Is knowledge of HIV status associated with sexual behaviours? A fixed effects analysis of a female sex worker cohort in urban Uganda

Abstract

Introduction: Female sex workers (FSWs) have strong economic incentives for sexual risk-taking behaviour. We test whether knowledge of HIV status affects such behaviours among FSWs.

Methods: We used longitudinal data from a FSW cohort in urban Uganda, which was formed as part of an HIV self-testing trial with four months of follow-up. Participants reported perceived knowledge of HIV status, number of clients per average working night, and consistent condom use with clients at baseline, one month, and four months. We measured the association between knowledge of HIV status and FSWs' sexual behaviours using linear panel regressions with individual fixed effects, controlling for study round and calendar time.

Results: Most of the 960 participants tested for HIV during the observation period (95%) and experienced a change in knowledge of HIV status (71%). Knowledge of HIV status did not affect participants' number of clients but did affect their consistent condom use. After controlling for individual fixed effects, study round and calendar month, knowledge of HIV-negative status was associated with a significant increase in consistent condom use by 9.5 percentage points (95% CI 5.2 to 13.5, p < 0.001), while knowledge of HIV-positive status was not associated with a significant change in consistent condom use (2.5 percentage points, 95% CI -8.0 to 3.1, p = 0.38).

Conclusions: In urban Uganda, FSWs engaged in safer sex with clients when they perceived that they themselves were not living with HIV. Even in communities with very high HIV prevalence, the majority of the population will test HIV-negative. Our results thus imply that expansion of HIV testing programmes may serve as a behavioural HIV prevention measure among FSWs.

Keywords: Knowledge of HIV status; Uganda; condom use; key and vulnerable populations; sex workers; sexual behaviours; testing; women.

Figure 1. Participants’ knowledge of HIV status at baseline, one month, and four months

HIV‐negative status knowledge (black stripes); HIV‐positive status knowledge (black); HIV status knowledge unknown (grey). The light grey lines between the bars show the flows of participants across the different categories of knowledge of HIV status between the three study rounds.

Figure 2. The association between FSWs’ knowledge of HIV status and sexual behaviours with clients

The associations between knowledge of HIV status and sexual behaviours (number of clients per average working night and consistent condom use) were measured using linear panel regressions with individual fixed effects, controlling for study round and calendar month. Standard errors are adjusted for clustering at the level of the peer educator. Number of clients per average working night: the bars show the mean differences in the number of clients between (i) those with knowledge of HIV‐negative status and those with knowledge of unknown HIV status (black striped bars) and (ii) those with knowledge of HIV‐positive status and those with knowledge of unknown HIV status (black bars). Consistent condom use: the bars show the average percentage point differences in the probability of consistent condom use between (i) those with knowledge of HIV‐negative status and those with knowledge of unknown HIV status (black striped bars) and (ii) those with knowledge of HIV‐positive status and those with knowledge of unknown HIV status (black bars). The vertical lines indicate the 95% confidence intervals.

Figure 3. The association between changes in FSWs’ knowledge of HIV status and number of clients from baseline knowledge and sexual behaviours

For these sub‐group analyses, participants were sub‐divided by their knowledge of HIV status at baseline and their sexual behaviours at baseline (i.e. low risk vs. high risk). The reference for each sub‐group is participants’ knowledge of HIV status at baseline. The associations between participants’ changing knowledge of HIV and number of clients on an average working night were measured using linear panel regressions with individual fixed effects, controlling for study round (baseline, one month, and four months) and calendar month. Standard errors are adjusted for clustering at the level of the peer educator. The bars show the mean differences in the number of clients for participants whose knowledge of HIV status changed from different states at baseline (listed by sub‐group along the x‐axis) to HIV‐negative (black striped bars), HIV‐positive (black bars), or unknown (grey bars). The vertical lines indicate the 95% confidence intervals.

Figure 4. The association between changes in FSWs’ knowledge of HIV status and condom use with clients from baseline knowledge and sexual behaviour

For these sub‐group analyses, participants were sub‐divided by their knowledge of HIV status at baseline and their sexual behaviours at baseline (i.e. low risk vs. high risk). The reference for each sub‐group is participants’ knowledge of HIV status at baseline. The associations between participants’ knowledge of HIV status and condom use with clients were measured using linear panel regressions with individual fixed effects, controlling for study round (baseline, one month, and four months) and calendar month. Standard errors are adjusted for clustering at the level of the peer educator. Consistent condom use was defined as not using a condom with at least one client on an average working night. The bars show the average percentage point differences in the probability of consistent condom use for participants whose knowledge of HIV status changed from different states at baseline (listed by sub‐group along the x‐axis) to HIV‐negative (black striped bars), HIV‐positive (black bars), or unknown (grey bars). The vertical lines indicate the 95% confidence intervals.