The advent of the human immunodeficiency virus (HIV) pandemic has led to increased awareness about the violence experienced by women 1 2 and by female sex workers (FSWs) worldwide 3 4 . Research on domestic violence from developing countries suggests that anywhere from 10-60% of married women of reproductive age report experiencing some form of domestic violence 2 , and that domestic violence is strongly associated with physical and mental health morbidity, including homicide, suicide, physical injuries and emotional distress 5 6 , as well as HIV seropositivity 1 7 8 .

FSWs are frequently marginalised from society due to sex work lacking social or moral approval. In addition to experiencing physical and sexual violence from their intimate partners 9 , they can also experience violence from others in their personal and working lives, including clients, pimps, madams and the police 4 10 11 . Compared with the general population, HIV/STI prevalence rates are frequently higher among FSW populations and their clients due to high partner turnover and concurrent sexual partnerships, leading to substantial HIV/STI transmission, particularly when condom use is low 12 . As well as negatively impacting on their mental health and emotional wellbeing 11 13 , violence against sex workers can heighten their vulnerability to HIV and other sexually transmitted infections (STIs) through multiple mechanisms: (i) coerced sex is rarely protected; (ii) coerced sex can result in injuries that can increase the transmission of STIs, which in turn can increase the risk of HIV transmission 14 ; (iii) men who are sexually violent may be more likely to have multiple partners and be infected with HIV and/or STIs 8 15 16 ; (iv) sex work is usually illegal, and thus reporting of sexual and physical violence to the authorities is difficult 3 11 , meaning violence can continue unchecked; (v) the fear of violence from regular partners (husbands/lovers) resulting from inadvertent disclosure of sex work can deter sex workers from negotiating condom use with these partners and from accessing sexual health services 13 17 ; and (vi) mental health morbidity arising from violence can reduce the ability of sex workers to negotiate condom use and to access STI services for testing and treatment.

India is experiencing an emerging, predominantly heterosexual HIV epidemic, with approximately 2.5 million individuals infected 18 19 . Although sex work per se is not illegal in India (the emphasis of the Indian Immoral Traffic Prevention Act, 1956 being on penalising the act of solicitation, as well as penalising those who profit from sex workers such as madams, pimps, brothel keepers and traffickers, rather than penalising sex work itself), the majority of police officials and sex workers wrongly understand sex work to be against the law 20 . In a setting where gender-based domestic violence is common 8 21 22 , marital rape could not (until 2008) be prosecuted, and sex work is widely perceived as illegal and immoral, sex workers often feel that they have little recourse against violence and rape, and the perpetrators of violence against sex workers are unlikely to be punished for their actions 23 .

Karnataka state in south India has a population of approximately 54 million and ranks among the top four states in India with regard to HIV epidemic severity 24 . Although the estimated HIV/AIDS prevalence among the general population is approximately 1%, rates among FSWs have reached >25% in some districts. The Karnataka Health Promotion Trust (KHPT), supported by the Bill & Melinda Gates Foundation, has worked since 2003 to deliver targeted HIV preventive intervention programs to over 60,000 FSWs and 20,000 MSM (men who have sex with men)/transgenders in 21 of the 30 districts in the state. FSWs in Karnataka state are often from the lower castes, and are often poor, uneducated, and may have children to support, leaving few economic alternatives for survival if they are deserted by their husbands or become widowed 25 . In our early dialogue with sex workers, it became clear that violence was a key concern. In addition, finding shelter, food and child-care concerns were major priorities, whereas health issues rarely featured. The program realised that physical and sexual violence against the sex worker community could potentially undermine the efforts of HIV prevention programming in multiple ways. The violence experienced by sex workers was considered a priority issue to address, as part of the provision of sexual health education and services. The program therefore created a framework and program to do so, beginning in 2005. In this paper we provide details of a multi-layered strategy involving policy makers, secondary and primary stakeholders, to stem and address violence against the sex worker community as part of a wider HIV intervention program. We then examine the impact of these violence intervention efforts on levels of violence against FSWs, and examine associations between violence and condom use, HIV/STI rates and exposure to the HIV prevention program components.

