Background Self-reported survey data on sexual behaviours have limitations which biological measures can complement. We assessed the feasibility of using qualitative interviews to understand discordance between a biological marker of semen exposure and self-reported sexual behaviours among women enrolled in a trial of STI counselling messages. We also investigated reasons women did not follow counselling messages.

Methods Data are from the ACME (Assessing Counselling Message Effectiveness) study. Women age >18 years treated for infection at a Jamaica STI clinic were randomly assigned to receive counselling messages to: (1) abstain from sex or (2) abstain but use a condom if they did have sex, during a 6-day treatment period. At follow-up, women reported sexual behaviours in the last 3 days in a quantitative survey and consented to testing for recent semen exposure using a rapid, on-site test for prostate-specific antigen (PSA). Individual, qualitative interviews were subsequently conducted with a purposive sample of 17 PSA positive women, ten who did not report unprotected sex in the survey (discordant) and seven who did (concordant). Interviewers asked about recent sexual behaviour, sexual partners, counselling messages, and how they may have been exposed to semen. We analysed interview transcripts to assess whether discordance could be explained and to identify reasons counselling messages were not followed.

Results From 8/10 to date, 262 participants completed the ACME trial. Of the 10 qualitative interview participants whose PSA and survey results were discordant, five had reported condom use and five reported abstinence. Interviewers asked these women directly about the inconsistency. This approach variably elicited little concern, discomfort, plausible and implausible explanations from participants. Three condom users and one abstinent participant gave likely reasons for semen exposure, while positive PSA results for two condom users and four abstinent participants remained unexplained. Main reasons given by participants with concordant results for having unprotected sex despite counselling messages were: wanting to please partner, delaying the start of treatment, and not being regular condom users with their partner.

Conclusions Interviewing women about inconsistencies between their self-reported and biologic data are feasible in this context and yielded information for improving future self-reported surveys and understanding the primary study outcomes.