Prevalence, characteristics, and predictors of early termination of cardiovascular clinical trials due to low recruitment: Insights from the ClinicalTrials.gov registry

    Prevalence, characteristics, and predictors of early termination of cardiovascular clinical trials due to low recruitment: Insights from the ClinicalTrials.gov registry Sabrina Bernardez-Pereira MD, MSc, Renato D. Lopes MD, PhD, Maria Julia Machline Carrion MD, MSc, Eliana Vieira Santucci MS, Rafael Marques Soares CN,MSc, Matheus de Oliveira Abreu MS, Ligia Nasi Laranjeira MS, Dimas T. Ikeoka MD, PhD, Ana Denise Zazula MD, Frederico Rafael Moreira BSc, Alexandre Biasi Cavalcanti MD, PhD, Evandro Tinoco Mesquita MD, PhD, Eric D. Peterson MD, MPH, Robert M. Califf MD, Otavio Berwanger MD, PhD PII: DOI: Reference:

S0002-8703(14)00222-1 doi: 10.1016/j.ahj.2014.04.013 YMHJ 4614

To appear in:

American Heart Journal

Received date: Accepted date:

30 December 2013 28 April 2014

Please cite this article as: Bernardez-Pereira Sabrina, Lopes Renato D., Carrion Maria Julia Machline, Santucci Eliana Vieira, Soares Rafael Marques, de Oliveira Abreu Matheus, Laranjeira Ligia Nasi, Ikeoka Dimas T., Zazula Ana Denise, Moreira Frederico Rafael, Cavalcanti Alexandre Biasi, Mesquita Evandro Tinoco, Peterson Eric D., Califf Robert M., Berwanger Otavio, Prevalence, characteristics, and predictors of early termination of cardiovascular clinical trials due to low recruitment: Insights from the ClinicalTrials.gov registry, American Heart Journal (2014), doi: 10.1016/j.ahj.2014.04.013

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Prevalence, characteristics, and predictors of early termination of cardiovascular clinical

T

trials due to low recruitment: Insights from the ClinicalTrials.gov registry

RI P

Bernardez-Pereira S et al • Early Termination of Cardiovascular Trials

SC

Sabrina Bernardez-Pereira, MD, MSc,*,† Renato D. Lopes, MD, PhD,‡,§ Maria Julia Machline Carrion, MD, MSc,* Eliana Vieira Santucci, MS,* Rafael Marques Soares, CN, MSc,* Matheus de Oliveira Abreu,

NU

MS,* Ligia Nasi Laranjeira, MS,* Dimas T. Ikeoka, MD, PhD,* Ana Denise Zazula, MD,* Frederico

MA

Rafael Moreira, BSc,* Alexandre Biasi Cavalcanti, MD, PhD,* Evandro Tinoco Mesquita, MD, PhD,† Eric D. Peterson, MD, MPH,‡ Robert M. Califf, MD,II Otavio Berwanger, MD, PhD,* on behalf of the

ED

Methodological Evaluation of clinical TriAls (META) Study Group

Research Institute, HCOR–Hospital do Coração, Sao Paulo, Brazil; †Fluminense Federal University,

PT

*

Niteroi, Rio de Janeiro, Brazil; ‡Duke Clinical Research Institute, Duke University Medical Center,

CE

Durham, NC, USA; §Brazilian Clinical Research Institute, Sao Paulo, Brazil; IIDuke Translational

AC

Medicine Institute, Duke University Medical Center, Durham, NC, USA.

Address for correspondence: Dr. Otavio Berwanger, Research Institute, HCOR–Hospital do Coração, Rua Abilio Soares, 250 12 andar 04005-909, Paraiso Sao Paulo, SP, Brazil. Phone: 55 11 30536611 Ext. 8201; Fax: 55 11 38864695. E-mail: [email protected]

Word count: 3993

ACCEPTED MANUSCRIPT

AHJ-D-13-01462 R2 2

ABSTRACT Background: Early termination of clinical trials due to low recruitment represents an understudied

T

challenge for clinical research. We aimed to describe characteristics of cardiovascular trials terminated

RI P

because of low recruitment and identify the major predictors of such early termination. Methods: We reviewed all cardiovascular clinical trials (7,042 studies) registered in ClinicalTrials.gov

SC

from February 29, 2000, to January 17, 2013, and assessed information about trials that were completed and those that were terminated early. Logistic regression models were developed to identify independent

NU

predictors of early termination due to low recruitment.

