RESUMEN
INTRODUCTION: Female sex workers (FSWs) frequently experience violence in their work environments, violating their basic rights and increasing their vulnerability to HIV infection. Structural interventions addressing such violence are critical components of comprehensive HIV prevention programmes. We describe structural interventions developed to address violence against FSWs in the form of police arrest, in the context of the Bill and Melinda Gates Foundation's India AIDS Initiative (Avahan) in Karnataka, South India. We examine changes in FSW arrest between two consecutive time points during the intervention and identify characteristics that may increase FSW vulnerability to arrest in Karnataka. METHODS: Structural interventions with police involved advocacy work with senior police officials, sensitization workshops, and integration of HIV and human rights topics in pre-service curricula. Programmes for FSWs aimed to enhance collectivization, empowerment and awareness about human rights and to introduce crisis response mechanisms. Three rounds of integrated behavioural and biological assessment surveys were conducted among FSWs from 2004 to 2011. We conducted bivariate and multivariate analyses using data from the second (R2) and third (R3) survey rounds to examine changes in arrests among FSWs over time and to assess associations between police arrest, and the sociodemographic and sex work-related characteristics of FSWs. RESULTS: Among 4110 FSWs surveyed, rates of ever being arrested by the police significantly decreased over time, from 9.9% in R2 to 6.1% in R3 (adjusted odds ratio (AOR) [95% CI]=0.63 [0.48 to 0.83]). Arrests in the preceding year significantly decreased, from 5.5% in R2 to 2.8% in R3 (AOR [95% CI]=0.59 [0.41 to 0.86]). FSWs arrested as part of arbitrary police raids also decreased from 49.6 to 19.5% (AOR [95% CI]=0.21 [0.11 to 0.42]). Certain characteristics, including financial dependency on sex work, street- or brothel-based solicitation and high client volumes, were found to significantly increase the odds of arrest for participants. CONCLUSION: Structural interventions addressing police arrest of FSWs are feasible to implement. Based on our findings, the design of violence prevention and response interventions in Karnataka can be tailored to focus on FSWs, who are disproportionately vulnerable to arrest by police. Context-specific structural interventions can reduce police arrests, create a safer work environment for FSWs and protect fundamental human rights.