SEARCH

SEARCH BY CITATION

To obtain clear and definitive answers to clinically important questions, randomised controlled trials (RCTs) with sufficiently large sample sizes are needed, which usually requires recruitment in several centres. A common problem in multicentre trials is that recruitment is slower or more difficult than expected, and many trials fail to reach their planned sample size within the timescale and funding originally envisaged. If the target sample size is not achieved, the trial has less power to discover differences between the groups, so the results may be less useful. If recruitment has to be extended to reach the required sample size, the trial will take longer and cost more than planned, delaying the use of the results in clinical practice. The more expensive trials become and the longer they take, the fewer can be conducted with the funding and resources available. It is therefore important to understand the reasons for poor recruitment and find effective strategies to improve it.1

Factors affecting clinicians' participation

  1. Top of page
  2. Factors affecting clinicians' participation
  3. Factors affecting patients' participation
  4. Improving recruitment to multicentre trials
  5. References

Multicentre studies rely on clinicians to recruit patients in centres remote from the co-ordinating centre, which therefore has only limited control over recruitment. Trials may fail either because too few clinicians are willing to take part or because they do not recruit enough patients. There could be many reasons for these problems. Clinicians may not perceive a trial's question as important, as they may be unaware of the poor evidence base for established practices, and hence may not understand the rationale for the trial and feel no uncertainty about their own practice. Trials evaluating treatments that are already being used may suffer from poor uptake by clinicians for this reason. In many cases, there may be an optimum time for a trial, and if this opportunity is missed, evaluation may become impossible, as the new intervention is adopted into clinical practice. For example, a trial to determine the best mode of delivery for very low birthweight infants2 failed because practice had already changed towards delivery by caesarean section before the trial started. Clinicians were unwilling to randomise, despite the lack of evidence from RCTs supporting this policy. This problem is encapsulated in ‘Buxton's law’:3 it is always too early (to evaluate) until, unfortunately, it is suddenly too late!

There is some evidence that one of the major factors that discourages clinicians from participating in trials is lack of time. Two surveys found that clinicians felt that their clinical and management duties did not leave them enough time to participate in trials, especially if they expected that obtaining consent or carrying out the trial protocol would be complex or time consuming.4,5 NHS reforms in the UK may have decreased clinicians' time for research and hence made recruitment to RCTs more difficult.5 Understaffing of units may lead to increased pressure on the workforce and lack of time for research. Shortage of staff with particular skills (e.g. those that are needed to carry out the trial protocol6 or experience of recruiting to trials) may also preclude participation. If only a few members of the staff recruit to a trial, eligible patients will be missed and the recruitment rate will be well below what is achievable. Dissemination of information within a unit and training of all staff so that they understand the trial's rationale and procedures may be crucial for good recruitment. Complicated (and hence time consuming) trial procedures are also likely to decrease participation,7 and these should be simplified wherever possible. Trials may be most acceptable to clinicians if they involve minimal additional procedures, and hence can be integrated into clinical practice.

A further disincentive to clinicians' participation may be the perceived lack of rewards for themselves or their department from taking part in collaborative multicentre trials. Clinicians in academic departments may not view collaboration in such trials as fulfilling the criteria on which they are judged, and they may therefore prefer to devote their research time to their own single-centre research projects rather than joining a multicentre collaboration. There is generally no financial reward (such as a fee per recruit that is paid to the department or hospital) from participation in publicly funded RCTs, so there may be more pressure on clinicians to participate in research that brings more tangible rewards, such as pharmaceutical industry sponsored trials. This may have a further impact on recruitment, as it is becoming increasingly common that participation in pharmaceutical company trials precludes joining other trials. These patients may therefore be lost as potential recruits to academic research, when there is no scientific reason for their exclusion.8

Even if clinicians are in principle keen to participate in a trial, there are a number of administrative hurdles that may be time-consuming and frustrating. For instance, although the trial co-ordinating team will be able to obtain multicentre research ethics approval, local clinicians will still need to apply for approval from their local research ethics committee. Although this should be straightforward, in practice there are often long delays. Plans to speed up this stage will be introduced in 2004.9 A recent development is the need for approval from the Trust's R&D Committee in addition to local research ethics committee approval. In many cases, this committee only receives the project after it has been approved by the local research ethics committee, so there is a further layer of bureaucracy and more delay before recruitment can begin.