MA

Results: Our search strategy identified 6,279 cardiovascular clinical trials, of which 684 (10.9%) were terminated prematurely. Of these halted trials, the main reason for termination was lower than expected recruitment (278 trials; 53.6%). When comparing trials that terminated early because of low recruitment

ED

with those that were completed, we found that studies funded by the National Institutes of Health or other

PT

US federal agencies (odds ratio [OR] 0.35, 95% confidence interval [CI] 0.14–0.89), studies of behavior/diet intervention (OR 0.35, 95% CI 0.19–0.65), and single-arm design studies (OR 0.57, 95% CI

CE

0.42–0.78) were associated with a lower risk of early termination. University/hospital-funded (OR 1.52, 95% CI 1.10–2.10) and mixed-source-funded studies (OR 2.14, 95% CI 1.52–3.01) were associated with

AC

a higher likelihood of early termination due to lower than expected recruitment rates. Conclusions: Low recruitment represents the main cause of early termination of cardiovascular clinical trials. Funding source, type of intervention, and study design are factors associated with early termination due to low recruitment and might be good targets for improving enrollment into cardiovascular clinical trials. Keywords: epidemiology; trials; cardiovascular diseases.

ACCEPTED MANUSCRIPT

AHJ-D-13-01462 R2 3

Cardiovascular disease is the primary cause of death and disability globally.1 High-quality evidence has established that interventions which reduce major cardiovascular events typically have only

T

moderate effects on outcomes; therefore, large numbers of participants are needed in clinical trials to

RI P

adequately control for random error. As clinical trials in cardiology become more complex and expansive, with increasingly large cohorts to allow for accurate assessment of the treatment effect of a given therapy,

SC

a common challenge for such trials is lower than expected recruitment rates, which can culminate in early termination of trials because of failure to reach the initially planned sample size within the expected

NU

timeframe.2 Currently, it is unknown what proportion of cardiovascular trials are halted early due to low

MA

recruitment. Moreover, the reasons why certain trials recruit well while others do not are not fully understood.

The publicly available ClinicalTrials.gov registry was launched more than a decade ago and has

ED

emerged as a key element of many public health policy initiatives aimed at improving clinical research.

PT

ClinicalTrials.gov represents the most comprehensive source for information about ongoing and completed trials within and outside the United States. Currently, about 10% of trials registered in

CE

ClinicalTrials.gov are cardiovascular studies.3 A growing number of researchers and authors of systematic reviews use ClinicalTrials.gov data to generate empirical evidence regarding clinical trials methodology,

AC

results, and publication rates.4 Analysis of the ClinicalTrials.gov database offers a unique opportunity to study a cohort of contemporary cardiovascular trials from registration to completion. Our study aims were to use ClinicalTrials.gov to (1) determine how frequently cardiovascular clinical trials terminate early, both overall and specifically because of lower than expected recruitment; (2) describe the characteristics of trials terminated due to low recruitment; and (3) identify predictors for early termination because of low recruitment.

METHODS ClinicalTrials.gov

ACCEPTED MANUSCRIPT

AHJ-D-13-01462 R2 4

The methods and quality assurance strategies used by ClinicalTrials.gov to register trials have been described in detail previously.5,6 ClinicalTrials.gov was developed and is maintained by the National

T

Library of Medicine on behalf of the National Institutes of Health (NIH). ClinicalTrials.gov uses a Web-

RI P

based system to facilitate registration of clinical trials by the study sponsor or the principal investigator,

SC

and includes both mandatory and optional data elements (see online Appendix A).

Eligibility criteria and search strategy

NU

We searched the ClinicalTrials.gov database to identify cardiovascular clinical studies registered from February 29, 2000, until January 17, 2013. Clinical or intervention studies were defined as clinical

MA

trials in which patients received an intervention in order to determine the generated results.7 Cardiovascular trials were defined as studies designed to test interventions aimed at the treatment or

ED

prevention of cardiovascular diseases. We considered cardiovascular diseases, according to MeSH terms,

PT

as being pathological conditions involving the cardiovascular system, including the heart, blood vessels, and pericardium, and including cardiovascular abnormalities (heart defects, vascular malformations),

CE

cardiovascular infection, heart diseases (cardiomyopathies, heart valve diseases, myocardial ischemia, pericardial diseases, arrhythmias, pulmonary heart disease, and others), and vascular diseases (aneurysm,

