This study has provided a clear understanding of the perspectives of clinical staff attempting to recruit patients to pragmatic complex intervention RCTs. The findings highlight the fragility of the process of recruitment, and the overt/covert, witting/unwitting influences that recruiters can have on the success or otherwise of recruitment to RCTs. Recruiters readily identified organisational difficulties, fewer than expected eligible patients and strong treatment preferences expressed by patients as the key barriers inhibiting recruitment, but were seemingly unaware of their own influences on these obstacles. A key finding from this research was how much discomfort and emotion recruiters expressed about their research and clinical roles while also indicating strong commitment to the RCT and research in general. Doctors emphasised the complementarity of their roles as scientists and clinicians, but were uncomfortable about particular groups or individual patients, clashes with their clinical judgments, and their own preferences for particular interventions. Nurses identified conflicts arising from their roles as caring nurses, patient advocates, and (somewhat denigrated) recruiters/researchers. They were uncomfortable and defensive about the decisions they made about whom to approach about the RCT, sometimes avoiding ‘bothering’ patients, making ostensibly ‘clinical’ decisions that strayed into personal views, even prejudices, and tended to transmit their own views about treatments and organisational difficulties.
Recruitment was very difficult in these RCTs, as it is in many. There were many examples in the interviews of how recruiters privileged their clinical roles and judgments over recruitment, leading to large numbers of patients who were technically eligible for the RCT but who were not informed about it, and many others who were subtly (or not so subtly) manoeuvred towards a particular intervention for reasons that might have been clinically appropriate but more often reflected the recruiters’ views. It seems plausible that these findings would also be present in other pragmatic RCTs. They are novel because they help to explain why problems with recruitment can persist even when CIs work hard to resolve organisational problems that appear to be the source of the difficulties [2–6]. Recruiters in these RCTs often expressed strong commitment to the RCTs, but they found the issues raised by actually recruiting individual patients so discomforting that they found ways to avoid it, keeping their discomforts away from colleagues or RCT leaders.
Patients’ treatment preferences were identified as a major reason for poor recruitment in all these RCTs, as they were in other studies [16, 17]. The nuanced analysis in this study suggested that some (perhaps many) of these preferences might actually be contributed to by the recruiters’ own views. A recent study of 93 consecutive recruitment appointments in T6 showed that patients’ treatment preferences ranged on a continuum from a strong desire to receive a treatment to a mild wish, based on a range of detailed or sparse, accurate or inaccurate information, gleaned from a variety of formal and informal sources . When nurse recruiters carefully elicited and explored these preferences and provided detailed information about the RCT and its interventions, only a minority of patients remained committed to their initial preference, and the majority (75 % of those who had originally expressed a preference) became open to other options and 80 % of these then consented to randomisation . The evidence from this synthesis was that recruiters often expected patients to have preferences, and when they heard preferences that seemed to make sense because they accorded with their own views, it was easy to accept them on face value, without further exploration. Some recruiters justified this by stating that it would be coercive to pursue things further; the balance between coercion and accepting stated preferences without checking they are based on accurate information would benefit from further investigation.
Doctors’ accounts showed that they were particularly uncomfortable with the formal RCT inclusion/exclusion criteria, reflecting concerns about ‘equipoise’. The literature on equipoise in relation to RCTs has largely been theoretical, focusing on its use as the ethical justification for randomisation [19–21]. The concept has been strongly criticised for its lack of relevance to contemporary healthcare [22, 23], but there is remarkably little empirical research investigating how doctors talk about or operationalise equipoise during RCT recruitment (see, for an exception, ). In this study, most of the doctors could explain the concept in a scientific sense and their reliance on a sense of uncertainty as a community of experts, but they often went on to indicate their lack of uncertainty (certainty) in relation to specific patients or subgroups of patients or particular interventions. As a result, they made their own decisions about whether to apply the criteria fully or not, but did not raise these issues with the RCT CIs. This ‘variable enactment of the protocol’ was also found in another qualitative study of clinical centres in a single RCT , and is probably widespread.
Nurses were particularly vexed by the conflicts they perceived in their roles. This has been found in other small-scale studies, particularly for nurses working on the boundary between clinical work and research [26–28], but the influence on recruitment has not been shown previously. Most nurses, whether employed by central networks or by specific RCTs, saw research roles as of lower status and were clear that their clinical role would always come first ‘at the expense of the research’. Their discomfort was covertly expressed through their concerns about ‘not bothering’ patients with information about the RCT, or making ‘clinical’ judgments. For most, the discomfort caused anxiety; for a small number this anxiety became quite severe. Ultimately, however, their views and actions meant that many patients were not approached for recruitment, or were influenced in their decisions about treatments outside the RCT. It will be interesting to see if this changes with the introduction of initiatives such as the UK’s ‘OK to ask’ campaign encouraging patients to request research participation .