A variety of other operational problems may hinder recruitment. For example, in a recent multicentre trial, the pharmacist in one centre insisted that the trial drugs were stored in the hospital pharmacy. This made it almost impossible for staff to recruit out of office hours, with the result that potential recruits were missed and staff did not become familiar with the trial's procedures. This centre recruited very poorly. Another example is disruption of recruitment due to drug supply problems, which meant that recruitment had to be stopped for a period and then re-started; we found that, in many centres, dissemination of the news that recruitment had re-started was slow. Many members of the staff were unsure of the trial's status and did not approach potentially eligible patients.

Factors affecting patients' participation

  1. Top of page
  2. Factors affecting clinicians' participation
  3. Factors affecting patients' participation
  4. Improving recruitment to multicentre trials
  5. References

The factors that affect participation of patients in trials are likely to be different from those that influence clinicians. Any cost, inconvenience or increased risk that patients perceive that they may incur if they participate is likely to affect their decisions about joining a trial. Additional procedures or extra hospital visits are an obvious disincentive to participation, especially if they are likely to involve additional expense. Some patients may view research as inherently risky and prefer to choose existing treatments, although the limited available evidence suggests, if anything, that patients may benefit from participation in trials.10,11 In taking decisions about participation, patients may give greater weight to very small risks of serious adverse outcomes than to much greater risks of less serious events.

Some patients may have strong preferences for treatment (either for one of the trial's interventions, another treatment or no treatment) and may be unwilling for their treatment to be determined by randomisation. A report has been published of a trial that failed principally because of patients' preferences for one treatment,12 despite the enthusiasm of participating clinicians. In this case, the choice was between two treatments, but preferences may be more of a problem in placebo-controlled trials, where some patients may be unwilling to take a placebo. There is some evidence for this; one study found that fewer women would agree to join a trial of HRT including a placebo arm than a trial without a placebo, although the difference between the groups was not very large (30%vs 39%).13 Random allocation of treatment may also be a problem, as some patients may not understand what it involves14 and may decline trial participation because of erroneous concerns.

Patients' personal knowledge and opinions about research are likely to play a role in their decision about trial participation. Research is generally portrayed in a negative way by the media,15 particularly in relation to the care of babies and children, and this may reduce the number of recruits to trials.

Improving recruitment to multicentre trials

  1. Top of page
  2. Factors affecting clinicians' participation
  3. Factors affecting patients' participation
  4. Improving recruitment to multicentre trials
  5. References

It is unlikely that there will be changes to the structure of health services that will make recruitment to multicentre trials easier in the near future. In the long term, health professionals' training may change to give more prominence to research. If clinicians are encouraged to view recruitment to trials as part of their normal clinical practice, more patients may be randomised. In a few fields, this process is well advanced; evaluation of treatments is well integrated into clinical practice, and a high proportion of patients are offered the chance to participate in one or more RCTs.16

Poor recruitment may have many different causes, and it is often difficult to predict before a trial starts what problems will arise. There has been relatively little research on the reasons for poor recruitment. Several reports of ‘failed trials’ that recruited very poorly or not at all included surveys of clinicians or patients asking about their reasons for non-participation.2,6,8 It is difficult to draw any general lessons from these studies, but they suggest that a variety of mechanisms may be operating to hinder recruitment to trials and triallists may find it difficult to predict whether a trial will be able to recruit adequately.

Many possible interventions to improve recruitment have been suggested and some used with apparent success, but there is an almost complete lack of rigorous evaluation of these strategies. A systematic review of barriers to participation in trials provided a number of recommendations (e.g. using multiple recruitment strategies, setting recruitment targets and providing training),1 but most of these were based on anecdotal evidence. Little high-quality evidence exists to suggest which interventions are likely to be most effective. Moreover, an intervention that is effective in one trial may not work in another, where the problems may be different. For example, intensive site visits by the study team may help in situations where the main problem is dissemination of information in participating units, but may be ineffective if the problem is patients' unwillingness to take a placebo.

