A Comparison of Participant Demographics Across Co-Designed Recruitment Methods to Two Student Mental Health Trials:...

Using social media platforms has been demonstrated to be a successful recruitment method, especially for young people. Two cited benefits of using social media for recruitment are its ability to quickly increase sample size and engage hard-to-reach participants.

This study aimed to (1) provide a pragmatic depiction of co-designing and implementing a social media strategy with our advisory group and (2) compare demographic information of participants recruited via social media and other methods. Our objective was to provide evidence for future trials to implement social media recruitment with maximum efficiency.

Participants were 2369 UK university students who consented to take part during the recruitment timeframe of 2 mental health trials. Our student advisory group advised on content, platform, and timing of engagement. We trialed 10 different adverts over a 12-month recruitment period. Descriptive analysis evaluated advert reach and link clicks using Meta/TikTok business data, website traffic using Google Analytics, screening consent, and enrollment using REDCap (Research Electronic Data Capture; Vanderbilt University) software. A cross-sectional observational analysis used chi-square and t tests to compare ethnicity, gender, sexual orientation, disability status, and university attended among 842 participants recruited via social media and 1527 participants recruited by other methods.

Through extensive student advisor input, an Instagram carousel advert led to a boost in participant recruitment. However, this fluctuated over the academic year, with numbers dropping completely over the summer months. All tests used α=.05. There was a difference in gender among those recruited through social media versus other recruitment methods on campus (χ²2=8.34, P=.02), with social media resulting in a higher proportion of gender-diverse students (27/370, 7% vs 30/711, 4%; 95% CI 3.4%-13.7%), but fewer male students (35/370, 4%, 95% CI [3.4%-13.7%] vs 99/711, 7%, 95% CI [1.6%-9.8%]). Those recruited from social media were younger than those recruited through other methods, with a mean difference of -3.49 years (SE=0.31, 95% CI [-3.94 to -3.04]; t1927.5=15.146, P<.001). A significantly higher proportion of students in the social media sample were from the LGBTQIA+ community (180/351, 51%, 95% CI [41.3%-60.6%] vs 350/711, 37%, 95% CI [28.2%-46.8%]; χ²1=17.87, P<.001). There was also a significant difference in the reported disability (103/375, 27.5%, 95% CI [19.7%-37.0%] vs 154/723, 21.3%, 95% CI [14.4%-30.3%]; χ²1=4.90, P=.03). There was no difference in ethnicity between the 2 groups (χ²1=2.4609, P=.12).

Our study describes how different recruitment approaches influence participant characteristics in clinical trials and highlights challenges in implementing a co-designed recruitment strategy in a university setting. This contributes to the field by providing research evidence on the efficacy of different recruitment strategies for planning future trials. Our key real-world recommendation is to allow time and resources for planning multiple recruitment strategies to ensure a diverse range of participants take part in research.