Perceived Social Support, Contact Violence Experience, and Sexually Transmitted Infection Risk among Gambian Female Sex Workers: A LASSO-Guided, Causal Inference, and Multilevel Analysis

Abstract

Female sex workers (FSWs) face disproportionate risks of sexually transmitted infections (STIs) including HIV/AIDS and various forms of violence. Although social support is generally considered protective against adverse health outcomes, its relationship with STI risk in high-risk population remains unclear. This study was conducted to determine and quantify key determinants of lifetime STI occurrence among FSWs in The Gambia, with a particular focus on violence experience and social support. A cross-sectional design was used, with secondary analysis of data from 429 FSWs recruited via snowball and venue-based non-probability sampling methods in urban and peri-urban areas of The Gambia. Lifetime STI occurrence, lifetime contact violence experience (physical or sexual), and perceived social support (using the Multidimensional Scale of Perceived Social Support) were assessed through structured interviews. Multiple analytical approaches, including LASSO-guided variable selection, conventional logistic regression, marginal structural models using inverse probability weighting, targeted maximum likelihood estimation (TMLE), and multilevel modeling, were employed to address potential confounding and clustering effects. The prevalence of lifetime STI occurence was 71.8%, and 69.2% of the participants reported ever experienced contact violence. LASSO penalized regression identified eight key predictors of STI occurence, with contact violence experience and perceived social support consistently emerging as significant factors. Contact violence experience was strongly associated with STI occurrence across all analytical approaches: conventional logistic regression (aOR = 7.55, 95% CI [3.05–19.43]), inverse probability-weighted models (aOR = 1.99, 95% CI [1.26–3.14]), and TMLE (risk difference = 0.206, 95% CI [0.129–0.283]). Contrary to our hypothesis, higher perceived social support was associated with increased STI risk (aOR = 1.57, 95% CI [1.14–2.19]). Multilevel analysis revealed significant geographical clustering, with 12.0% of variance in STI experience attributable to residential area. Sensitivity analyses using E-values indicated that findings were robust to potential unmeasured confounding, particularly for violence experience (E-value = 21.27, CI bound = 9.27). This study reveals a counterintuitive positive relationship between social support and STI risk among Gambian FSWs, potentially reflecting limitations of Western-derived measurement instruments in capturing culturally specific support structures among stigmatized populations. The robust association between contact violence experience and STI occurrence highlights the urgent need for integrated interventions that address violence prevention alongside STI screening and treatment within sexual health services for FSWs. Interventions leveraging peer networks should carefully assess network norms to ensure they reinforce protective rather than risk-enhancing behaviors.