Breaking through the barriers to recruit research participants in community settings: a qualitative exploration using the Theoretical Domains Framework and Behaviour Change Wheel


 Background Recruitment of pregnant or postnatal women and young families into health research is a challenge. Community midwives and health visitors are well placed to invite service users to participate, but evidence suggests that they do not always invite all potentially eligible service users. Our aim was to use the Theoretical Domains Framework to explore health visitors’ and community midwives’ perceived barriers and enablers to approaching service users about participation in research and to use the Behaviour Change Wheel to suggest theory-based strategies to improve future recruitment. Methods Health visitors and community midwives working in four NHS Trusts and one community partnership in England were invited to complete an anonymous, online survey. The sample comprised health visitors (n=39) and community midwives (n=22). Qualitative data from open-ended questions about recruitment behaviour informed by the Theoretical Domains Framework were analysed using directed content analysis and inductive coding to identify salient Theoretical Domains Framework domains and specific barriers and enablers. Strategies to address these barriers and enablers were identified using the Behaviour Change Wheel. Results Six key Theoretical Domains Framework domains were identified as salient to service user approach: (a) environmental context and resources; (b) social/professional role and identity; (c) social influences; (d) goals; (e) beliefs about capabilities; (f) knowledge. Intervention strategies were identified to address specific barriers (insufficient time and staff capacity, inadequate study materials, rejection of the study’s relevance to practitioners’ role, negative influence of researchers and managers, and competing priorities) and to leverage specific enablers (additional staff resource, the relevance of service user approach to professional role, positive influence of team members, managers and researchers, and a belief in the link between service user approach and improvements in healthcare). Conclusions This study provides new insights into why community midwives and health visitors do, and do not, invite service users to participate in research. Using the Theoretical Domains Framework and the Behaviour Change Wheel we identified that resourcing and social support for staff together with education and communication about the health benefits of approaching service users about research participation are key to improving research recruitment involving these health professionals.


in healthcare).
Conclusions This study provides new insights into why community midwives and health visitors do, and do not, invite service users to participate in research. Using the Theoretical Domains Framework and the Behaviour Change Wheel we identified that resourcing and social support for staff together with education and communication about the health benefits of approaching service users about research participation are key to improving research recruitment involving these health professionals.
Background It is estimated that fewer than half of all clinical research studies achieve their participant recruitment targets [1]. This has economic, ethical and clinical consequences [2] and undermines potential benefits to population health. Recruitment to research involves three stages: (i) identification of eligible service users (ii) approaching them about the study (iii) seeking their agreement to participate and gaining their consent [2]. There is potential for loss of participants at any of these stages, but the stage where most potential participants are lost [3,4] and about which least is known [5] is the approach, in which eligible people are told about the study and invited to take part. The NHS Constitution for England [6] pledges to inform patients of research studies in which they may be eligible to participate. Data protection legislation prevents researchers approaching potential participants directly [2,7,8] and therefore it is common practice for healthcare professionals to approach service users, delivering research invitations, in the course of their clinical care [2]. However, there is substantial evidence that when performing this function, healthcare professionals approach only a proportion of eligible service users [2,[9][10][11][12]. This adversely affects recruitment rates, introduces biases to the sample, and deemed too vulnerable for research [2,9].
There is a growing body of literature addressing why healthcare professionals do not 4 invite all eligible service users to take part in research. Reasons identified include: a protective bias and overprotection of some groups of service users [2,9,[11][12][13][14]; unclear or complex eligibility criteria and insufficient study information [15]; poor understanding of equipoise, randomisation, allocation, eligibility criteria, informed consent [16]; lack of time or resources to approach service users [15][16][17]; misgivings about the study methods or rationale [10]; role conflict [4]. Attempts to intervene to address these barriers have, however, had little success [1,16].To date, most interventions to improve recruitment rates have been developed as ad-hoc, pragmatic responses to poor recruitment in a particular trial [28]. Consequently, there have been calls for greater understanding of barriers and facilitators to healthcare professionals inviting service users to research [16].
French and Stavropoulou [15] made an important first step towards that goal in their investigation of specialist nurses' perceptions of inviting patients to research as a general concept rather than in relation to a specific study [15]. However, as the authors note, use of an established theoretical framework, such as the Theoretical Domains Framework (TDF) [18][19][20][21] to guide data collection and analysis would have enhanced and strengthened their study.
The TDF is an evidence-based tool that provides a systematic approach to understanding healthcare professionals' behaviours and identifying what needs to change [18][19][20][21].
Building on this behavioural analysis, it is then possible to use the Behaviour Change Wheel (BCW) to identify the appropriate change mechanisms. Changing behaviour is acknowledged to be challenging, but it is more likely to be successful where interventions are rooted in evidence-based models of behaviour change [22,23] -an approach that is yet to be used in the development of strategies to help healthcare professionals recruit service users to research [2,24].
In this study we address that gap, focussing on community midwives and health visitors, professionals who provide community healthcare for women and their children in the UK, from pregnancy to five years of age. Delivering universal services, these practitioners have very high potential reach [25] and are well placed to approach pregnant women, new parents and families about participation in research. However, where health visitors and community midwives have been involved in participant recruitment, disappointing recruitment and limited representativeness of the study sample has been a concern [26][27][28][29]. There is very little previous research exploring the experiences of these community health professionals in research, with only one study looking at barriers to community midwives identifying potential participants in a specific randomised controlled trial (RCT) [35] and none that we have found looking at the experiences of health visitors. There are also no interventions designed to improve the potential participant approach rates by these community healthcare professionals.
The aim of this study was to explore health visitors' and community midwives' perceived barriers and enablers to inviting service users about participation in research in order to suggest theory-based strategies to improve future recruitment. Following the BCW approach, the study's objectives were: i.
To use the TDF [18][19][20][21] to conduct a behavioural analysis of what needs to change to enable health visitors and community midwives to approach service users about research participation; ii.
To use the BCW [19,20] to link the results of this behavioural analysis to identify intervention elements and policy changes to improve service user approach.

