether our findings areequally applicable to other interventions such as surgery,behavior change, or complex interventions. Finally, al-though we found evidence for saturation in our qualitativeanalysis, the size of our study sample was mainly deter-mined by practical issues of our approach.
Comparison with other studies investigating poorrecruitment in RCTsOur results confirm several findings of previous studiesusing different methods. Investigator-sponsored (in thesense of investigator-initiated) RCTs, for instance, werefound to be at higher risk for discontinuation due topoor recruitment than industry-sponsored RCTs by twostudies using a quantitative approach with multivariableregression analysis [1, 2]. The common interpretation isthat industry sponsorship is associated with sufficientfunding and better planning and conduct, factors that fa-cilitate successful patient recruitment. Moreover, theacute care setting (e.g., emergency rooms, intensive careunits, care for preterm neonates) seems particularlyprone to insufficient recruitment [10].Similar to the present study, a systematic review of
published reports of RCTs discontinued due to poor re-cruitment found that overly narrow eligibility criteria
and prejudiced views of recruiters and patients on trialinterventions were the most frequent reasons for poorrecruitment [3]. Prejudiced views of patients and re-cruiters may come from different sources. Our studyfound that RCTs with poor recruitment were oftenlaunched relatively late in the sequence of RCTs on thesame research question. As evidence accumulates overtime, the uncertainty about the benefits and harms of acertain intervention or about the superiority of oneintervention over another (equipoise) may become in-creasingly compromised. Some control interventionswere even considered outdated right from the start of anRCT (research questions #10 and #11 in Table 2), whichconfirms the observation by Habre et al. [11]. In someinstances, the route of application of an intervention orthe more complex logistics associated with an interven-tion were less appealing to patients or recruiters. This isalso consistent with the finding by Bernardez-Pereiraet al. that single-arm clinical trials were less prone todiscontinuation due to poor recruitment compared tomultiple-arm trials. That is because in single-arm trialsall patients receive the same intervention and, thus, arenot confronted with the fact that they could be random-ized to a different, maybe less preferred, treatment [2].Another common finding in RCTs with poor recruit-
ment was a high burden or inconvenience for patients orrecruiters due to trial procedures (e.g., many follow-upvisits, blood draws, lengthy questionnaires or case reportforms). This was mentioned previously as a problem inseveral other qualitative studies [3, 12–14].Our study highlights two aspects about recruitment
challenges to RCTs that are, so far, not prominent in thepublished literature. First, investigators need to be awareof all other RCTs on a research question, their timing,and the accumulating evidence base, so that potentiallycompromising effects on the recruitment to their owntrial can be minimized. Second, apart from the notionthat well-established networks of collaborating studycenters are an asset for the successful recruitment of pa-tients to RCTs [4], it seems that a larger number of cen-ters is not always advantageous. The performance of astudy center typically depends on its commitment to-ward an RCT, the enthusiasm, training, and quality ofcommunication of staff; therefore, challenging settingssuch as acute care, vulnerable patients, or complex logis-tics require careful selection and close monitoring ofparticipating centers.Finally, our study documents the urgent need for a more
detailed reporting of participant recruitment in RCTs, whichis in line with previous reports [3, 15, 16]. Furthermore, wedid not find that poor reporting of the recruitment processwas a particular issue of RCTs discontinued or revised forpoor recruitment. Indicating that, if they do indeed get pub-lished, the reporting quality of the recruitment process is
Briel et al. Trials (2019) 20:800 Page 10 of 12
similarly poor as in RCTs that were completed as planned.The current Consolidated Standards of Reporting Trials(CONSORT) statement [17] explicitly recommends report-ing the number of patients assessed for eligibility, number ofeligible patients, and number of consenting patients. In thecontext of RCT discontinuation due to poor recruitment,however, investigators should additionally describe how theyprojected the number of eligible and consenting patients;whether a pilot study including informed consent was done;whether recruitment was closely monitored; which measureswere undertaken to improve recruitment; and the specificroot causes for recruitment failure in their case, so that fu-ture recruitment failures in that area of research can effi-ciently be prevented.Given its magnitude and global presence of the problem,
the evidence base on what actually works to improve re-cruitment in RCTs is still astonishingly thin [18, 19]. Basedon the numerous analyses about the nature, extent, andcauses for recruitment failure, it is time for internationalcollaborative efforts to overcome the problem. Specifically,we need more randomized ‘studies within a trial’ (SWATs),i.e., promising interventions to improve recruitment needto be empirically evaluated within a host trial as propagatedby the Trial Forge initiative, the Medical Research Council’sNetwork of Hubs for Trials Methodology Research in theUK, and the Health Research Board’s Trials MethodologyResearch Network in Ireland [20, 21].
