Background: The extent to which participants in randomized clinical trials (RCTs) resemble their broader target groups is of particular concern when studying stigmatized conditions such as substance use disorders (SUDs). The present study compared patients who enrolled in a 4-arm clinical trial of SBIRT (screening, brief intervention, and referral to treatment) with eligible patients who declined study participation.
Methods: Participants (N = 1338) were primary care patients who anonymously completed a computer-delivered health survey and screened positive for heavy/problem alcohol or drug use. Those who consented to the RCT (n = 713) were compared with those who declined (n = 625) on a variety of demographic, substance use, and psychosocial characteristics. Variables significant at P < .20 in univariate analyses were then examined using multivariate logistic regression to determine their combined effect.
Results: The sample was 60% female and 76% African American, with a mean age of 45.2 years. Patients who consented to participate differed from those who declined on 34 (60%) of the 57 variables studied. The most parsimonious model by multivariate regression found that those who consented were older, more likely to be unemployed, and more likely to endorse prescription drug misuse, problems related to drug use, family history of alcohol problems, trouble falling asleep, and a health professional recommendation to lose weight.
Conclusion: Patients consenting to the RCT reported a greater number and more severe psychosocial and mental health problems than those who declined study participation. If the higher level of risk among study participants was found in other studies as well, it would raise questions regarding the generalizability of RCT results to broader clinic samples.
Keywords: Generalizability; SBIRT; representativeness; substance use.