Design
We used a self-reported, anonymous, online, cross-sectional survey to collect data from the health visitor and community midwife participants about their professional and 6 demographic characteristics, self-reported involvement in study recruitment and the factors that influence their approaches to potential participants about a research study.
The questions about recruitment behaviour were informed by the TDF: a validated, systematic approach to assessing the barriers and enablers to desired behaviours. Based on psychological theory, the TDF synthesises the key theoretical constructs from 33 behaviour change theories into 14 domains [21]. Data collection for TDF analysis can be done using interviews, focus groups or surveys [19,30].

Sampling
Consistent with the exploratory nature of this qualitative study, we used purposive sampling to identify a maximum variation sample of health visitors and community midwives working in different community settings with experience of inviting service users to participate. Participating organisations could not identify which staff had experience of the service user approach, therefore they sent an email to all of their health visitors and community midwives inviting them to take part in the study, and we used a screening 7 question to select staff who had experience of the service user approach.

Participants and setting
We invited staff delivering community public health nursing (health visiting) and community midwifery services for four NHS Trusts and one social enterprise to complete the questionnaire. These organisations covered both rural and urban areas, in different regions of England.

Research team and reflexivity
At the time of data collection, the researchers included two registered nurses (SR and FM), one of whom is also a health visitor (SR), a social worker (JA) and a research fellow (JR).
JR, JA and SR were all employed as researchers at Anglia Ruskin University; FM was employed by NIHR as Research Delivery Manager and Continuous Improvement Lead for CRN Eastern. All researchers have a PhD and experience working in non-academic roles with families, in clinical or community settings. We also all have experience of working with health and social care professionals to recruit participants to research projects, and these experiences prompted our interest in this study. Our application of an approach rooted in psychological theory to frame the study is influenced by our training as psychologists (JR and SR).

Data collection
A hyperlink in the email opened to the participant information sheet. Potential participants were informed that the survey was anonymous, no personally identifiable information would be captured and once submitted their survey answers could not be withdrawn.
Recipients were asked to confirm their consent before starting the survey and again before submitting their completed surveys. The survey was open for four weeks, and a reminder sent after two weeks. The few differences in coding were discussed, and a consensus opinion reached. Specific barriers and enablers to service user recruitment were then identified within each domain.

Analysis
Salient TDF domains, identified as the domains which best explained the variation in service user approach behaviour, were mapped to intervention functions using the BCW, and components for an intervention to support the desired behaviour were selected by applying the APEASE criteria (affordability, practicability, effectiveness, acceptability, safety and equity) [19].

