Population Surveys of STIs: Bias due to Missing Biological Specimens

Objective: The marriage of NAAT assays to population survey methodology permits monitoring of undiagnosed STIs in populations. We explore subpopulation variation in provision of biological specimens in a new research paradigm.

Methods: The Monitoring STIs Survey Program (MSSP) tracks three STIs (trichomonas, chlamydial and gonococcal infection) in probability samples of adolescents and young adults in Baltimore, MD. Participants complete telephone audio computer-assisted survey interviews (T-ACASI) and provide biological specimens through the mail. We report results for two years of the study (Y1-2, 2006 - 2008).

Results: 1,956 of 2,281 survey respondents (86% unweighted) agreed to provide biological specimens. 1,637 (84% unweighted) of “agreers” mailed biological specimens adequate for testing. Overall, 71% (weighted; 95% CI: 68.6%, 73.2%) of sampled population provided biological specimens. The specimen return rate was significantly higher in Y2 (75% vs. 67% wtd.; p = 0.002) due, we suspect, to changes in survey procedures. Relatively few statistically significant or borderline differences were found in specimen return rates across Baltimore subpopulations (e.g., by education: 78% for < HS, 70% for HS, 67% for >HS, p =0.0003; currently married vs. not: 67% vs. 72%, p = 0.093; had 6+ lifetime sex partners vs. not: 74% vs. 68%, p = 0.019).

Conclusion. MSPP's specimen return rate of 71% suggests that T-ACASI population surveys with mail-back collection of biological specimens provides a feasible methodology for monitoring population trends in undiagnosed STIs. Statistical models for assessing bias introduced by the small number of observed subpopulation variations in specimen non-response rates are evaluated in the presentation.