The need for large, multicentre RCTs is likely to increase in the future as maternal and neonatal mortality and morbidity become rarer. Trials need to be run as efficiently as possible, so that the limited funding available will be able to answer more questions, ultimately bringing greater benefit to more patients. At present, there is little or no high quality evidence about which strategies are effective in improving recruitment to trials and when they should be used. There is therefore a clear need for evaluation of interventions to improve recruitment, using RCT methodology and testing interventions in a wide range of situations to explore their generality.

References

  1. Top of page
  2. Factors affecting clinicians' participation
  3. Factors affecting patients' participation
  4. Improving recruitment to multicentre trials
  5. References
  • 1
    Prescott RJ, Counsell CE, Gillespie WJ, et al. Factors that limit the number, quality and progress of randomised controlled trials. Health Technol Assess 1999;3(20): .
  • 2
    Lumley J, Lester A, Renou P, Wood C. A failed RCT to determine the best mode of delivery for preterm infants. Control Clin Trials 1985;6: 120127.
  • 3
    Buxton MJ. Problems in the economic appraisal of new health technology: the evaluation of heart transplants in the UK. In: DrummondMF, editor. Economic appraisal of health technology in the European Community. Oxford: Oxford University Press, 1987.
  • 4
    Dickinson CJ. Clinical research in the NHS today. J R Coll Physicians Lond 1994;28: 460463.
  • 5
    Smyth JF, Mossman J, Hall R, et al. Conducting clinical research in the new NHS: the model of cancer. UK Coordinating Committee on Cancer Research. BMJ 1994;309: 457461.
  • 6
    Penn ZJ, Steer PJ. Reasons for declining participation in a prospective randomized trial to determine the optimum mode of delivery of the preterm breech. Control Clin Trials 1990;11: 226231.
  • 7
    Hetzel MR, Lee T, Prescott RJ, et al. Multicentre clinical respiratory research: a new approach? J R Coll Physicans Lond 1998;32: 412416.
  • 8
    Brocklehurst P. Randomised controlled trials in perinatal medicine 2: recruitment of a pregnant woman or her newborn child into more than one trial. Br J Obstet Gynaecol 1997;104: 765767.
  • 9
  • 10
    Braunholtz DA, Edwards SJL, Lilford RJ. Are randomized clinical trials good for us (in the short term)? Evidence for a “trial effect”. J Clin Epidemiol 2001;54: 217224.
  • 11
    Vist GE, Hagen KB, Devereaux PJ, Jackowski D, Oxman AD. Outcomes of patients who participate in randomised trials versus those of similar patients who do not participate[Cochrane methodology review protocol]. The Cochrane Library, Issue 4. Chichester, UK: John Wiley and Sons Ltd, 2003.
  • 12
    Rogerson L, Duffy S, Crocombe W, Stead M, Dassu D. Management of menorrhagia—SMART study (Satisfaction with Mirena and Ablation: a Randomised Trial)[letter]. Br J Obstet Gynaecol 2000;107: 13251326.
  • 13
    Welton AJ, Vickers MR, Cooper JA, Meade TW, Marteau TM. Is recruitment more difficult with a placebo arm in randomised controlled trials? A quasirandomised, interview based study. BMJ 1999;318: 11141117.
  • 14
    Snowdon C, Garcia J, Elbourne D. Making sense of randomization; responses of parents of critically ill babies to random allocation of treatment in a clinical trial. Soc Sci Med 1997;45: 13371355.
  • 15
    Bartlett C, Sterne J, Egger M. What is newsworthy? Longitudinal study of the reporting of medical research in two British newspapers. BMJ 2002;325: 8184.
  • 16
    Department of Health Cancer Working Group. Strategic priorities in cancer research and development. www.doh.gov.uk.