Results
Sample characteristics 347 community midwives and health visitors were informed of the survey by their employer and 114 completed the survey. Of these, sixty one respondents (54%) had experience of approaching service users about participation in research; this figure was higher among community midwives (N=22, 76% of all community midwife respondents) than among health visitors (N=39, 51% of all health visitor respondents). We did not calculate a response rate for health visitors and community midwives with experience of service user approach, as the employing organisations did not have data on the numbers of staff with that experience. Table 1 presents the demographic and other characteristics of the health visitor and community midwife respondents with experience of service user approach. 1 Participants could choose more than one option, so % does not add up to 100.
Behavioural analysis of factors influencing the desired behaviours using the TDF Across the dataset, 408 responses were mapped to 14 TDF domains and forty five specific barriers/enablers. Table 2   Another barrier mentioned by health visitors and community midwives alike was language, encountered when attempting to inform service users who spoke little or no English about potential research opportunities.
It is a difficulty when English isn't their first language (HV87) There was also evidence for the social influence of researchers on the respondents' service user approach behaviour. Researchers who fail to engage with and support the health professionals appear to be a particular barrier to the involvement of health visitors, whereas community midwives identified communicative and supportive researchers as an enabler. These practitioners felt that it was the researchers' responsibility not only to provide the information and physical resources for recruitment, but also to provide support and encouragement in person.
We need more involvement from the researchers rather than just handing it to us! (HV27) Researchers being visible and approachable, using easy to-understand language and making it relevant to our clinical area, and help in the recruitment process is important (CM102) A desire to contribute to the team was an enabler for both health visitors and community midwives. However, only health visitors mentioned that the influence of managers could act as both an enabler and a barrier to service user approach.

I do this as it supports my colleagues (CM106)
We share enthusiasm about research and how it impacts on all of us, practitioner and patient alike (HV50) Some managers encourage participation whereas others are mindful of time restraints (HV91)

(d) Goals
Across both professional groups it was apparent that introducing research opportunities to service users was not a high priority. This posed a significant barrier to some, but not all respondents, and was the fourth most frequently identified TDF Domain for both professional groups (

(f) Knowledge
Knowledge could act as a barrier or enabler for both health visitors and community midwives. In particular both professional groups emphasised the need for good procedural knowledge about the study, and knowledge of the scientific rationale for the study. A few participants also mentioned needing knowledge of the research topic.
It must be with midwives who have the relevant knowledge (CM4).
Knowing that participation is voluntary and that participants can with-draw within defined boundaries gives me greater confidence in approaching service users.
Understanding the process of ethical approval and how research projects are planned, their protocols laid out etc. gives me greater confidence in explaining participation to others. I would also want to understand the research project objectives and the implication/commitment required from participants to feel confident that I could then 'sell' this to a patient (HV59).

