Recruiting to a Randomized Controlled Trial of a Web-Based Program for People With Type 2 Diabetes and Depression: Lessons Learned at the Intersection of e-Mental Health and Primary Care

BACKGROUND: E-mental health (eMH) interventions are now widely available and they have the potential to revolutionize the way that health care is delivered. As most health care is currently delivered by primary care, there is enormous potential for eMH interventions to support, or in some cases substitute, services currently delivered face to face in the community setting. However, randomized trials of eMH interventions have tended to recruit participants using online recruitment methods. Consequently, it is difficult to know whether participants who are recruited online differ from those who attend primary care.

OBJECTIVE: This paper aimed to document the experience of recruiting to an eMH trial through primary care and compare the characteristics of participants recruited through this and other recruitment methods.

METHODS: Recruitment to the SpringboarD randomized controlled trial was initially focused on general practices in 2 states of Australia. Over 15 months, we employed a comprehensive approach to engaging practice staff and supporting them to recruit patients, including face-to-face site visits, regular contact via telephone and trial newsletters, and development of a Web-based patient registration portal. Nevertheless, it became apparent that these efforts would not yield the required sample size, and we therefore supplemented recruitment through national online advertising and promoted the study through existing networks. Baseline characteristics of participants recruited to the trial through general practice, online, or other sources were compared using the analysis of variance and chi square tests.

RESULTS: Between November 2015 and October 2017, 780 people enrolled in SpringboarD, of whom 740 provided information on the recruitment source. Of these, only 24 were recruited through general practice, whereas 520 were recruited online and 196 through existing networks. Key barriers to general practice recruitment included perceived mismatch between trial design and diabetes population, prioritization of acute health issues, and disruptions posed by events at the practice and community level. Participants recruited through the 3 different approaches differed in age, gender, employment status, depressive symptoms, and diabetes distress, with online participants being distinguished from those recruited through general practice or other sources. However, most differences reached only a small effect size and are unlikely to be of clinical importance.

CONCLUSIONS: Time, labor, and cost-intensive efforts did not translate into successful recruitment through general practice in this instance, with barriers identified at several different levels. Online recruitment yielded more participants, who were broadly similar to those recruited via general practice.