Using Drug Use Testing to Validate Self-Reported Sexual Behavior in a Sample of MSMs in Chicago

The validation of self-reported substance use in epidemiologic surveys using biological assays is well-documented. The accuracy of other survey self-reports, however, are more difficult due to the absence of gold standard auxiliary information. Sexual practices are a case in point, as it is impractical to confirm survey reports of these behaviors. We present analyses that indirectly evaluate the quality of self-reported sexual behavior by examining the reports provided by survey respondents who do and do not provide accurate reports of drug use behaviors. Data come from a random community sample of n=172 men who have sex with men (MSM) conducted in Chicago in 2003. Findings indicate that those respondents documented as under-reporting their drug use were found to also report fewer sexual experiences. Specifically, MSMs under-reporting current marijuana or cocaine use were in addition less likely to indicate sexual experiences with men during the past six months, compared to those with validated use of these substances (61.5% vs. 84.3%, respectively, chi-square = 4.3, p < 0.05). Logistic regression models suggest that these differences are mediated by respondent age and income. We will consider the utility of these findings for collecting more valid information regarding sensitive behaviors in epidemiologic research.