Summary of main findings
The 20 interviewees had positive views of research and of the aims of the REFRAMED trial in particular. Many had experience of research participation. The interviews enabled us to identify four stages in the process of deciding whether or not to participate in the REFRAMED trial. At each stage some respondents concluded their deliberation and opted out. In stage 1 the ‘prior decliners’ opted out, who have an established position of declining trial participation, stemming from personal circumstances, for example viewing themselves as ‘too old’. In stage 2, the ‘self-excluders’ who use the trial eligibility criteria to declare themselves ineligible opted out; they see their illness and its management differently from the clinical team who invited them to participate. In stage 3 the ‘treatment decliners’ opted out, who perceive that they may be eligible, but focus on their health needs and decide that they do not need the trial therapy. In stage 4 the ‘trial decliners’ opted out, who perceive that they may be eligible and in need of the trial therapy, but focus on the burden of trial participation and decide that that outweighs potential benefits.
Strengths and limitations
Our study adds to the very sparse literature on non-participation in randomised trials. To our knowledge this is the first qualitative study to explore explicitly how decisions to decline invitations to mental health trials were made and to present the results in a conceptual framework of decision making.
There are gender and age differences in the presentation and diagnosis of depression [
44,
45], and most primary care depression trials enrol many more females than males [
46]. Our sample of 14 (70 %) women and six (30 %) men, with ages ranging from 18 to 77 years, reflects the demographics of depression trials and is a strength of this study.
We used telephone and e-mail interview methods and it is possible that, compared with face-to-face interviews, these may compromise rapport, probing and interpretation of interview responses [
47]. However, using these methods enabled us to interview a hard-to-reach group who otherwise may not have engaged [
48,
49] and to achieve a degree of anonymity which arguably helped interviewees to disclose their experiences.
It is possible that interviewees present themselves as rational deliberators in studies of this sort, because that is what they perceive is expected of them. We minimised this risk by asking interviewees simply to report what happened when they received the trial invitation, rather than to provide detailed elaborations of their decision making process and some – the ‘prior decliners’ who had previously made similar decisions to decline other trials – clearly reported that they made the decision with little deliberation. Some interviews occurred months after the initial refusal. While some respondents had difficulty recalling details, most recalled the invitation and decision process in detail and provided vivid accounts.
As in all studies of volunteers, informants selected themselves. However, participants represented only 16 % of decliners, which may limit the transferability of our findings. Interviewees expressed very positive views of research, presumably because, like other studies of non-participation, we could not access those averse to research. However, we doubt whether research-averse individuals could help to enhance recruitment, as they would not respond to recruitment interventions.
Patients who declined after being directly approached by clinicians to participate in REFRAMED also could not contribute to this study. Such patients may have offered different views, particularly around eligibility and self-exclusion issues, since clinicians were perhaps more likely to approach those whom they were confident would meet the trial eligibility criteria.
Whilst we undertook purposive sampling, the small numbers of patients who responded limited the scope of that. Despite this we did reach data saturation with those interviewed. Finally, the novel treatment in REFRAMED was aimed at patients with refractory depression and was particularly intense, so findings may not be transferable to other depression trials.
Comparison with existing literature
Our meta-synthesis [
23] shows that patients’ decisions to enter depression trials depend on: their health at the time of the invitation; their attitudes towards the research and trial interventions; and the demands of the trial. Our conceptual framework describes how decisions to participate require judgment between ‘risk and reward’. This qualitative study supports that meta-synthesis by showing that in making their decisions, respondents balanced their current health and whether they would benefit from the trial therapy against the burden of participating in the therapy including travel and time. In planning this study, we sought to contribute to existing knowledge. For example, we focused on patients under-represented in the previous literature by exploring how those who opted out of REFRAMED made their decisions.
Our findings reflect the wider decision-making literature, in particular the ‘deliberation and determination’ framework [
25]. This framework differentiates between the pre-decisional process of deliberation, the act of determination and post-decisional outcomes. Our findings and the stages appear to match this process of ‘deliberation’, in which the person considers the invitation in light of their eligibility, experiences and need; and determination, which is the act of choosing to not participate. Our classification of individuals as ‘prior decliners’, ‘self-excluders’, ‘treatment decliners’ and ‘trial decliners’ appears to reflect the ‘determination’ phase of the deliberation and determination framework.