AC

aortic diseases, arteritis, embolism and thrombosis, hypertension, cerebrovascular disorders, and others).8 Eligibility decisions were conducted by teams of independent adjudicators (SBP, MJMC, EVS, RMS, MOA, LNL, DTI). Disagreements were resolved by consensus or third-party adjudication. Included studies were further classified as completed or terminated according to ClinicalTrials.gov criteria (online Appendix A). By this definition, completed studies were defined as a study that ended normally and participants are no longer being examined or treated. Terminated studies were those that ended before inclusion of the predetermined sample size and without intention to resume or restart.7 For terminated studies, four independent adjudicators (SBP, MJMC, EVS, LNL) attempted to define the reason for termination according to the following categories: low recruitment, safety or ethics issues, sponsor or business decisions, financial or administrative reasons, futility, benefit, study

ACCEPTED MANUSCRIPT

AHJ-D-13-01462 R2 5

competition or replaced for another study, treatment withdrawn from the market, or other reasons. Disagreements were resolved by consensus or third-party adjudication.

T

For studies that did not provide the recruitment status, a Medline database search was done

RI P

manually using the National Clinical Trial (NCT) number or through the item information (publications) to verify whether the study had been published. If we did not identify a publication, we searched Medline

SC

again with the protocol title, intervention, condition studied, and name of the principal investigator (when provided in response to the “study official” field). Articles identified through the search were matched to

NU

the corresponding trial (when possible) using the following information from ClinicalTrials.gov: detailed

MA

description, location, enrollment, study start and completion dates, and primary and secondary outcome measures. Studies were excluded from the analysis if it was not possible to define the reason for

ED

termination.

PT

Data extraction

We downloaded the dataset in eXtensible Markup Language (XML) containing all cardiovascular

CE

clinical studies registered with ClinicalTrials.gov by January 17, 2013. We then exported the dataset into R software (version 0.97.318)9 to extract the values. Data definitions and comprehensive data dictionaries

AC

were extracted from the database for Aggregate Analysis of ClinicalTrials.gov (AACT).10

Study variables The following clinical trial characteristics were assessed by independent investigators: allocation (randomized and non-randomized), endpoints (safety, efficacy, safety/efficacy, pharmacokinetics/ pharmacodynamics, and others, which included bioavailability and bioequivalence studies), intervention model (parallel, crossover, factorial, and single-arm assignment), masking (double-blind, single-blind, and open-label studies), primary purpose (diagnosis, treatment, prevention, and others, which included basic science, educational/counseling/training, health services research, supportive care, and screening), interventions (behavior or diet, drug, devices, biological, and others, which included genetic, procedural,

ACCEPTED MANUSCRIPT

AHJ-D-13-01462 R2 6

and radiation), sample size (1-100, 101-1,000, and >1,000), gender (female, male, or both), age group (adult, adult/senior, and others, which included child), sponsors and collaborators, and phases of clinical

T

trials (0 or 1, 1/2, 2, 2/3, 3, 4). Variables were defined according to ClinicalTrials.gov criteria available in

RI P

online Appendix A.

Organizations listed as sponsors for a study were considered the funder of the study and classified

SC

by funder type (e.g., NIH, other US federal agency, industry, university/hospital, and all others including individuals and community-based organizations). Probable trial funding source was determined using the

NU

following algorithm: If the lead sponsor was from the NIH or other US federal agency without

MA

collaboration of industry, university/hospital, or others, then the study was categorized as NIH funded. If the lead sponsor was from industry without collaboration of the NIH or other US federal agency, university/hospital, or others, then the study was categorized as industry funded. If the lead sponsor was a

ED

university/hospital without collaboration of the NIH or other US federal agency, industry, or others, then

PT

the study was categorized as university/hospital funded. If more than 1 funding source was considered as a collaborator, then the study was categorized as mixed-source funded. Otherwise, if the lead sponsor and

AC

Statistical methods

CE

collaborator fields were non-missing, the study was considered to be funded by “other.”3

To identify predictors of early termination due to low recruitment, we compared terminated trials and completed trials by means of unadjusted and adjusted analyses. Multiple logistic regression was performed including all covariates in a full model (from a set of variables with p<0.20 in univariate logistic regression analysis). Covariates with high percentages of missing values (>10%) were not included in the multiple logistic regression. We examined multicollinearity by assessing the variance inflation factor (VIF). A VIF > 2.5 for the logistic regression model was used as an indicator of multicollinearity.11 The discriminative power of the logistic regression model was checked with a receiver operating characteristic curve. Calibration was tested using the Hosmer-Lemeshow goodness-of-fit test.12

ACCEPTED MANUSCRIPT

AHJ-D-13-01462 R2 7

The logistic regression results are presented as odds ratios (OR) with their respective Wald 95% confidence intervals (CI) and p values. Two-tailed p values < 0.05 were considered to be statistically

T

significant. All statistical analyses were performed using SAS software (version 9.3; SAS Institute, Cary,

RI P

NC, USA).

