Abstract
Background:
With large collaborations needed to reach sample size requirements for relatively rare events, a major challenge for multi-centre clinical trials is efficiency of recruitment at individual sites. We used data from an international, multi-centre, randomised trial of preterm prelabour rupture of membranes to assess any impact on recruitment following the introduction of a new Clinical Trial Agreement and to identify site-specific predictors of recruitment to the trial for the purpose of targeting future recruitment sites and strategies.
Methods:
The outcome measure was recruitment rate per 10,000 births, and according to this, an average recruitment rate was determined. Factors that were considered potentially predictive of recruitment above the average rate were classified according to three broad themes: ‘ethics and regulatory requirements’, ‘characteristics of site investigators’ and the ‘research culture’ at the collaborating site. Data were analysed using contingency tables and logistic regression modelling.
Results:
At 31 January 2009, following the introduction of the Clinical Trial Agreement, 39 centres had obtained ethics approval to commence recruitment, and 38 centres had enrolled at least one woman. Time to first recruit ranged from 25 days to 584 days. Recruitment rates ranged from 0.18 to 6.0 per 10,000 births (mean 1.71/10,000 births) per month. Factors most associated with above-average recruitment rate were the following: implementation of a clearly defined ‘system’ of recruitment, engagement of other staff, time from ethics approval to first recruit and provision of a dedicated trial coordinator.
Conclusion:
A delay of greater than 3 months in approval of the new Clinical Trial Agreement had an effect which extended into the third year of the trial. Characteristics that were indicative of the presence of a ‘system’ were the best predictors of recruitment. It may be more effective to limit recruitment sites and focus resources on those sites where investigators are engaged with trial processes and have adequate resources and structures to support them.
Introduction
Recruitment of participants is crucial to the success of clinical trials. For investigator-led, non-pharmaceutical trials, recruitment is dependent on the goodwill and commitment of individuals participating at collaborating sites. Where outcome events are rarer or large samples sizes are required, collaboration with other sites is vital. Slower than expected recruitment is a major barrier to successful and timely completion of clinical trials.1,2 This can potentially lead to underpowered investigations of clinically relevant conditions. Less than one-third of trials recruit to target. 3 This has implications for resources and morale levels at the coordinating centre as well as the collaborating sites. Barriers to, and enablers of, recruitment to clinical trials require investigation.
Trials designed to demonstrate improvements in perinatal mortality and morbidity are often large and require to be multi-centred. 4 This is especially the case in the Australia and New Zealand networks where populations are comparatively small. Pregnancy studies are often individual investigator led and attract no commercial sponsorship. Gaining the required approvals for multi-centre studies is time-consuming and potentially costly if there is little return in the form of patient enrolment. Delay in transmitting research evidence into practice, however, is the primary concern. A method for identifying centres that are likely to be recruit well would be useful.
Research data about trial recruitment are few. There are four reviews on the subject of predicting recruitment to clinical trials.1,2,5,6 These reviews, however, deal with strategies to increase recruitment, rather than prediction of what characteristics specific to a recruiting site would result in differential recruitment to a trial. Flyers, email shots and physician referrals are suggested to be effective forms of patient recruitment.6,7 However, before these strategies are executed, are there inherent factors at a collaborating site, or characteristics of an investigator, that will generate momentum for effective strategies to be initiated? We evaluated the indicators of recruitment in a multi-centre trial called preterm prelabour rupture of membranes close to term (PPROMT).
The ‘PPROMT’ Trial is a large multi-centre, international randomised controlled trial (RCT) investigating two standard forms of management for the condition of preterm prelabour rupture of the membranes (PPROM) in women who are between 34 and 37 weeks of gestation 8 operating principally from Australia. PPROM is a relatively uncommon complication of pregnancy, and women with PPROM may present at any time throughout the day or night. Effective enrolment of such women therefore requires appropriate awareness and engagement of general clinical staff. The aim of this study was to identify site-specific indicators of recruitment to the PPROMT Trial.
Methods
A priori hypotheses were developed about factors that may influence recruitment within sites, and these factors were divided into three broad categories of investigation: (1) Ethics and Regulatory Documents, (2) Investigator Characteristics and (3) General Research Culture.
Study sites
A total of 39 tertiary obstetric hospitals in Australia, New Zealand, the United Kingdom, South Africa, Norway, Argentina and Egypt participated in the PPROMT Trial at the time these data were collected at the end of January 2009. These study sites had all obtained ethical and regulatory approval to begin recruitment to the PPROMT Trial by 30 January 2009. Regulatory submissions were prepared by the coordinating centre, with any additional information provided by the site. At site-induction, all sites were provided with the same study materials, processes for recruitment and randomisation, and promotional items. Sites were given study training by telephone or in person, depending on location. All identified personnel involved in carrying out the trial were invited to attend. All sites had access to telephone numbers and email addresses for the principal investigator (PI) and study coordinator to be used by any personnel at the sites.
