Background
Surgical trials have typically experienced recruitment difficulties compared with other types of oncology trials in the UK [
1]. There are several reasons for this suggested in the medical literature. On the one hand, the culture amongst surgeons is considered to be less research oriented, and surgeons less experienced in conducting trials, compared with other clinical specialties engaged in cancer medicine [
1]. There are also specific challenges with using randomised trials to compare types of surgery, such as the impracticalities of blinding, resistance to randomisation amongst surgical teams, difficulties attaining clinical equipoise, and the enhanced training that is necessary for new treatments [
1]. Following this, clinicians may also need to accept that there is uncertainty or equipoise about treatments that are established in usual practice [
2]. That said, there are examples of current recruitment ‘success stories’ in surgical oncology trials [
3], and novel attempts to incorporate blinding into such trials [
4], suggesting that such challenges are not necessarily insurmountable. Finding ways to improve the evidence base for surgical procedure is imperative [
1,
3], as is the need to do so in ways which minimise ‘research waste’ [
5‐
7]. Improving the efficiency of trial processes, including recruitment, is one such way of accomplishing this, although to date this remains a ‘largely evidence-free zone’ [
8].
Qualitative research has an important role to play in investigating and addressing recruitment issues in trials. Studies investigating patient perceptions and experiences of surgical trials have helped illuminate why such trials experience difficulties recruiting [
2,
9‐
13]. In the ProtecT, SPARE and QUEST trials (trials in prostate cancer, bladder cancer and breast reconstruction surgery) patient preferences for particular treatment options undermined recruitment into these trials [
2,
10‐
13]. In the ProtecT and SPARE trials recruiters were found to unwittingly pass on their preferences to patients [
10,
12], including the use of ‘loaded’ terminology [
2,
11]. In the CLasP trial (which compared treatments for men with symptoms of benign prostatic disease) it was also observed that clinicians played a role in eliciting preferences and deciding who could participate [
9].
Participants in these trials also experienced difficulties with the concept of randomisation. Whilst participants understood the principles of randomised design (i.e, chance, comparison and equipoise) [
10], they struggled to accept randomisation because they could not accept ‘equipoise’; that the clinician was genuinely uncertain and the treatments similarly effective [
2,
10,
11]. Participants in one trial also had expectations that their clinical appointments would be used to make individualised treatment decisions [
9]. Such examples, therefore, suggest that it is not necessarily patient (mis)understandings of trial/medical information that is critical to patient decision-making, as is often assumed [
11], but the culturally situated expectations that patients have of their ‘experts’; an argument which has been well-developed in qualitative research investigating public uptake of science [
14‐
16]. This and other research also usefully highlights the different types of expertise, knowledge and forms of reasoning that people draw upon and develop when confronted with illness or perceived threats to their health and wellbeing [
17‐
20], including ‘experience based expertise’ [
17] and the culturally framed, personal, subjective and social knowledge, often described as ‘lay knowledge’ [
18,
19].
The BOLERO trial (Bladder cancer: Open versus Laparoscopic or RObotic cystectomy) is a feasibility study that was developed on behalf of the National Cancer Research Institute (NCRI) Bladder Clinical Studies Group and funded by Cancer Research UK, coordinated by the Wales Cancer Trials Unit and Marie Curie Palliative Care Research Unit. The study aim was to determine the feasibility of randomisation to open versus minimal access (laparoscopic or robotic) cystectomy in patients with invasive bladder cancer, aged 18 or over. The secondary aims were to assess the safety and efficacy of minimal access cystectomy, and the reasons for non-acceptance of randomisation [
21]. Open and minimal access surgery were routinely available to all patients in our six participating trial sites, regardless of whether or not they entered the study. Some of these centres were also offering robotic surgery to their patients (trial and non-trial); at the other centres laparoscopic was the only minimal access option available.
The trial was designed with a quantitative primary end-point of feasibility of randomisation, defined as >60 % of patients accepting randomisation, which would be used to determine whether a future randomised phase III trial, requiring many hundreds of patients, was possible to achieve. Within this trial an embedded qualitative study was included to explore factors relating to a decision to decline randomisation and patient experiences of trial recruitment processes. Although not designed to identify and implement changes to the feasibility trial, as in other studies [
2,
10‐
13], it was intended that the findings could be used to make trial specific recommendations for improving recruitment practices and documentation should the trial have progressed to a full trial and more broadly be used to inform the design and conduct of similar trials in the future by contributing to the small amount of evidence in this area. A more detailed description of trial design is available on-line [
21] and the quantitative outcome, including activity and safety of the treatment arms, will be reported elsewhere. For the background purposes of this paper, however, it is reported that the threshold of >60 % of patients accepting randomisation was not reached, and a full phase III trial would not, therefore, be considered feasible. This paper reports the qualitative outcome which not only enhances understandings of recruitment issues relating to this particular trial, but more widely, adds to the qualitative evidence described above, and identifies implications for the evaluation of similar surgical/technology interventions in the future.
