Effect of different recruitment sources on the composition of a bipolar disorder case registry.

BACKGROUND: Conducting research on low-prevalence conditions presents an ongoing challenge for clinical and services researchers. Recruitment through health professionals versus other forms of self-referral may affect study group composition.
METHODS: We compared members of a voluntary case registry for bipolar disorder who were recruited through a variety of sources including health professionals, support groups, an Internet website, and mailings, brochures, and other general public relations activities. We also compared the cost of recruitment methods. We hypothesized that self-referred registry members would be of higher socioeconomic status and less likely to be in treatment compared to members recruited through health professionals.
RESULTS: Registrants referred through the Internet and patient support groups were better educated and more likely to be married than other registrants. However, Internet registrants were younger, had fewer lifetime hospitalizations and were more likely to be working. Nearly all registrants were in treatment with a psychiatrist. Local registrants were predominantly recruited through health professionals and public presentations. Registrants outside of the local region most often learned about the registry from patient support groups and the Internet. Local registrants were less likely to be using non-lithium mood stabilizers. Recruitment through public relations efforts was the most expensive method of recruitment, and the Internet website was the cheapest.
CONCLUSIONS: Diverse recruitment methods can expand the population available for clinical trials. For services research, the Internet and patient support groups are less expensive ways to identify persons served in diverse settings and health plans, but these recruitment methods yield a sample that is better educated than the remainder of the population. It remains a difficult task to identify minorities and persons not in treatment.