Development of an intervention strategy
The intervention functions which map to the salient TDF Domains are shown in Table 3, together with example behaviour change techniques (BCTs) and policies for inclusion in a complex intervention to support health visitors and community midwives to approach eligible service users offering them the choice to take part in a research study. How these could combine into a complex intervention is depicted in Figure 1. Although the TDF and BCW are widely used in implementation science to understand the behaviour of healthcare professionals, to our knowledge this is the first attempt to use them to understand participant recruitment behaviour. Our study goes beyond previous studies looking at barriers and enablers to research recruitment to make theory-based recommendations for future interventions. It also addresses an important and underresearched aspect of health visitors' and community midwives' behaviour. There are a few investigations exploring the research recruitment behaviour of hospital midwives [32][33][34] but we have identified only one study focusing on community midwives' [35] barriers and one looking at why health visitors declined to take part in a cluster RCT [28].
Studies of barriers to research recruitment involving other healthcare professionals have identified many of the same barriers that were evident in our study. Time constraints, staff shortages and workload are widely reported barriers to research recruitment across health specialities and services in the UK, Finland and US [28,32,[36][37][38]. Others have also reported that inaccessible study materials [32,34,35], language barriers [14] and inadequate support from researchers [37] pose barriers to research recruitment. Against the backdrop of increased demand and limited resources for healthcare, these may further impede healthcare professionals' engagement with research. However, evidence from systematic reviews suggest that support and additional training alone are not effective in changing service user approach behaviour of healthcare professionals [2,39]. As respondents clearly communicated to us, healthcare professionals need time to be allocated to this activity, and that in turn demands funding for the staff resource it uses.
Whilst there is an established mechanism for the recovery of costs of research in the NHS [40], no additional funding is attributable for staff asked to approach potential participants during scheduled clinical encounters. Like others in the UK and the US, we call for salary support which releases time for research-related tasks for healthcare professionals asked to assist with research recruitment [36,38]. It is also important that any salary support funding is visible to the healthcare professionals involved in service user approach.
Our analysis suggests that leveraging an enabling belief in the link between successful research recruitment and improvements in health care could help to address the difficulty of navigating research responsibilities alongside a clinical role. This could be delivered through education and persuasion, supported by service provision and a communications campaign. Importantly, this provides an opportunity to address our finding that some health visitors and community midwives choose not to approach all eligible service users about research opportunities, a tendency which is apparently quite widespread among other healthcare professionals [10-12, 35, 41]. Authors of a thematic meta-synthesis [14] suggest that this behaviour may be underpinned by a widespread belief among healthcare professionals that service users should not decide for themselves whether to take part in research. While the intention may be to protect, a decision by professionals not to offer the opportunity to take part in research necessarily results in a loss of autonomy for that service user and a shift towards paternalism.
Healthcare professionals choosing not to approach some eligible service users can also cause exclusion of certain group perceived as being too vulnerable for research [10]. Such excluded groups have included children [10], low-income and minority women [42], potentially vulnerable pregnant women [35], patients with depression [12] and people who are socially disadvantaged [11]. This necessarily undermines the representativeness of the study sample, the generalisability of the findings and the scientific and social value of the study. To redress this it is imperative that healthcare professionals who are asked by researchers to invite service users into a study are fully aware of the negative consequences of choosing not to invite all eligible service users to take part in a study.
Strengths and limitations 20 The online survey was a pragmatic way of gathering qualitative data without requiring extensive time from these time-poor professionals. Others have found that, when compared to more in-person interviews, online surveys can collect richer and more detailed data and this method of qualitative data collection [43] and are especially useful with participants for whom travel to a common research site would difficult and for topics where anonymity will likely promote disclosure [44]. Both of these circumstances apply to our study population. The anonymity of the online survey may also have enabled us to collect responses from participants with a wide range of views on research, including negative, ambivalent and positive views, which some interview studies of this topic have not previously accessed [15]. Whilst interviews may have yielded richer data, we found that our respondents were generous in their free-text responses, which delivered ample material to enable us to apply the TDF to analyse the behaviour in question.
The lack of a respondent denominator meant we were unable to calculate an overall response level and the sample sizes, though adequate for a qualitative study using the TDF and BCW [30], are small for samples collected via an online survey. The survey respondents were self-selecting and consequently our findings may have been influenced by response bias, and this limits the generalisability of our findings. Nevertheless, the samples were diverse with respect to the range of environments where the respondents were practicing, their experience in their professional role, and, particularly for community midwives, ethnicity. Whilst the majority of respondents were female, this reflects the demographics of the NHS workforce in these specialisms.

Conclusions
This study uses a theory-informed approach to gain new insights into improving research recruitment where health visitors and community midwives invite service users to take part in a study. We found that resourcing and social support for staff, together with

Consent for publication
Participants consented to the publication of individual direct quotes (anonymised).

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests
None declared

Funding
This research was funded by a Seedcorn Grant from Anglia Ruskin University. The funder had no input into the design of the study and collection, analysis, and interpretation of data and in writing the manuscript.

Authors' contributions
JR wrote the protocol, obtained ethics and governance permissions, assembled the research team, liaised with participants' employers, project managed the study, conducted the data analysis with KL and wrote the paper. SR, JA and FM commented on the protocol, contributed to the questionnaire design and study design and commented on the paper. All authors read and approved the final manuscript.   Figure 1 Outline of a complex intervention to support health visitors and community midwives to approach service users about research participation.

Supplementary Files
This is a list of supplementary files associated with the primary manuscript. Click to download.