Methods
The aim of the qualitative study was to explore factors relating to a decision not to consent to randomisation between open surgery and minimal access surgery (robotic or laparoscopic), and participants’ experiences of recruitment processes in the BOLERO trial. The embedded qualitative design involved semi-structured interviews with up to 20 patients and an in-depth thematic approach to analysis of the data. This research was informed by a phenomenological perspective and interest in exploring the lived experiences of participants and their decision- making in the context of their life experiences, values and beliefs. For the purposes of this paper and our clinical audience, however, our results are presented as group themes which cut across patients’ views and experiences, but where relevant also identify the salient life context and stories of individual participants in relation to the theme being discussed.
Ethical approval for the BOLERO trial (including the qualitative study) was granted by South East Wales Research Ethics Committee in May 2010, and the trial was sponsored by Cardiff University. Ten participants were recruited into the qualitative study from three (out of a potential six) sites which were recruiting into the main BOLERO trial. Robotic surgery was not available at one of these three sites. The interviews were conducted between October 2011 and December 2012, and included eight male and two female participants, with an age range of 44 to 74, and a mean age of 63.5 years. Participants (P)2 and P10 were the only patients from the site without robotic surgery; all other participants had the option of receiving robotic surgery.
Informed consent was obtained for all participants. All patients who declined randomisation into the main BOLERO trial were given information about the qualitative study by their research nurse. Patients who wanted to participate in the qualitative study were consented into the study by their research nurse before being contacted by the qualitative researcher to make interview arrangements. When the minimum number of participants was deemed sufficient for meaningful analysis, and theoretical saturation was reached, recruitment was closed to the qualitative study. Just under a quarter of the patients who were eligible for this qualitative study participated in an interview.
In response to initial slow recruitment to the qualitative study, the study protocol was also amended to include a further study involving semi-structured interviews with the six research nurses recruiting to the BOLERO trial. Research nurses have immediate experience of screening, providing trial information to, and consenting prospective participants. In this trial, the research nurses would talk through the trial information in detail with patients, after the surgeon or another member of the clinical team had introduced and explained the trial to the patient during their clinical consultation. They were typically also the first point of contact for patients once they had made a decision on whether or not to enter the trial. It was thus considered that they would have good experience-based knowledge of why patients declined or accepted randomisation to the trial, as well as views on the process of presenting and explaining the trial information, and any difficulties that this may present. Semi-structured interviews with the research nurses involved in this trial were used to explore their views and experiences with regards to recruitment and to provide insights into any problems and issues specific to this aspect of the trial.
Ten semi-structured interviews were carried out with patients in a location of the participants’ choice; patient’s homes (
n = 4); their place of work (
n = 1) and by telephone (
n = 5). Interviews were conducted by a female researcher with experience in cancer and healthcare research, but no clinical background. Interviews lasted between 20 and 45 minutes and explored patients’ experiences of trial information and recruitment processes, reasons for declining randomisation into the main trial and attitudes towards medical research (interview topics summarised in Table
1).
Table 1
Summary of interview topics
Trial information and recruitment processes • Experience of being told about the trial • Understandings of trial purpose and principles, e.g. randomisation • Views on participant information sheet • Other sources of information accessed about types of surgery/trial • Prior knowledge of and preferences for surgical methods Decision-making • Reasons for declining randomisation • Choice of surgery and reasons given • Influence of clinical team and others on decision-making • Reflection on choices made and decision-making process Experience of and attitudes towards medical research • Views on randomisaton as process of allocating treatment • Participation in previous and (potential) future research studies | Recruitment practices • Professional background and time on trial • Own and others’ duties and responsibilities with regard to the trial • Description and reflection on information giving process for trial/other options available to patients Patient responses • Recall of patient responses to trial information • Views on patients’ reasons for declining randomisation • Views on patients’ reasons for accepting randomisation • Reflection on differences between patients who accept/decline trial General reflections • Compare and contrast with recruitment experiences of other trials • Key challenges of this trial and suggestions for future trial design and conduct |
Interviews were audio-recorded, transcribed verbatim and uploaded to NVivo 9 software. Data analysis was undertaken by two researchers who shared the analysis of transcripts, following an in-depth thematic approach [
22]. Group results were analysed for consistent themes using techniques of coding and comparison. A coding framework for emergent themes was developed by the researcher. This was an iterative process, moving between the data and the analytical concepts to develop codes and concepts grounded in the data. Higher level abstractions of codes were decided and results were verified by the research team by independent review of a selection of transcripts.