SC

Sources of funding

NU

No extramural funding was used to support this work.

MA

RESULTS

Our search strategy retrieved 7,042 trials, from which we were able to identify 6,279 studies that met the inclusion criteria of cardiovascular clinical trials registered in ClinicalTrials.gov between

ED

February 29, 2000, and January 17, 2013. From the eligible trials, 5,595 (89.1%) were classified as

PT

completed and 684 (10.9%) trials as terminated. From the total of terminated trials, 165 trials (24.1%) did not report the reason for their early

CE

termination. After performing a Medline search, we were able to reclassify 35 of these trials when a related publication with results was identified. A total of 130 studies had no clear reason for termination

AC

and were excluded from the analysis (Figure 1). From the remaining 519 trials (75.9%), the main reason for termination was low recruitment (278 trials; 54%), followed by safety or ethics issues (9%), financial or administrative reasons (8%), sponsor or business decision (6%), treatment withdrawn from the market (6%), futility (6%), replaced or study competition (2%), benefit (2%), or other reasons (8%).

Characteristics of studies Overall characteristics of included studies are shown in Table I. As compared with completed trials, studies terminated early due to low recruitment were more often of parallel design (72.7% vs. 58.0%), assessed safety/efficacy endpoints (57.1% vs. 48.9%), were phase 4 studies (36.4% vs. 27.4%),

ACCEPTED MANUSCRIPT

AHJ-D-13-01462 R2 8

had diagnostic assessment as the primary trial objective (8.3% vs. 5.0%), and were funded by mixed

T

sources (23.0% vs. 14.5%).

RI P

Predictors of early termination

Our multivariate logistic regression adjustment with the variables assessed for inclusion at

SC

univariate regression (online Appendix B) found that mixed-source funding (OR 2.14, 95% CI 1.52–3.01; p<0.001) and university/hospital funding (OR 1.52, 95% CI 1.10–2.10; p=0.001) were independently

NU

associated with a higher risk of study termination due to low recruitment (Figure 2). Conversely, NIH/US

MA

federal funding (OR 0.35, 95% CI 0.14–0.89; p=0.027), behavior/diet intervention (OR 0.35, 95% CI 0.19–0.65; p<0.001), and single-arm design (OR 0.57, 95% CI 0.42–0.78; p<0.001) were factors independently associated with lower risk for early termination due to low recruitment. The area under the

ED

receiver operating characteristic curve of our final model was 0.65 (95% CI 0.62-0.68), and the Hosmer-

CE

DISCUSSION

PT

Lemeshow goodness-of-fit test showed a p value of 0.1485.

Our study showed that among a contemporary cohort of cardiovascular clinical trials registered in

AC

ClinicalTrials.gov, over 10% were terminated prematurely. The most common reason for early termination was lower than expected recruitment rates. Factors associated with lower risk of early termination due to low recruitment were funding by the NIH or other US federal agencies as well as trials assessing behavioral or dietary interventions and those with a single-arm design. On the other hand, studies funded by a university/hospital or from mixed sources were less often completed. Failure to recruit an adequate number of participants is a common cause of clinical trials being terminated prematurely. In a recent analysis of 114 multicenter trials from different specialties, the authors found that 45% of those studies failed to reach 80% of the prespecified sample size, and less than one third of the studies reached their sample size within the time allotted in the schedule.13 Our study confirms that these findings hold true among contemporary cardiovascular trials and adds important

ACCEPTED MANUSCRIPT

AHJ-D-13-01462 R2 9

information to these findings by showing that lower than expected recruitment rates represent the main reason for early termination of trials.

