Recruiting hard-to-reach pregnant women at high psychosocial risk: strategies and costs from a randomised controlled trial

Background Recruiting participants to randomised controlled trials (RCTs) is often challenging, particularly when working with socially disadvantaged populations who are often termed 'hard-to-reach' in research. Here we report the recruitment strategies and costs for the Trial for Healthy Relationship Initiatives in the Very Early years (THRIVE), an RCT evaluating two group-based parenting interventions for pregnant women. Methods THRIVE aimed to recruit 500 pregnant women with additional health and social care needs in Scotland between 2014 and 2018. Three recruitment strategies were employed: (1) referrals from a health or social care practitioner or voluntary/community organisation (practitioner-led referral), (2) direct engagement with potential participants by research staff (researcher-led recruitment) and (3) self-referral in response to study advertising (self-referral). The number of referrals and recruited participants from each strategy is reported along with the overall cost of recruitment. The impact of recruitment activities and the changes in maternity policy/context on recruitment throughout the study are examined. Results THRIVE received 973 referrals: 684 (70%) from practitioners (mainly specialist and general midwives), 273 (28%) from research nurses and 16 (2%) self-referrals. The time spent in antenatal clinics by research nurses each month was positively correlated with the number of referrals received (r = 0.57; p < 0.001). Changes in maternity policies and contexts were reflected in the number of referrals received each month, with both positive and negative impacts throughout the trial. Overall, 50% of referred women were recruited to the trial. Women referred via self-referral, THRIVE research nurses and specialist midwives were most likely to go on to be recruited (81%, 58% and 57%, respectively). Key contributors to recruitment included engaging key groups of referrers, establishing a large flexible workforce to enable recruitment activities to adapt to changes in context throughout the study and identifying the most appropriate setting to engage with potential participants. The overall cost of recruitment was 377 pound per randomised participant. Conclusions Recruitment resulted from a combination of all three strategies. Our reflections on the successes and challenges of these strategies highlight the need for recruitment strategies to be flexible to adapt to complex interventions and real-world challenges. These findings will inform future research in similar hard-to-reach populations.