Data collection
The PPROMT Trial began recruitment in May 2004 and completed recruitment of 1812 women in June 2013. For the purpose of this survey, we examined recruitment at each site from January 2005, taking into account each site’s start date. This time frame was chosen as it followed an extensive change in the Clinical Trial Agreement (CTA) that was implemented between institutions, and the impact this had on trial recruitment was of interest. Data were collected from two sources. First, we emailed a survey to the named investigator or coordinator at each site. The survey collected information about personnel involved in the trial, including their role, qualifications and experience, size of institution in births per year over the 2004/2005 period, and level of obstetric and neonatal care. An average birth rate over these 2 years was then assigned to each collaborating site. Responses were obtained from all sites.
Second, we examined specific indicators which may influence recruitment at collaborating sites. The characteristics for each factor were determined over a period of time beginning with submission of the ethics application and continuing over the induction and recruitment to the trial at each site up until 31 January 2009, when the survey period ended. Recruitment at this point was 650 women from a target recruitment of 1812 women.
Non-survey information was ascertained from the PPROMT Trial database. These indicators were classed in three categories: Ethics and Regulatory Documents, Investigator Characteristics and General Research Culture. Ethics and Regulatory Documents included time taken for the PPROMT Trial to gain ethics approval, period of time the site had been eligible to recruit, time from approval till first recruit, delay of greater than 3 months in ethics approval due to negotiations regarding the CTA and average birth rate for the site over the 2004/2005 period. Investigator Characteristics included the presence or absence of the following items: local PI–specific inquiry into details of the protocol; local PI uptake of offer of in-service for themselves and other staff members; local PI’s description of the specific ‘system’ or process for the identification, counsel, recruitment and collection of trial data that he or she planned to implement for the purposes of the PPROMT Trial; local PI–provided responses to local ethics committee questions; personal relationship of local PI with the chief investigator as basis for collaboration in trial; average email response time of local PI regarding trial-related matters less than or equal to 5 days and involvement of additional staff not directly related to the PPROMT Trial at the local site. For General Research Culture, we determined the presence or absence of the following items: site’s engagement in other multi-centre perinatal trials, dedicated trial coordinator for PPROMT Trial, experience of dedicated trial coordinator, timeliness of provision of datasheets, Consolidated Standards of Reporting Trials (CONSORT) statements 9 and other PPROMT Trial–related materials.
Outcomes
Recruitment rate was calculated by expressing the number of recruits per month until 31 January 2009 per 10,000 births. The primary outcome was an indicator of whether or not the recruitment rate at the site was above or below the mean recruitment rate (Table 1).
Hospital characteristics.
NICU: neonatal intensive care unit.
Level II: specialty newborn care greater than 32 weeks’ gestation. Do not provide assisted ventilation. Level III: care for babies born greater than 26–28 weeks. Differing types of mechanical ventilation and can provide advanced surgeries.
Analysis
Statisticalanalysis was carried out using SAS software, version 9.1. Each potential predictor was analysed to determine its association with the outcome. Continuous variables were analysed using Student’s t-tests, and dichotomous variables were analysed using chi-squared tests.
Results
All 39 sites included had the necessary approvals to commence enrolment. The average (mean) recruitment rate was 1.7 per month per 10,000 births, with a median of 1.4. There was a total enrolment from all sites of 534 women. A total of 21 sites were categorised as below average for recruitment and 18 as above average.
In the Ethics and Regulatory Documents category (Table 2), outcomes differed only with regard to time to first recruit, above-average recruiting sites having an average of 143 days as compared with 253 days for below-average sites. In the category of Investigator Characteristics, investigators from above-average recruiting sites were significantly more likely to have an identifiable system of recruitment in place. This factor indicated whether the investigator was able to outline, prior to beginning the trial, a system by which counselling, recruitment and randomisation of patients would occur, as well as subsequent data collection, and those staff who would be responsible for these roles. This factor was considered to be part of the local investigator’s responsibility for the running of a trial at the site. Above-average recruiting sites also demonstrated an increased likelihood of having engaged other staff in the running of the trial and had faster response times to emails from the coordinating site. In the Research Culture category, while there was no association between recruitment rates and having engagement in other multi-centre trials, there was a significant association with the presence of a dedicated trial coordinator or research midwife. Additionally, above-average recruiters were more likely to have an experienced trial coordinator in the position and more likely to be timely with return of datasheets.
Comparison between below- and above-average recruitment sites for all factors.
PI: principal investigator.
With the exception of the top four factors, the columns display the percentages of sites that possess the characteristic.
