Behavioural science is cross-disciplinary and has been considered as an umbrella term that includes contributions from various disciplines (including psychology, economics, sociology, political science, and anthropology). The field concerns how and why people behave as they do. Behavioural science as applied to health seeks to use the theories, methods, and knowledge from these disciplines to design more effective health care interventions. Within this article, we have focussed primarily on contributions from psychology but recognise that many of the other disciplines may have important contributions for clinical trials. The application of behavioural science to complex problems in health care has clearly been effective in changing both patient (e.g., smoking cessation) and health care professional behaviour (e.g., following recommendations for acute stroke care) as well as improving patient outcomes on both the short and long terms [5, 6]. For decades, implementation science has informed how to improve the uptake of trial results into practice, but lessons from the wider field of behavioural science have only recently been applied to problems of trial design and delivery.
Clinical trials are complex and made up of multiple processes at various stages of the trial lifecycle. These include, but are not limited to, question conception, trial design, grant and protocol writing, planning trial delivery, recruitment, intervention delivery, data collection, retention, analysis, dissemination of findings, and closedown. Understanding the influences on trial processes as multiple behaviours (performed by multiple actors), across the trial life cycle, has the potential for developing more effective evidence-based strategies for improvement. For example, recruitment can be further broken down to designing recruitment marketing materials (performed by investigator teams), approaching all eligible participants (performed by trial recruiters), signing of the consent form (performed by recruiters and participants), etc. Once a trial process is broken down in this way, it becomes more amenable to study and improvement with the tools of behavioural science. In what follows, we will present detailed examples of how behavioural science has been applied to trial processes in just this way.
When is a trial needed?
It can be difficult to determine when the evidence is strong enough to support the widespread implementation of an intervention or when further RCTs may be required. A study by Cuthbertson and colleagues sought to identify why the ICU community had not widely adopted the use of selective decontamination (SDD) of the digestive tract in ICU patients given the substantial evidence supporting the effectiveness of SDD from 12 meta-analyses of 36 RCTs [7]. Using a Delphi survey based on the Theoretical Domains Framework (TDF), the team were able to assess the factors affecting the clinical behaviour and the appetite for a further RCT [7]. In brief, the TDF is a comprehensive framework that proposes 14 theoretical domains that may influence behaviour (e.g. knowledge, behavioural regulation, emotion) [8]. Priority domains can be determined with regard to facilitators or barriers to performing the behaviour, which are then targeted when developing behavioural interventions [9]. The Delphi study concluded that the behaviours (in this case actively delivering SDD to ICU patients) would not be more widely implemented without further supportive evidence given the concern regarding the lack of appropriate/relevant outcomes in the existing trial contexts. This work directly informed the successful funding of an international trial of SDD in ICU patients with hospital mortality (primary) and antibiotic usage/resistance (secondary) as outcome measures. This approach of analysing the profile of behavioural responses to determine whether further (or indeed preliminary) trials are needed could be adapted for many clinical questions as one of the first steps in designing an RCT.
Is a trial feasible?
There are many ways that assessments of trial feasibility can be conducted. One of the benefits of assessing trial feasibility using a behavioural science approach is that it offers detailed identification of barriers and potential facilitators to performing key behaviours, which, in turn, drive the development of highly tailored, specific solutions with real potential to overcome feasibility challenges. To inform a future large-scale evaluation of a prehospital trauma intervention, ongoing work by Gillies et al. is developing a detailed behaviour specification and ‘diagnosis’ to identify the key challenges and opportunities for improving the feasibility and ultimate success of the future trial [10]. Specifically, interviewing health care professionals who are currently (or potentially would be) delivering the intervention will allow an understanding of behavioural challenges in intervention delivery and will provide evidence to help future strategies succeed for the future randomisation of participants.
Understanding the broad challenges for potential trial participants is also an important barrier to overcome when recruiting to a trial. Some studies have used surveys informed by behavioural theory, such as the health belief model (a model that assumes people’s subjective health considerations determine health-related behaviour), to investigate why patients choose to participate in trials [11, 12]. Brehaut and colleagues have taken this a step further by developing a theory-guided TDF survey to identify the challenges and opportunities to trial participation amongst potential participants, rather than amongst those who have participated [13]. The use of tailored surveys (which could be informed by the Brehaut approach) can be applied pretrial to determine what the main barriers to trial recruitment are likely to be and to facilitate recruitment strategies to address these barriers.
Do trial teams involve patients and public partners?
There are many motivators for involving patients and/or the public as research partners, not least of all to ensure the research is relevant for those it seeks to serve. The involvement of the patient and public partners in trials is now commonplace, but the extent and depth of that involvement vary significantly. A recent study by Goulao et al. surveyed trial teams to investigate the behavioural determinants of involving patient partners in numerical aspects of trials using a TDF-based survey [14]. The survey highlighted several domains that act as barriers (knowledge; skills and beliefs about capabilities; resources; reinforcement) which could be targeted with behaviourally specified interventions to improve current practice. This approach could be extended to the involvement of other stakeholders in the trial design and delivery process.
What are the challenges to trial recruitment?