T

The main implications of low recruitment rates are potential lack of external validity, increased

RI P

probability of type II error, and ethical issues. Furthermore, when the time needed to reach the required sample size is extended in clinical trials, delays can arise in the introduction of new technologies and

SC

better therapies into clinical practice.14 Our results showed that studies funded by the NIH/US federal agencies were more likely to be completed. The NIH represents the leading supporter of biomedical

NU

research in the world, investing more than $31 billion annually in medical research for the American

MA

people.15 Inadequate funding or other economic considerations can be responsible for the decision to close a study early, as was demonstrated in the CONVINCE (Controlled Onset Verapamil Investigation of Cardiovascular End Points) trial.16 Attaining recruitment objectives in most multicenter clinical trials

ED

requires significant effort and commitment from the research team. Investigators may consider that

PT

NIH/US federal agency funded trials are more connected to public health priorities than market-driven initiatives.17 Also, the operational and regulatory requirements involved in industry-sponsored

CE

cardiovascular trials are enormous, which can affect enrollment. A recent study demonstrated that important barriers to participation in cardiovascular clinical trials are longer trial duration and intensive

AC

trial-related testing.18

Unlike trials funded by NIH/US federal agencies or industry, trials conducted by universities may not be able to count on sufficient funding to afford an adequate research infrastructure and a dedicated staff. Another explanation for the negative association between academic funding and trial completion may be related to lack of appropriate statistical planning, since sample size calculations of such studies are commonly based upon unrealistic event rates and overoptimistic effect sizes.19 Finally it is possible that NIH/US federal agency funded trials tend to employ a higher level of monitoring of trial conduct and timeline goals than other academic research projects. 20,21 We observed that trials assessing the effects of behavior or diet interventions were more likely to be completed. In such trials, a greater interaction between research subjects and healthcare professionals

ACCEPTED MANUSCRIPT

AHJ-D-13-01462 R2 10

may reduce patients’ perception of being “part of an experiment”; there is also no testing of new drugs or devices in these trials, and adverse events tends to be less common. Patient preferences may also explain

T

why a single-arm study design (i.e., all participants receive the same intervention) has a lower risk of

RI P

being terminated due to insufficient recruitment. 22-24

Our study has several limitations that merit consideration. Our results are prone to selection bias.

SC

Although ClinicalTrials.gov is considered the most comprehensive registry of ongoing and completed trials within and outside of the United States, it does not contain all clinical trials, and other information

NU

sources were not included in our analysis. Moreover, registration is not legally required for trials not

MA

involving a drug or a device, as well as trials not under US jurisdiction.3 It must also be acknowledged that we did not perform additional manual screening to identify and exclude possibly misclassified observational studies. Our results are also prone to ascertainment bias. There have been changes over time

ED

in the data collected, the definitions used, and the control over missing data. Moreover, because

PT

ClinicalTrials.gov allows changes to reported data at any time, it is currently not clear what data can be considered final. In addition, we derived funding source for the trials from the lead sponsor and

CE

collaborator fields, but this classification is prone to errors. Furthermore, reason for termination is not a required field in ClinicalTrials.gov, as there is no defined classification. Finally, residual confounding

AC

may have also biased our results. Information regarding anticipated recruitment, a potential confounder, was missing for a large proportion of included trials, and other variables such as case report form length, amount of funding, and factors related to trial complexity are not readily available from the current ClinicalTrials.gov database. In summary, early termination of cardiovascular clinical trials is common, and low recruitment represents the most frequent reason for early termination. Trials funded by the NIH and other US federal agencies and trials of behavior and diet interventions face a lower risk of early termination due to low recruitment. Our findings may contribute to a better understanding of the etiology of premature trial termination due to low enrollment and could also set the stage for targeted approaches to improve clinical research recruitment.

ACCEPTED MANUSCRIPT

AHJ-D-13-01462 R2 11

Authorship

T

The authors are solely responsible for the design and conduct of this study, all study analyses, the

RI P

drafting and editing of the paper, and its final contents.

SC

Acknowledgments

We thank Peter Hoffmann of the Duke Clinical Research Institute for his editorial contribution to

NU

this manuscript. Appendix A: ClinicalTrials.gov criteria

MA

Current Variables

CE

PT

ED

Comments on 'CTTI_Notes': CTTI notes are provided for those variables that are either derived from ClinicalTrials.gov data elements (e.g., 'Design_Name' and 'Design_Value' are derived from 'Study_Design') or are not available in ClinicalTrials.gov Protocol Data Element Definitions (e.g., 'Agency_Class', 'First_Received_Results_Date', and 'MeSH_Term'), or are defined primarily for CTTI's database purposes (e.g., system generated sequential IDs). Comments on dates convention: The dates that are generated by NLM (LASTCHANGED_DATE, FIRSTRECEIVED_DATE, and FIRSTRECEIVED_RESULTS_DATE) are stored as “month dd, yyyy” and those submitted by the users (START_DATE, COMPLETION_DATE, PRIAMRY_COMPLETION_DATE, and VERIFICATION_DATE) seem to be stored as “month yyyy”.

