Subscribe to E-Bulletin Donate to CEGA

Biomarker Validation of Self-Reported Sexual Behavior among Women in Zimbabwe

Development Challenge

Inaccurate self-reporting of sexual behavior hinders HIV research and the design of effective interventions.  Several factors reduce the accuracy of self-reported sexual histories during interviews, including stigma associated with admitting non-adherence to HIV prevention strategies, poor recall of recent sexual activity, poorly worded questionnaires, and complex categorization of sexual behaviors.  To help increase reporting accuracy, researchers have increasingly replaced face-to-face interviewing (FTFI) with a new technology called audio computer-assisted self-interviewing (ACASI), which prior studies suggest shows promise to improve accuracy of self-reported sexual histories.[1] , [2] While FTFI requires a surveyor to read the participant sensitive survey questions and record their responses, in contrast, ACASI allows respondents to answer a questionnaire in privacy, can be designed for low-literacy populations, and can accommodate complex skip patterns.  This randomized evaluation assessed differences in the accuracy of self-reported sexual histories between FTFI and ACASI, using an objective biomarker of recent sexual activity as a validation of actual sexual history.

Context

This study took place in 2007 among a subset of Zimbabwean women who had recently participated in the Methods for Improving Reproductive Health in Africa (MIRA) trial, a randomized evaluation that tested methods to reduce susceptibility to contracting HIV.  Researchers recruited 910 women aged 18-49 years who were, at the time of the survey, HIV-negative, sexually active, and not pregnant, and who had not recently given birth. 

Evaluation Strategy

Researchers randomly assigned participants to the ACASI or FTFI interview group and then asked them a series of questions about their sexual activity and condom use over the past seven days.  Participants then provided a self-collected vaginal swab, which was tested for prostate-specific antigen (PSA), an enzyme that predicts recent sexual activity without a condom.  PSA is only detectable in an estimated 29 percent of women 24 hours after known exposure, and is rarely present beyond 48 hours after exposure.[3] 

Researchers hypothesized that the modal characteristics of ACASI would lead to higher reported levels of unprotected sex than FTFI, and would therefore lead to a lower level of discrepancy between self-reporting and PSA test results compared with FTFI.  First, to measure differences in self-reporting by interview mode, researchers compared the percentage of participants in each interview group who reported sexual activity only in the past two days.  Second, researchers restricted the sample to participants who tested PSA-positive in order to evaluate the proportion of discrepant results by interview mode.  A discrepant result refers to a case when a participant tested positive for PSA but reported having no unprotected sex in the two days prior to survey. 

Results and Policy Implications

The proportion of women who reported having no vaginal sex in the two days prior to the survey did not differ by interview mode.  However, as hypothesized, participants in the ACASI group did report a statistically significant lower level of condom-protected sex only than those in the FTFI group (32.7 percent vs. 40.0 percent, respectively).  These results suggest that ACASI may have a modest positive effect on the accuracy of self-reporting of sexual behavior. 

Among the 21.5 percent of women who tested positive for PSA, there was no statistical difference in self-reporting of sexual activity by interview mode.  The chart below notes the differences by reporting methods for those who reported no vaginal sex or condom-protected sex only, yet still tested positive for PSA. These results indicate a high level of discrepant reporting: nearly half of participants with biologic evidence of recent semen exposure reported that they had not had vaginal sex (12 percent) or that they had had condom-protected sex only (36 percent) in the two days prior to the survey. 

 

FTFI

ACASI

PSA positive & reported having no vaginal sex

10.9%

12.5%

PSA positive & reported condom-protected sex only

39.1%

33.7%

Overall, the study results indicate that ACASI yielded a lower reported level of consistent condom use during recent sexual activity than FTFI, but did not contribute to improved concordance between self-reported sexual behavior and PSA test results among those who tested positive for PSA. 

The researchers note the limitation of using PSA tests to analyze discrepancies in self-reported sexual behavior because the test does not evaluate condom usage sufficiently to assess user error or mechanical failure.  The authors conclude that additional research that incorporates in-depth qualitative interviews with PSA-discrepant participants would help identify the causes of the discrepancies, may help improve questionnaire wording, and could inform improvements to future ACASI usage.  

Timeline

2006 – 2007

[1] Mensch, B., Hewett, P., Erulkar, A. (2003). The reporting of sensitive behavior by adolescents: a methodological experiment in Kenya. Demography, Vol. 40(2), 247–268.

[2] Langhaug, L., Cheung, Y., Pascoe, S., Chirawu, P., Woelk, G., Hayes, R., Cowan, F. Comparing four questionnaire delivery methods for collection of self-reported sexual behavior data in rural Zimbabwean youth. (Abstract 0-038), Presented at the 17th Meeting of the International Society for Sexually Transmitted Diseases Research, Seattle, Washington, July 29–August 1, 2007.

[3] Macaluso, M., Lawson, L., Akers, R., Valappil, T., Hammond, K., Blackwell, R., Hortin, G. (1999). Prostate-specific antigen in vaginal fluid as a biologic marker of condom failure. Contraception, Vol. 59(3), 195–201.

Photo credit: PSI Zimbabwe