Recruiting a special sample with sparse resources: Lessons from a study of Gulf War veterans

Abstract

Background Recruitment is the most common failure point for clinical studies, with recruitment failure adversely affecting science, dollar costs, human capital, and the ethical risk–benefit trade-off to study participants. Added problems attend recruitment of special and/or challenging candidate populations, particularly in settings of sparse recruitment resources. Obstacles to study recruitment and participation of ill Gulf War veterans (GWVs) include health barriers, work and family obligations, mistrust of the medical/scientific community, and challenges to identifying/reaching potential participants. Purpose We sought to identify and implement a minimal-cost multipronged recruitment approach for a small single-site (<50 participants) study of a special group, ill GWVs, with approaches substantially applicable to other recruitment settings and larger multisite studies. Methods Categories of recruitment approach included directed as well as general media, collaborations with support groups/interest groups, local free advertising resources (Craigslist and Backpage), physician outreach, Internet-based approaches, and referrals from study participants and screenees. We describe the subcategories and yield of each approach within each approach. Results Each approach contributed candidates to the final recruitment tally, with the largest fractional contribution by directed media (52%). Among the remainder, no other individual approach was clearly dominant (largest contribution: 13%). Limitations Special population subsamples present special challenges; all approaches cited may not be useful in all settings and subpopulations. Conclusions A multipronged suite of minimal-cost approaches led to successful recruitment to target for this single-site clinical trial for a special population with significant recruitment challenges. It additionally yielded a nation-wide corpus of several hundred individuals interested in participation in future studies of GWVs. While certain approaches produced disproportionate yield, it was not possible to predict these a priori. We suggest that this model, which incorporates a suite of approaches, and delineates backup approaches in the event of recruitment shortfall, may provide a template applicable to recruitment of other special samples in settings of limited resources and also is germane to cost-effective recruitment in studies more generally.