Mitigating circumstances: A model-based analysis of associations between risk environment and infrequent condom use among Chinese street-based sex workers

Background: Little is known about how freelance street-based sex workers navigate condom use while soliciting. Traditional behavioural model may fail to account for the complex risk environment that most street-based sex workers work within. We examine first the association of self-efficacy and the infrequent condom use, then we investigated the roles of clients and venues frequented on this association. Method: Using a purposive chain-referral sampling method, we surveyed 248 street-based sex workers in Shanghai. The survey focused on sex workers HIV risk factors, sex work patterns, HIV knowledge, and related HIV self-efficacy. Clients types and behaviours, and characteristics of the venues frequented by these commercial sex workers were also collected. We conducted a series of multiple logistic regression models to explore how the association between a sex worker’s self-efficacy with infrequent condom use change as client and venue characteristics were added to the models. Results: We find that within the basic model, low self-efficacy was marginally associated with infrequent condom use (54.9% vs. 45.1%, AOR = 1.70, 95% CI = 0.95–3.03). As client- and venue- characteristics were added, the associations between self-efficacy and condom use were strengthened (AOR = 2.10 95% CI = 1.12–3.91 and 2.54 95% CI = 1.24–5.19 respectively). Those who reported middle-tiered income were more likely to report infrequent condom use compared to their peers of high income (AOR = 3.92 95% CI = 1.32–11.70) whereas such difference was not found between low income and high income sex workers. Visiting multiple venues and having migrant workers as clients were also associated with infrequent condom use. Conclusion: Our findings suggest sex worker’s self-efficacy matters in their HIV risk behaviours only when environment characteristics were adjusted. Risk environment for street-based sex workers are complex. Programming addressing behavioural changes among female sex workers should adopt holistic, multilevel models with the consideration of risk environments.