Our findings in this subgroup contrast with the general literature which suggests that altruism is a major reason for research participation [
50‐
52]. Our respondents initially assessed their eligibility for the trial, then focused on their need for the trial therapy, and their potential to benefit. There is evidence that perceived ineligibility can lead people with depression to decline trial participation [
24], and that patients participating in trials focus on the therapy under review and consider personal benefits from it [
53‐
57]. The term ‘conditional altruism’ describes willingness to help others that inclines people to participate in trials, but does not clinch trial participation unless they judge that this will benefit them personally [
57]. Whilst interviewees appeared to understand that randomisation meant that those who enrol might not receive the trial intervention, their accounts revealed the perception of randomisation in treatment trials as fundamentally unfair, even ‘cruel’ in cases where people may be seeking treatment through trial participation. Thus our group of decliners demonstrated similar attitudes to those who enrol to gain therapeutic benefit from trial participation. A relevant concept is the
therapeutic misconception – a blurring of research and treatment, and thus a threat to understanding the trial and its risks [
58‐
61]. There is some evidence that patients who decline participation often misunderstand the nature of the research [
62,
63]. More pertinent to our interviewees, however, may be the concept of the
therapeutic mis-estimation, which misunderstands the likelihood of risks and benefits rather than the general purpose of trials [
64].
We found that interviewees had positive attitudes to research and the trial. This contrasts with some literature on non-participation which reports that decliners are less supportive of research [
65‐
67]. Despite not participating, our interviewees generally did not mind being invited and felt free not to participate. There is evidence that most patients with mental health problems approve of psychiatric research [
50], and that non-participation does not reflect objection to research in principle [
63,
68]. Patients who opt out of trials have reported that they do not object to being asked to participate, nor do they feel any pressure to do so [
69].
Implications for recruitment practice and future research
Our findings have several implications for trial recruitment and ethical and methodological research on it. First it is important to recognise that those whom we term ‘prior decliners’ are unlikely to respond to any recruitment initiative as they have an established stance of declining all trial invitations. However, other factors leading patients to opt out of trials may be open to amelioration as they do not arise from a rejection of trials or personal stances of declining such invitations.
To improve responses to postal invitations in similar trials, the most successful interventions are likely to address patients’ assessments of their eligibility and their potential to benefit from the trial treatment, rather than reducing the burden of that treatment. Trialists can influence patients’ assessments of eligibility by exploring methods of:
(a)
managing electronic patient records to estimate eligibility more precisely;
(b)
influencing patients’ own assessment of eligibility and their judgments of their potential to benefit from the trial treatment; and
(c)
drafting trial invitations, for example to minimise the risk of excluding themselves as ineligible.
The wording of invitations could be evaluated to examine the effect of conveying broader criteria on the numbers initially expressing interest, and ultimately enrolled. It is unclear whether ‘self-excluders’ make the same decisions that the trial team would, and whether the trial team would also have excluded them as not meeting the inclusion criteria. Thus trialists could evaluate a trial invitation letter which lists the precise inclusion and exclusion criteria against a comparator invitation which lists only the condition under investigation (e.g. ‘depression’), to estimate how many people initially respond in each arm, how many are excluded by the trial team and how many are ultimately enrolled. While eligibility issues are complex, there may be a case for accepting the risk of attracting more patients who turn out to be ineligible rather than being too restrictive. However, our findings caution against raising patients’ expectations in a way that would be unrealistic.
We know from our study that most patients focus on their need for the trial therapy when deciding whether to participate, whatever their final decision. Thus Miller and Brody [
70] and Schlichting [
71] have argued for trials to serve health needs, by abandoning the traditional commitment to clinical equipoise and conducting research ‘with therapeutic intent’. This approach replaces the ethical framework of equipoise with that of non-exploitation, so as to achieve the goals of patients, clinicians and researchers [
71]. Though detailed examination of this ethical dilemma is beyond the scope of this study, trialists should know that our respondents effectively supported this radical proposal. The implication of accepting the principle of research ‘with therapeutic intent’ is that trials should aim, not only for a favourable benefit-risk ratio for society, but also to avoid an unfavourable benefit-risk ratio for each trial participant [
72,
73]. Our qualitative study suggests that trialists should prospectively monitor patients’ expectations of their trials and use that to inform design and delivery. Better, patient-centred explanations of the potential benefits of trial treatments may help [
74]. Engaging service users and members of the public in the design and conduct of trials alongside qualitative research may be the key to this [
75]. For example, qualitative research could explore patient treatment preferences [
76,
77]. Thus a priority for future research is the presentation and provision of accurate and effective trial information in which patients and the public play a seminal role [
78]. Retrospective but timely feedback from patients who opt out of trials can assess the acceptability of the treatment being evaluated [
24]. Early inclusion of such feedback into trial recruitment procedures can increase participation rates [
79]. However, all such interventions require robust evaluation, ideally through embedded randomised trials.