Discussion
This study identified that factors most associated with above-average recruitment rate were the following: implementation of a clearly defined ‘system’ of recruitment, engagement of other staff, time from ethics approval to first recruit and provision of a dedicated trial coordinator. Our findings strongly suggest that for trials to recruit effectively, participating centres need to have a defined research trial structure which supports the site investigator. In the case of investigator-led, non-commercially sponsored trials, if the investigator is engaged with the research question and has engaged other staff to fulfil their commitment to recruiting to the trial, the site will recruit well and consistently.1,3,10
While we grouped ‘system of recruitment’ under the ‘Investigator Characteristics’ label, it would seem that ‘system’ is a stand-alone characteristic of the site. We looked at the distribution of recruitment for sites that did and did not have a ‘system’ of recruitment (see Figure 1). Clearly, the presence of a system is the most influential factor in this analysis, and other factors analysed are probably supportive or indicative of the presence of a ‘system’.
Average recruitment for sites with a system for recruitment in place versus sites without a system for recruitment in place.
With investigator-led, non-commercially sponsored trials, there is often little consideration paid to the systems requirements for sites to implement the protocols effectively. As a result, the clinicians who are collaborating in trials are often isolated and have little support or guidance, and trial inertia becomes evident. 7 This is reflected in our findings that while size of institution (as measured by birth rate), time taken to gain ethics approval and period of time a site had been eligible to recruit did not have an impact on a site’s rate of recruitment to the PPROMT Trial, a time of less than 3 months to the first recruit was indicative of a subsequent above-average recruitment rate. The enthusiasm and momentum that are generated by staff having a quick beginning to the trial may carry the trial forward and indicate that a system of recruitment to the trial is present. Our findings are consistent with Fletcher et al., 1 who indicate that appropriate support for clinicians in terms of prioritising time, communication and understanding of the research is key to boosting recruitment.
There are many reasons why clinical trials recruit less well than anticipated. 11 Other research cites factors including bureaucratic barriers, clinician lack of time, lack of understanding of research or research dedicated time and patient unwillingness.2,7 Of these, the time-poor clinician is most significant, which would indicate that an effective system of research collaboration is not present, as we’ve demonstrated. Efficient recruitment requires adequate support and infrastructure for trials, and with increasing competition for grant funding, having the appropriate site-specific resources to maximise effectiveness is of value.
Equipoise is also seen as a vital prerequisite for successful trial completion,12,13 so physician preference for any treatment will necessarily impair participation. For the PPROMT Trial, a pre-trial survey of obstetricians 14 demonstrated that not only did equipoise exist, but there was a willingness to enrol to a clinical trial at the gestations specified.
Our investigation of delays created by institutional approval of CTAs is important as it relates to the growing complication of regulatory documents, and the implications this has for the institutions that are responsible for writing regulatory documents. When we considered the data over the course of the trial, delays created by the new and more complex Institutional CTA were predictive of a site’s recruitment up until the third year of the trial. In the fourth year, the association was no longer present. This is an important consideration, as many grants extend only to 4 years, and if recruitment is hampered, as in this trial, by institutional delay with effects right into the third year, this has a significant impact on capacity to complete projected recruitment on time. The PPROMT Trial took more than twice the anticipated length of time to complete recruitment, and an additional application for grant monies was required.
There is concern that the departments drafting regulations which have the purpose of protecting intellectual property may have become disconnected from the clinical and practical workings of clinical trials, including time limitations. The essence of investigator-led enquiry into the practice and management of obstetric conditions relies on the recognition of the clinical equipoise that exists and the goodwill of the participating institutions to contribute to the resolution by recruitment to the trial. These factors may not be generalisable beyond perinatal research, and infrastructure requirements may differ for different areas of research, for example, cancer; however, the research is consistent across disciplines with regard to lower than expected recruitment rates. The data we collected were during the 2008–2009 period, and these issues persisted to the end of the trial and remain relevant in the current climate of clinical trial recruitment.
Pragmatic trials such as this one are important contributors to clinical trials research and as such require appropriate attention. Systems and support for the effective management of the trial in terms of personnel, structures and strategy would appear to be crucial for its success. In many Australian centres, this is not regarded as core business for health-care providers. Clinical trials are designed to optimise health-care outcomes and therefore are integral to safety and quality of health care, and require timely delivery of results. Studies from the United Kingdom show that the establishment of the UK Clinical Research Network has led to the development of disease-specific networks for recruitment to clinical trials and have resulted in improved rates of recruitment and increased infrastructure, including roles for clinical trials officers for this specific purpose. 15 Priority should be given to the establishment of protected research positions in our institutions, whose role it is to recruit into studies and encourage clinical staff to do likewise. The resultant economic benefit of completing trials within the timeframes set and the flow on effect of the introduction of evidence-based medicine into clinical practice would make sequestered positions for research staff more than worthwhile. 15 It is necessary to conduct quality trials to determine the answers to important clinical questions, and therefore, the identification and designation of appropriate infrastructure are also required.
Footnotes
Declaration of conflicting interests
The authors declare that there is no conflict of interest.
Funding
Kate M Levett was supported by a project grant from the National Health and Medical Research Council of Australia at the time of this research.