Behavioral Health, Mental Health, Substance Abuse
Did I tell you the truth? Biomarker validation of self-reported sex among middle-aged female sex workers in China"
(School of Public Health (UMD) EPIB Doctoral Student)
Guida, Jennifer (UMD), Fukunaga, Ami (UMD), Liu, Hongjie (UMD)
Background: A common practice in public health research is to rely on self-reported measures to collect data. However, sensitive questions, particularly those that are attached to socially desirable behaviors may lead to measurement error and result in invalid outcomes. The objective of this study was to examine information bias arising from self-reported engagement in sexual intercourse and its association with syphilitic infections among a vulnerable population of female sex workers (FSW) in China. Methods: A questionnaire was administered to 1245 middle-aged FSWs. Respondents were asked to self-report whether they had sexual intercourse in the past 24 and 48 hours. The prostate specific antigen test (PSA) is detectable in vaginal fluids up to 48 hours after unprotected sex. FSWs were considered discordant if they indicated no sexual intercourse in the past 48 hours on the survey and had a positive PSA test. Incident syphilitic infection was determined by sero-positivity to both the TRUST and TPPA tests. Prevalent syphilitic infection was measured by the TPPA test. Logistic regression was used to test the associations between discordance and syphilis. Results: 320 FSWs self-reported no engagement in sexual intercourse in the past 48 hours. Among those who indicated no sex on the questionnaire, approximately 75% of respondents (213/283) were discordant (e.g., had a positive PSA test). Six percent (13/203) and 35.8% (15/52) of discordant FSWs tested positive for incident and prevalent syphilis, respectively. After adjusting for confounders, discordant FSWs had 3.8 times the odds of incident syphilis and 2.6 times the odds of prevalent syphilis, compared to concordant FSWs. Conclusions: FSWs who had incident or prevalent syphilis were more likely to be discordant compared to those who were concordant. Self-reporting sensitive questions on surveys, such as sexual intercourse, may not be a valid tool to assess sexual behavior.