Variable

AC

Appendix B: Univariable regression analysis assessing predictors of early termination of cardiovascular interventional trials due to low recruitment. Odds Ratio (95% CI*)

P Value

Study Designs Randomized

Non-randomized

1.0 (Ref.) 0.67 (0.46-0.97)

0.032

Efficacy

0.81 (0.62-1.07)

0.140

Pharmaco † Safety-efficacy

0.45 (0.21-0.97)

0.041

Safety

0.47 (0.25-0.87)

0.016

Intervention Model Crossover

0.44 (0.27-0.73)

0.001

Factorial

0.43 (0.16-1.18)

0.101

Parallel

1.0(Ref.)

Endpoints

1.0(Ref.)

ACCEPTED MANUSCRIPT

AHJ-D-13-01462 R2 12

Single group

0.56 (0.41-0.76)

<.001

Masking Double blind

0.948

Single blind

1 (0.67-1.48)

0.986

Diagnost

1.65 (1.04-2.6)

0.032

Prevention

0.96 (0.67-1.37)

0.800

Treatment

1.0(Ref.) 0.62 (0.33-1.14)

Intervention 0.26 (0.04-1.87)

Device

1.3 (0.95-1.78)

0.179

MA

Biologic

0.122

NU

Other

SC

Primary Purpose

0.099

1.0(Ref.)

Other

0.77 (0.54-1.11)

Behavior/diet

0.33 (0.18-0.59)

0.165 <.001

1.39 (0.82-2.37)

0.225

0.57 (0.13-2.48)

0.457

1.39 (0.82-2.37)

0.225

ED

Drug

Sample Size

PT

1-100 101-1000

CE

>1000 Gender Only female Both

0.53 (0.22-1.31)

AC

Only male

0.33 (0.12-0.89)

0.171 0.028

1.0(Ref.)

Age Adult Adult/senior Other (including child)

0.59 (0.37-0.96)

0.032

1.0 (Ref.) 0.73 (0.49-1.09)

0.122

Funded By Industry

1.0 (Ref.)

Other

1.18(0.77-1.79)

Mixed sources

1.84 (1.32-2.55)

0.449 <.001

NIH/US fed

0.25 (0.1-0.62)

0.003

University/hospital

1.28 (0.95-1.73)

0.109

Phases

RI P

0.99 (0.76-1.29)

T

1.0 (Ref.)

Open label

ACCEPTED MANUSCRIPT

AHJ-D-13-01462 R2 13

0 or I

0.39 (0.19-0.82)

I/II or II

1.18 (0.82-1.69)

II/III or III

0.013 0.375

1.0 (Ref.)

IV

0.024

T

1.47 (1.05-2.06)

*CI = Confidence Interval

AC

CE

PT

ED

MA

NU

SC

RI P

† Pharmacodynamics/pharmacokinetic

ACCEPTED MANUSCRIPT

AHJ-D-13-01462 R2 14

REFERENCES Centers for Disease Control and Prevention. Compressed mortality file: underlying cause of death,

T

1.

2.

RI P

1979 to 2005. http://wonder.cdc.gov/mortSQL.html. Accessed February 28, 2014. Avenell A, Grant AM, McGee M, et al. The effects of an open design on trial participant

SC

recruitment, compliance and retention—a randomized controlled trial comparison with a blinded, placebo-controlled design. Clin Trials 2004;1:490–8.

Califf RM, Zarin DA, Kramer JM, et al. Characteristics of clinical trials registered in

NU

3.

4.

MA

ClinicalTrials.gov, 2007-2010. JAMA 2012;307:1838–47. Zarin DA, Tse T, Williams RJ, et al. The ClinicalTrials.gov results database—update and key issues. N Engl J Med 2011;364:852–60.

Zarin DA, Tse T, Ide NC. Trial registration at ClinicalTrials.gov between May and October 2005.

ED

5.

6.

PT

N Engl J Med 2005;353:2779–87.

Zarin DA, Ide NC, Tse T, et al. Issues in the registration of clinical trials. JAMA 2007;297:2112–

7.

CE

20.

ClinicalTrials.gov protocol Data Element Definitions (DRAFT). January 2013.

8.

AC

http://prsinfo.clinicaltrials.gov/definitions.html. Accessed April 4, 2013. National Center for Biotechnology Information. MeSH terms: Cardiovascular Diseases. 2013. http://www.ncbi.nlm.nih.gov/mesh?Db=mesh&term=Cardiovascular+Diseases. Accessed April 4, 2013. 9.