ConclusionsThis qualitative comparison of RCTs discontinued or re-vised due to poor recruitment and RCTs completed asplanned on the same research question complementsprevious efforts to identify risk factors for recruitmentfailure in RCTs, and to better understand the underlyingmechanisms. Our study confirms previously identifiedcauses such as narrow eligibility criteria, investigator-sponsorship, and a high burden of trial procedures forpatients and recruiters, but also stresses the importanceof considering the accumulating evidence and the timingof other RCTs on the same topic as well as carefullyselecting and closely monitoring participating centers forRCTs in challenging settings. A more detailed reportingof patient recruitment in RCTs is urgently needed sothat recruitment failure can provide lessons for other re-searchers in the future.
Supplementary informationSupplementary information accompanies this paper at https://doi.org/10.1186/s13063-019-3957-4.
Additional file 1. Standards for Reporting Qualitative Research (SRQR)*
Additional file 2. Pre-specified checklist of items potentially associatedwith poor recruitment
Additional file 3. References of included randomized controlled trials(RCTs)
Additional file 4. Recruitment characteristics of included randomizedcontrolled trials by research question
Additional file 5. Recruitment characteristics of included randomizedcontrolled trials (RCTs)
Additional file 6. General characteristics of included randomizedcontrolled trials (RCTs)
Additional file 7. Reporting quality of the recruitment process inincluded randomized controlled trials
AbbreviationsCONSORT: Consolidated Standards of Reporting Trials; RCT: Randomizedcontrolled trial; SRQR: Standards for Reporting Qualitative Research;SWAT: Study within a trial
AcknowledgmentsWe thank PD Dr. Dr. Benjamin Kasenda (oncologist at the University HospitalBasel) for providing expert knowledge about chemotherapy regimens inincluded RCTs with cancer patients.
Authors’ contributionsMB and EvE conceived of the study, wrote the protocol, and acquiredfunding. VG and BS conducted the literature searches and extracted datafrom relevant articles. MB, BS, and VG qualitatively analyzed the data. MB andVG wrote the first draft of the manuscript. BS and EvE critically revised themanuscript. All authors approved the final version before submission.
FundingThis study was funded by the Swiss National Science Foundation (grant no.320030_149496/1). The funder had no role in the study design, datacollection and analysis, decision to publish, or preparation of this manuscript.
Availability of data and materialsThe data used for and/or analyzed during the current study is available inAdditional files 1, 2, 3, 4, 5, 6, and 7.
Ethics approval and consent to participateNot applicable.
Consent for publicationNot applicable.
Competing interestsThe authors declare that they have no competing interests.
Author details1Basel Institute for Clinical Epidemiology and Biostatistics, Department ofClinical Research, University of Basel and University Hospital Basel, Basel,Switzerland. 2Department of Health Research Methods, Evidence, and Impact,McMaster University, Hamilton, ON, Canada. 3Cochrane Switzerland, Centerfor Primary Care and Public Health (Unisanté), University of Lausanne,Lausanne, Switzerland.
Received: 22 December 2018 Accepted: 4 December 2019
References1. Kasenda B, von Elm E, You J, Blumle A, Tomonaga Y, Saccilotto R, Amstutz
A, Bengough T, Meerpohl JJ, Stegert M, et al. Prevalence, characteristics, andpublication of discontinued randomized trials. JAMA. 2014;311(10):1045–51.
2. Bernardez-Pereira S, Lopes RD, Carrion MJ, Santucci EV, Soares RM, deOliveira AM, Laranjeira LN, Ikeoka DT, Zazula AD, Moreira FR, et al.Prevalence, characteristics, and predictors of early termination ofcardiovascular clinical trials due to low recruitment: insights from theClinicalTrials.gov registry. Am Heart J. 2014;168(2):213–9 e211.
3. Briel M, Olu KK, von Elm E, Kasenda B, Alturki R, Agarwal A, Bhatnagar N,Schandelmaier S. A systematic review of discontinued trials suggested that
Briel et al. Trials (2019) 20:800 Page 11 of 12
most reasons for recruitment failure were preventable. J Clin Epidemiol.2016;80:8–15.
4. Briel M, Elger B, von Elm E, Satalkar P. Insufficient recruitment andpremature discontinuation of clinical trials in Switzerland: qualitative studywith trialists and other stakeholders. Swiss Med Wkly. 2017;147:w14556.
5. Prescott RJ, Counsell CE, Gillespie WJ, Grant AM, Russell IT, Kiauka S, ColthartIR, Ross S, Shepherd SM, Russell D. Factors that limit the quality, numberand progress of randomised controlled trials. Health Technol Assess. 1999;3(20):1–143.