Recruitment to clinical trials has been identified as the top methodological priority by UK Clinical Trials Units directors, evidencing its importance to many in the community [15]. Understanding the main challenges relating specifically to trial recruitment has been the focus of much research, but still very few high-quality, generalisable solutions exist [16]. A number of studies have applied behavioural science to understand the problems of trial recruitment. This has included conducting behavioural theory-informed qualitative interviews to understand the potential challenges to recruitment to early phase trials from the perspectives of clinicians and patients [17,18,19,20]. Findings from these studies were then used to refine the design and conduct of future trials. In addition to early phase trials, an exploratory TDF-based approach is currently being used to understand the challenges faced by health care professionals when recruiting pregnant women into clinical trials. The findings of the interviews will be used to develop, and subsequently test, a behaviour change intervention targeting professionals to improve the recruitment of pregnant women [21]. A similar approach has been used to develop an implementation intervention to address low recruitment to cancer clinical trials amongst rural and minority community urology practices [22]. This implementation intervention, termed ‘learn/inform/recruit’ was deemed appealing and acceptable by stakeholders [22]. The theory of planned behaviour (which proposes a model based on three variables: attitudes, subjective norms, and perceived behavioural control, which work together to predict the intention to perform a behaviour) has also been used to explore trial recruitment [23, 24]. TPB was used as a guiding framework to assess an intervention aimed at supporting patients in making fully informed decisions about lung cancer trials, highlighting that the application of this approach can be used with a range of theoretical approaches [23]. Using theoretical frameworks in this way is helpful for the individual trials as it enables more direct identification of possible strategies/techniques tailored to address the construct/factors more readily. A further advantage is this approach also allows the opportunity to combine data across studies and consider the meta-level findings of relevance across (possibly similar phased) trials.
Whilst many of the examples to date have been based on the TDF, a range of other behavioural theories and frameworks have been applied to problems of trial recruitment and retention. A recent mapping review identified 31 studies that used a range of theories/frameworks including the TDF, the theory of planned behaviour, social cognitive theory (describes the influence of the actions of others, experiences, and environmental contexts on an individual’s health behaviour), and others [Coffey et al. manuscript under review [25, 26]]. Establishing whether there are ‘best fit’ theories and frameworks for different trial problems is an important consideration for future work in this area.
How is the trial intervention delivered?
Process evaluations have long been embedded in randomised evaluations of clinical interventions to understand various aspects of delivery [27]. Many of these have included behavioural theories that have underpinned the behaviour change interventions being evaluated or indeed used theories (from a wide range of fields) to understand the mechanisms of change or barriers to implementation. However, less well addressed in this literature is the application of behavioural science to unpack the behaviours and behaviour change required for the delivery of clinical interventions within trials. Two recent studies have aimed to do just that. The first was with health care professionals delivering a trial of individualised temperature-reduced haemodialysis to explore the behaviours involved in adjusting the temperature on a dialysis machine [28]. The second was using a theory-based approach in data analysis gathered from both health care professionals and patients to explore trial experience and beliefs and experiences of the intervention, which in this case is catheter wash out policies [29].
What are the challenges to trial retention?
Similar to work on recruitment, a number of studies are now emerging that have conducted qualitative interviews informed by behavioural frameworks to understand trial retention behaviours such as postal questionnaire return and follow-up clinic attendance [30, 31]. Findings from the interview studies were then used to develop participant-centred, theory-informed interventions to promote trial retention that have been codesigned with stakeholders and will be tested in randomised evaluations [32].
The Cochrane reviews on interventions to improve recruitment to and retention in clinical trials have found very little evidence of effect [16, 33]. The reviews largely include interventions that were not designed as behaviour change interventions (BCIs) with only a minority (< 5%) conceptualised as BCIs, yet the implicit aim of the majority is to change participants’ recruitment or retention behaviour. For example, intervention categories in both reviews include incentives and rewards (which target the theoretical behavioural domain of reinforcement), reminders and prompts (target theoretical domain of memory, attention and decision-making, environment context, and resources), and improvements to information (target theoretical domain, knowledge). Yet, the design and delivery of these interventions do not include the explicit inclusion of behaviour change input, nor are these interventions informed by the bodies of knowledge in the behavioural sciences. Deconstructing interventions into their behaviour change techniques (BCT, defined as the smallest ‘active ingredient’ of an intervention that can be used alone or in combination) has the potential to identify possible ‘active ingredients’ which could be enhanced in future replications of evaluations or implementation [34]. Duncan et al. demonstrated the potential value of this approach with preliminary work identifying BCTs within interventions shown to improve retention [35]. The findings identified that BCTs were used amongst the interventions but not labelled as such (notably incentives and prompts—both behavioural strategies) and that several implicit BCTs were applied in both intervention and control strategies. The need to explicitly incorporate BCTs during the design of interventions to target recruitment and retention behaviours (and others relevant for trial conduct) is key. A small number of studies have developed behaviour change interventions for trial retention by incorporating BCTs into covering letters of questionnaires, newsletters, and also use of trial stickers on envelopes (to act as prompts) [36]. Preliminary evaluations of these behaviourally focussed trial process interventions are showing promise, but replication and further research to include patient input and assessment are required to maximise their potential [33, 36]. Creating a shift in the conceptualisation of recruitment and retention interventions to be considered (during design and delivery) as behaviour change interventions may provide more potential for more focused assessment of effectiveness and may enhance replicability.
It is important to highlight that the examples provided here are not an exhaustive list but are exemplars from key trial life cycle stages that serve to show where existing empirical studies have demonstrated the potential for a behavioural approach to address trial process problems (see Fig. 1). In particular, the challenges that many trials have faced during the COVID-19 pandemic (such as the move to remote delivery of recruitment, interventions, and follow-up) also provide a wealth of opportunities to apply behavioural approaches to generate evidence-informed solutions from the perspective of trial teams, regulators, and trial participants. A varied range of other trial process problems could also benefit from this approach including (but not limited to) choosing outcomes, participants’ experience, and sharing of trial results with trial participants. In addition, several behavioural approaches such as multiphase optimisation strategies (MOST), intervention mapping, and ‘nudging’ (a recent focus in the behavioural economics literature) could also warrant investigation in the future [37,38,39].