R Core Team. R: A language and environment for statistical computing. 2013. http://www.Rproject.org/. Accessed April 4, 2013.

10.

AACT Database. Clinical Trials Transformation Initiative Web site. 2013. http://www.ctticlinicaltrials.org/what-we-do/analysis-dissemination/state-clinical-trials/aact-database. Accessed April 4, 2013.

ACCEPTED MANUSCRIPT

AHJ-D-13-01462 R2 15

11.

Allison PD. Logistic Regression Using the SAS System: Theory and Application. Cary, NC: SAS Institute; 2001. Hosmer DW, Lemeshow S. Applied Logistic Regression. 2nd ed. New York: John Wiley & Sons;

T

12.

13.

RI P

2000.

McDonald AM, Knight RC, Campbell MK, et al. What influences recruitment to randomised

14.

SC

controlled trials? A review of trials funded by two UK funding agencies. Trials 2006;7:9. Reynolds T. Clinical trials: can technology solve the problem of low recruitment? BMJ

National Institutes of Health. Our economy. 2013. http://www.nih.gov/about/impact/economy.htm.

MA

15.

NU

2011;342:d3662.

Accessed August 15, 2013.

Black HR, Elliott WJ, Grandits G, et al. Principal results of the Controlled Onset Verapamil

ED

16.

Investigation of Cardiovascular End Points (CONVINCE) trial. JAMA 2003;289:2073–82. Kitterman DR, Cheng SK, Dilts DM, et al. The prevalence and economic impact of low-enrolling

PT

17.

18.

CE

clinical studies at an academic medical center. Acad Med 2011;86:1360–6. Martin SS, Ou FS, Newby LK, et al. Patient and trial specific barriers to participation in

19.

AC

cardiovascular randomized clinical trials. J Am Coll Cardiol 2013;61(7):762–9. Bowman SJ, Macfarlane GJ; Arthritis Research Campaign Data Monitoring Committee. Successful patient recruitment in investigator-led clinical trials. Rheumatology (Oxford) 2007;46(7):1207–8. 20.

DeMets DL, Califf RM. Lessons learned from recent cardiovascular clinical trials: Part II. Circulation 2002;106(7):880–6.

21.

Als-Nielsen B, Chen W, Gluud C, Kjaergard LL. Association of funding and conclusions in randomized drug trials: a reflection of treatment effect or adverse events? JAMA 2003;290(7):921– 8.

22.

Jack WJ, Chetty U, Rodger A. Recruitment to a prospective breast conservation trial: why are so few patients randomised? BMJ 1990;301:83–5.

ACCEPTED MANUSCRIPT

AHJ-D-13-01462 R2 16

23.

Llewellyn-Thomas HA, McGreal MJ, Thiel EC, et al. Patients’ willingness to enter clinical trials: measuring the association with perceived benefit and preference for decision participation. Soc Sci

RI P

Yeomans-Kinney A, Vernon SW, Frankowski RF, et al. Factors related to enrollment in the breast cancer prevention trial at a comprehensive cancer center during the first year of recruitment. Cancer

CE

PT

ED

MA

NU

SC

1995;76:46–56.

AC

24.

T

Med 1991;32:35–42.

ACCEPTED MANUSCRIPT

AHJ-D-13-01462 R2 17

Figure Legends Figure 1. Study design – flowchart of systematic analysis of trials registered in ClinicalTrials.gov.

T

Figure 2. Multivariable logistic regression – analysis assessing predictors of early termination of

AC

CE

PT

ED

MA

NU

SC

RI P

cardiovascular clinical trials due to low recruitment.

ACCEPTED MANUSCRIPT

CE AC

Figure 1

PT

ED

MA

NU

SC

RI P

T

AHJ-D-13-01462 R2 18

ACCEPTED MANUSCRIPT

MA

NU

SC

RI P

T

AHJ-D-13-01462 R2 19

AC

CE

PT

ED

Figure 2

ACCEPTED MANUSCRIPT

AHJ-D-13-01462 R2 20

Table 1: Characteristics of completed and terminated cardiovascular studies registered in ClinicalTrials.gov