Unfortunately, only two of the six research nurses agreed to take part in an interview and one of these interviews failed to record; the data from this interview are from notes taken during the interview and written up by the researcher from memory immediately afterwards. This seriously limits the contribution of these interviews to the study results. However, although this data cannot be interpreted as ‘findings’ in their own right, they do provide some useful context for the findings of the patient interview study and are reported as such.
Discussion
This paper contributes to the small amount of evidence from embedded qualitative studies which help to explain recruitment difficulties experienced by surgical trials. It identifies a perceived lack of equipoise between trial arms amongst our ‘non-consenter’ participants and draws attention to a number of factors which explain patient preferences and choices. As such, this study adds new insight to previous similar findings in this area and identifies important implications for future research aimed at evaluating the effectiveness of surgical/technology innovations, particularly those already available as part of routine practice.
In general, these study participants were well-informed and had a good understanding and appreciation of the trial and medical research more broadly. Most participants seemed to correctly understand the purpose of the trial as being to compare the different types of surgery, and felt happy with their level of understanding. As in previous surgical trials, participants understood the basic concept of randomisation as having treatment allocation left to chance, but could not accept this method of treatment allocation because they could not accept the supposed ‘equipoise’ in this trial and that their clinician was genuinely uncertain [
2,
10,
12].
As in previous studies, the main barrier to recruitment into this surgical trial seems to have been the emergence of clear treatment preferences amongst patients [
2,
9‐
13]. In this embedded study, most of the patients who declined randomisation did so because they had preferences for the surgical method they would be given, which in most cases was the robotic option. The ability of these patients to choose their preferred treatment option in usual care meant that they were unwilling to relinquish control by accepting randomisation and with it the possibility that they might not receive their preferred option, as similarly demonstrated in a feasibility trial of childhood strabismus surgery [
23]. Patient preferences for type of surgery were linked to their understandings of the different risks and benefits attached to each method. Most patients saw the benefits of keyhole/robotic surgery to be that it was less invasive, requiring smaller cuts and producing less scarring, smaller risk of infection and a shorter recovery period than for open surgery.
Patients drew on multiple forms of knowledge and expertise to inform their decision-making. Many had sought additional information from ‘accredited’ sources of expertise, such as published research studies available on the Internet, and friends and family members with medical backgrounds, whilst a few looked to the narrative accounts and ‘experience-based’ expertise of patients who had already undergone surgery [
17]. Patients also drew on their own personal, subjective and culturally framed knowledge [
18,
19]. For several patients their preference for robotic surgery seemed intuitive. It made sense to them (within their cultural frames of reference) that this would be the better option; it was new and technologically advanced whilst also perceived as sufficiently ‘tried and tested’. Patients also took into account their own illness histories and lifestyles when weighing up the different options available to them, to make what could be considered personalised treatment choices. These findings highlight the context dependent nature of knowledge and decision-making [
24,
25], and illustrate well the argument that patients judge treatment outcomes subjectively based on their personal circumstances [
25]. This causes difficulties for the practice of clinical equipoise in trials in so far as different people will attach different meanings to presented outcomes and may expect that these are taken into consideration when planning treatment pathways.
Participants had also been influenced by their clinical team, from whom they had expected and sought expert guidance and direction. This influence occurred through direct and indirect messages. Only a small number of patients felt that their consultants conveyed equipoise or messages of uncertainty with regards to the different surgery options. As in previous surgical trials many patients felt that their surgeon had openly favoured a particular trial arm [
2,
10‐
12], in this case the robotic method. Many patients also interpreted indirect or subliminal messages during their clinical encounters. These ‘indirect’ cues included positive comments made about fitness and suitability for robotic surgery during pre-operative assessments; the strong reputations of their surgeons in delivering pioneering robotic surgery (and assumed preferences for this technique), and the fact that the ‘new method’ had become well- established at these centres. In some ways this contrasts with the findings of an embedded study in another surgical trial, in which participants responded to loaded terminology in favour of the established (as opposed to ‘experimental’) treatment [
2]. What both studies more importantly illustrate, however, is the salience of hidden messages in unwittingly shaping patient preferences, and the openness of patients to such cues as they actively seek out ‘expert’ influence and confirmation.