6. Anvari M, Allen C, Marshall J, Armstrong D, Goeree R, Ungar W, GoldsmithC. A randomized controlled trial of laparoscopic Nissen fundoplicationversus proton pump inhibitors for the treatment of patients with chronicgastroesophageal reflux disease (GERD): 3-year outcomes. Surg Endosc.2011;25(8):2547–54.
7. Grant AM, Wileman SM, Ramsay CR, Mowat NA, Krukowski ZH, Heading RC,Thursz MR, Campbell MK, Group RT. Minimal access surgery compared withmedical management for chronic gastro-oesophageal reflux disease: UKcollaborative randomised trial. BMJ. 2008;337:a2664.
8. Connolly SJ, Dorian P, Roberts RS, Gent M, Bailin S, Fain ES, Thorpe K,Champagne J, Talajic M, Coutu B, et al. Comparison of beta-blockers,amiodarone plus beta-blockers, or sotalol for prevention of shocks fromimplantable cardioverter defibrillators: the OPTIC Study: a randomized trial.JAMA. 2006;295(2):165–71.
9. Sandercock P, Wardlaw JM, Lindley RI, Dennis M, Cohen G, Murray G, InnesK, Venables G, Czlonkowska A, Kobayashi A, et al. The benefits and harms ofintravenous thrombolysis with recombinant tissue plasminogen activatorwithin 6 h of acute ischaemic stroke (the third international stroke trial [IST-3]): a randomised controlled trial. Lancet. 2012;379(9834):2352–63.
10. Schandelmaier S, von Elm E, You JJ, Blumle A, Tomonaga Y, Lamontagne F,Saccilotto R, Amstutz A, Bengough T, Meerpohl JJ, et al. Prematurediscontinuation of randomized trials in critical and emergency care: aretrospective cohort study. Crit Care Med. 2016;44(1):130–7.
11. Habre C, Tramer MR, Popping DM, Elia N. Ability of a meta-analysis toprevent redundant research: systematic review of studies on pain frompropofol injection. BMJ. 2014;348:g5219.
12. Spaar A, Frey M, Turk A, Karrer W, Puhan MA. Recruitment barriers in arandomized controlled trial from the physicians' perspective: a postalsurvey. BMC Med Res Methodol. 2009;9:14.
13. Mills E, Wilson K, Rachlis B, Griffith L, Wu P, Guyatt G, Cooper C. Barriers toparticipation in HIV drug trials: a systematic review. Lancet Infect Dis. 2006;6(1):32–8.
14. Kaur G, Smyth RL, Williamson P. Developing a survey of barriers andfacilitators to recruitment in randomized controlled trials. Trials. 2012;13:218.
15. Garde A, O'Hearn K, Nicholls S, Menon K. Reporting of consent rates incritical care studies: room for improvement. J Clin Epidemiol. 2016;74:51–6.
16. Toerien M, Brookes ST, Metcalfe C, de Salis I, Tomlin Z, Peters TJ, Sterne J,Donovan JL. A review of reporting of participant recruitment and retentionin RCTs in six major journals. Trials. 2009;10:52.
17. Schulz KF, Altman DG, Moher D, Group C. CONSORT 2010 statement:updated guidelines for reporting parallel group randomised trials. BMJ.2010;340:c332.
18. Treweek S, Pitkethly M, Cook J, Fraser C, Mitchell E, Sullivan F, Jackson C,Taskila TK, Gardner H. Strategies to improve recruitment to randomisedtrials. Cochrane Database Syst Rev. 2018;2:MR000013.
19. Gardner HR, Fraser C, MacLennan G, Treweek S. A protocol for a systematicreview of non-randomised evaluations of strategies to improve participantrecruitment to randomised controlled trials. Syst Rev. 2016;5(1):131.
20. Treweek S, Altman DG, Bower P, Campbell M, Chalmers I, Cotton S, Craig P,Crosby D, Davidson P, Devane D, et al. Making randomised trials moreefficient: report of the first meeting to discuss the Trial Forge platform.Trials. 2015;16:261.
21. Treweek S, Bevan S, Bower P, Campbell M, Christie J, Clarke M, Collett C,Cotton S, Devane D, El Feky A, et al. Trial Forge Guidance 1: what is a StudyWithin A Trial (SWAT)? Trials. 2018;19(1):139.
Publisher’s NoteSpringer Nature remains neutral with regard to jurisdictional claims inpublished maps and institutional affiliations.
Briel et al. Trials (2019) 20:800 Page 12 of 12