222 (87.1%) 33 (12.9%) 255

183 (82.8%) 38 (17.2%) 221

107(89.2%) 13 (10.8%) 120

87 (35.5%) 7 (2.9%) 140 (57.1%) 11 (4.5%) 245

61 (27.9%) 7 (3.2%) 140 (63.9%) 11 (5.0%) 219

25 (22.5%) 7 (6.3%) 70 (63.1%) 9 (8.1%) 111

580 (11.1%) 140 (2.7%) 3018 (58.0%) 1464 (28.1%) 5202

17 (6.2%) 4 (1.5%) 200 (72.7%) 54 (19.6%) 275

15 (6.3%) 4 (1.7%) 156 (65.8%) 62 (26.2%) 237

10 (8.1%) 0 (0%) 88 (71.0%) 26 (21.0%) 124

2003 (37.9%) 2617 (49.5%) 670 (12.7%) 5290

105 (38.0%) 136 (49.3%) 35 (12.7%) 276

111(46.3%) 101 (42.1%) 28 (11.7%) 240

62 (49.2%) 52 (41.3%) 12 (9.5%) 126

268 (5.0%) 778 (14.6%) 3934 (73.7%) 359 (6.7%) 5339

22 (8.3%) 37 (13.9%) 196 (73.7%) 11 (4.1%) 266

8 (3.4%) 31 (13.2%) 190 (80.9%) 6 (2.6%) 235

9 (7.3%) 16 (12.9%) 98 (79.0%) 1 (0.8%) 124

680 (12.1%) 72 (1.3%) 784 (13.9%)

12 (4.3%) 1 (0.4%) 55 (19.8%)

11 (4.6%) 6 (2.5%) 49 (20.3%)

1 (0.8%) 0 (0%) 21 (16.2%)

AC

CE

PT

ED

No reason for stopping early (N=130)

RI P

SC

MA

1663 (37.4%) 243 (5.5%) 2175 (48.9%) 367 (8.3%) 4448

Stopped early for other reasons (N=241)

T

3998 (81.7%) 893 (18.3%) 4891

NU

Variable

Allocation Randomized Non Randomized Total Endpoints Efficacy Pharmaco* Safety/Efficacy Safety Total Intervention Model Crossover Factorial Parallel Single group Total Masking Double Blind Open Label Single Blind Total Primary Purpose Diagnostic Prevention Treatment Other Total Intervention Behavior/Diet Biologic Device

(N=5630)

Stopped early due to low recruitment (N=278)

Completed studies

ACCEPTED MANUSCRIPT

AHJ-D-13-01462 R2 21

149 (61.8%) 26 (10.8%) 241

88 (67.7%) 20 (15.4%) 130

33 (53.2%) 27 (43.5%) 2 (3.2%) 62

139 (58.4%) 71 (29.8%) 28 (11.8%) 238

70 (57.4%) 45 (36.9%) 7 (5.7%) 122

3 (1.3%) 1 (0.4%) 236 (98.3%) 240

1 (0.8%) 1 (0.8%) 128 (98.5%) 130

19 (6.8%) 231 (83.1%) 28 (10.1%) 278

14 (5.8%) 201 (83.4%) 26 (10.8%) 241

9 (6.9%) 110 (84.6%) 11 (8.5%) 130

94 (33.8%) 30 (10.8%) 64 (23.0%) 5 (1.8%) 85 (30.6%) 278

121 (50.2%) 17 (7.1%) 51 (21.2%) 8 (3.3%) 44 (18.3%) 241

71 (54.6%) 9 (6.9%) 15 (11.5%) 4 (3.1%) 31 (23.9%) 130

8 (3.8%) 57 (27.3%) 68 (32.5%) 76 (36.4%) 209

16 (7.9%) 63 (31.2%) 84 (41.6%) 39 (19.3%) 202

11 (9.5%) 40 (34.5%) 32 (27.6%) 33 (28.4%) 116

ED

PT

CE

AC

SC

RI P

T

174 (62.6%) 36 (12.9%) 278

NU

5 (1.8%) 4 (1.4%) 269 (96.8%) 278

MA

Drug 3228 (57.3%) Other 866 (15.4%) Total 5630 Sample Size 1-100 398 (59.1%) 101-1000 234 (34.7%) > 1000 42 (6.2%) Total 674 Gender Only Female 181 (3.2%) Only Male 236 (4.2%) Both 5203 (92.6%) Total 5620 Age Adult 598 (10.6%) Adult/senior 4315 (76.6%) Other (with child) 717 (12.7%) Total 5630 Funded By Industry 2197 (39.0%) Other 596 (10.6%) Mixed source 814 (14.5%) NIH/US Fed 469 (8.3%) University/Hospital 1554 (27.6%) Total 5630 Phases 0 or I 453 (10.9%) I/II or II 1068 (25.7%) II/III or III 1500 (36.1%) IV 1138 (27.4%) Total 4159 * Pharmacodynamic/pharmacokinetic