Following this, a number of patients were explicit about wanting greater input from their consultants when making their treatment decisions [
23]. This preference aligns with research on decision-making preferences in healthcare, which notes that whilst patients welcome the opportunity to participate in decision-making it appears they do not necessarily want to be responsible for decisions, with a sizeable minority to a half preferring passive roles [
26‐
28], compared with a much smaller (and generally better educated, younger) minority who prefer purely autonomous roles [
27‐
29]. In this study, it was the older (but well-educated) patients who expressed greatest frustration at the lack of direction, even reflecting on what could be conceptualised as a ‘culture clash’ between the values and expectations of their generation and current medical models of patient-led choice and calculated risk assessment [
14‐
16]. These patients struggled to accept randomisation not just because of the preferences which they formed but because they had an underlying expectation that their clinicians would make treatment decisions based on individual clinical assessments [
9], thus further highlighting the need to contextualise knowledge, risk and decision-making [
14,
16,
24].
Limitations and implications for further research
Recruitment to the qualitative study was slow and just ten patient interviews were carried out. Although these patients represented a homogenous group and sufficient data were gained from these interviews to reach saturation and enable meaningful analysis, it should also be noted that a larger proportion of non-consenters to the clinical trial declined to take part in the qualitative interview. It is, therefore, possible that our sample included participants who were more positive towards medical research than the wider population of non-consenters who were eligible for this embedded study. This in turn may have biased our results in other ways given that a key concern of the study was with patient perceptions of, and engagement with research. However, it is unlikely that this bias should detract significantly from our main findings on patient preference and availability of choice as barriers to trial recruitment.
Whilst our interviews with research nurses were informative and provided some useful context it was disappointing that only two out of six nurses agreed to be interviewed, and only one of these interviews was recorded, thus limiting any conclusions that can be drawn from this data. Interviewing patients who decline the trial is only one part of the picture, a more comprehensive investigation of recruitment issues could have included observations of recruitment settings and practices, analysis and patient assessments of study documents and consultations and interviews with a larger sample of clinician recruiters, as well as participants who enter the trial. Ideally, this type of in-depth qualitative investigation of trial processes should take place before, or in the early stages of, the feasibility trial with formal mechanisms for feeding back key findings and designing and implementing practical changes to recruitment materials (e.g. DVDs, communication checklists) and staff training which could help to achieve a more effective presentation of equipoise. Such approaches have been shown to improve recruitment into trials [
10,
12], and are recommended for further studies of this kind. Given the apparent recruitment ‘success’ of a similar trial being conducted in the US [
3], exploration of inter-continental differences in trial recruitment practices could also be worthwhile.
Conclusion
Patients like to exercise informed choice over the type of surgery they receive, and they like this choice to be informed by their clinical team. In contrast with many cancer ‘drug trials’, where trial entry might be the only chance of receiving the ‘experimental’ treatment, for surgical trials where the ‘new technology’ is routinely available to patients there will likely be difficulties with recruitment. Going forward, an alternative option for similar technology trials could be to use centres which do not yet routinely use the technology under investigation, with access to the ‘new’ procedure restricted to trial patients at these sites, thus incentivising patient entry into the trial.
Greater attention could also be paid to how messages about treatment options and the trial are conveyed, including the more subliminal messages that pervade the whole clinical setting; a strategy which was tried and found to be effective in improving recruitment for another surgical trial [
10,
12]. However, if it is too difficult to challenge such messages, for example if the ‘experimental’ treatment is already entrenched in culture and practice (as seemed to be the case in this trial), then questions should be asked about whether equipoise can be presented and perceived with conviction in such trials. Moreover, accommodating patient wishes for clinician input and personalised consideration of individual circumstances and characteristics, as is expected by patients and practised in other areas of medical care, presents a challenge in clinical trials and can leave patients feeling frustrated and anxious if they perceive that this is not being accomplished in their consultations. Clinical equipoise is a difficult concept for patients to accept and feelings of anxiety and frustration are probably more likely to occur where there is not genuine conviction and belief in clinical uncertainty amongst recruiting clinicians.
These issues challenge the feasibility of similar trials in the future [
11], and have implications for current agendas on minimising ‘trial waste’; absent in which have been questions concerning the appropriateness of using randomised control trial (RCT) designs when evaluating certain types of interventions [
5‐
8]. These findings call for consideration of whether alternative trial designs or methods of generating evidence could be used when evaluating surgical techniques which have already become established and routinely available, and where there are likely to be clear patient and physician preferences [
1,
11,
13,
30].
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
EH carried out the interviews, analysed the data and drafted the paper. JK is chief investigator for the trial and helped to draft the paper. GG and AC are statisticians who were closely involved with the trial and made contributions to the content of the paper. AN designed and oversaw the qualitative study, and contributed to the data analysis and drafting of the paper. All authors read and approved the final manuscript.