These issues have relevance for the increasing number of RCTs that are currently being launched. RCTs are needed so that only the most effective and cost-effective healthcare interventions are used, but they are extremely expensive to undertake. Failure to complete recruitment wastes considerable resources, and fears of recruitment difficulties mean that some key healthcare questions are avoided. These findings suggest that training and support for recruiters based on integrated qualitative studies to investigate the enacted recruitment process, might help with completing recruitment (as it did in T6, and to some degree in T3 and T5), or in clearly identifying the issues that led to the closure of T1 and T7.
The findings are also important in the context of the increasingly important contribution that nurses make to RCT recruitment. Over the past decade, nurses have become the main workforce in network organisations such as the UK NCRN, which co-ordinates recruitment for cancer RCTs, including T1, T2, and T7 . Many have taken on this role without much formal training – often simple courses in ‘good clinical practice’ , or occasionally more advanced specialist courses . As recruitment to RCTs is increasingly placed in the hands of nurses, if the views expressed here have currency elsewhere, then considerable opportunities for RCT recruitment are not being taken. Nurses had considerable power to decide which patients would be given information about the RCT, justified by being the patient’s advocate. Advocacy is a relatively new addition to nursing, and while it is often promoted as a moral imperative, it is fraught with dilemmas [31, 32]. It will be important for further research to examine the role of advocacy and clinical decision-making in nurse-led recruitment. It was remarkable how little insight nurses seemed to have about the judgments they were making, and how little evidence there was that they had taken time to check the veracity of their opinions with patients. Training and support in T6 enabled nurses to achieve very high rates of recruitment and informed consent . There was evidence in this synthesis that the nurses still continued to exhibit some concerns about their roles, but they had access to support that mitigated their anxiety. These nurses, and those in T3, were easiest to train because they were employed by the RCTs and time was dedicated to it by the CI; it was very difficult to meet with groups of nurses who were employed by the national networks such as the NIHR NCRN because they were responsible for recruitment to many RCTs and their supervisors were unwilling for them to devote time to particular RCTs. The intensive training and support in T6 may not always be possible, but this synthesis has emphasized the need for standard and continuing training for all nurse recruiters, to include aspects of RCTs, roles, and social awareness, and not just good clinical practice. It has also been suggested that standard and continuing education is needed for physicians about how participation in RCTs varies from standard clinical practice . The evidence from this synthesis suggests that all recruiters (physicians, surgeons, nurses, and other professionals) would benefit from training incorporating basic and advanced details of the RCT design and how to present treatments and randomisation fairly and clearly.
Strengths and limitations
This study is the first detailed and wide ranging exploration of the perspective of recruiters involved in recruitment in several (six) pragmatic RCTs underway with recruitment difficulties. A strength of the study was in the use of qualitative methods to understand the perspectives of those actively engaged in recruitment, producing a detailed and rich synthesis that allowed a nuanced understanding of the processes underlying recruitment in each RCT across a range of clinical interventions, recruitment methods, and contexts (Table 1). Qualitative research methods are particularly suitable for gaining an understanding of complex issues and systems and the perspectives of particular individuals and groups.
The study has a number of limitations. The RCTs were included pragmatically because they included or added a qualitative study with the aim of improving their recruitment, and so might be different from other RCTs. There was diversity in the RCTs in terms of clinical contexts, but four involved oncology and three surgery. There were differences in the numbers and characteristics of those who were interviewed in the RCTs and interviews and group discussions were analysed pragmatically – a function of the different management and recruitment arrangements. It is likely that only those most willing to speak about the RCT and recruitment issues agreed to be interviewed, and so there could be different views among other recruiters. The data were collected across quite a long time period because of the reliance on the funding of the qualitative studies. Accumulating evidence of this sort across complex studies is difficult and this is the largest synthesis to date. The consistency of findings over time suggests that the insights are robust, but it cannot be ruled out that other or more recent RCTs recruit better (although recruitment problems remain the key challenge to RCTs today). It is important to acknowledge that the RCTs here were all having recruitment difficulties. Others may have ‘quick fix’ issues that simple interventions such as telephone reminders, etc., may solve [2, 6]. However, for RCTs where the topic is controversial or the evidence is contested, or interventions are very different (including ‘no’ or ‘control’ intervention arms), it is likely that patients and clinicians will have strong views or preferences, and it may be that the findings have most relevance for these most ‘challenging’ RCTs.