We found strong evidence of a reduction in the proportion of FSWs who reported violence at follow-up compared with baseline, following the implementation of a multi-layered violence intervention as part of a wider HIV prevention program. In addition, we found that FSWs who reported experiencing violence in the past year were less likely to report using condoms with their clients, were more likely to be infected with gonorrhea, and were less likely to have accessed the HIV prevention services than FSWs who did not report violence.

There is growing recognition of the impact of physical and sexual violence on the mental and physical health of women, including HIV infection 3 13 33 , and studies with non-sex workers have reported reductions in violence within intimate partnerships as part of broader HIV intervention programs 34 35 36 . However, although several studies have reported violence against marginalised groups such as sex workers 10 11 13 37 38 , to our knowledge, this is the first study to report on a multi-layered strategy to address violence against sex workers, as part of HIV prevention programming. Consistent with studies from elsewhere, data from our IBBA surveys suggest that FSWs who reported being beaten or raped in the past year were more likely to be infected with gonorrhea than FSWs who did not 22 . The possible mechanisms through which violence can enhance vulnerability to STIs are multiple and complex, and are likely to be both direct (for example, having unprotected forced intercourse with more risky partners, or being in unsafe situations for negotiating condom use) 10 11 13 15 38 ; and indirect (for example depression resulting from victimisation may reduce motivation to access HIV intervention services or negotiate condom use) 2 . Indeed, similar to findings from elsewhere, FSWs in our study who reported violence in the previous year were less likely to have been contacted by a peer educator or to have accessed the project sexual health clinics, and were much less likely to report using condoms with their clients 10 . In addition to increased knowledge and awareness of sexual risk leading to increase condom use following participation in an HIV prevention program, the conversion of a hostile environment to a more facilitative enabling environment for sex work has been reported elsewhere as a contributing factor to improved condom use 39 . Furthermore, reducing violence can translate into sex workers accessing services more easily, as well as carrying condoms without fear of their being used as evidence of sex work. In addition, working in an environment less threatened by violence supports FSWs to make choices based on their own volition 20 .

As this study was evaluating a violence intervention program which has been implemented throughout Karnataka state, it was not possible for us to include a control group for comparison. With unlinked cross-sectional survey data such as these, and without a control group, it is always possible that the findings presented here represent natural trends over time and are independent of the violence intervention efforts. However, the strength and consistency of the associations across virtually all districts surveyed suggests that this is unlikely. Moreover, when we stratified participants in the IBBA surveys according to length of exposure to the HIV intervention program, violence in the preceding year was negatively associated with increasing duration of program exposure. Another limitation of the study is that it was not possible to separate out the impact on violence of individual aspects of the program, such as the policy level advocacy vs. the crisis response program vs. the police officer training vs. increasing access to social entitlements. Future qualitative and policy-oriented research would be important to undertake in the future to address this issue. Another important issue to address is program cost. It has recently been reported that the median cost per female sex worker registered in the first two years of Avahan programs in four south India states (including Karnataka) was $53, which is comparable to that of similar programs elsewhere 40 . Future work should examine the proportion of costs attributable to the components of the program directed at violence reduction.

Rates of reported violence were very high, and were much higher in the PBS compared with the IBBA, likely reflecting social desirability bias in the face-to-face interviews, and highlighting the importance of using alternative survey methods when collecting information on sensitive behaviours 6 29 30 31 . The elevated rates of reported violence in the 2007 PBS survey compared with the 2006 PBS survey may have reflected increased violence during this time period, as perpetrators tried to re-assert their power 20 . Alternatively, it may be that women became more comfortable with reporting violence, and more understanding of the nature of violence, and so were more likely to report it. In any case, reported violence in 2008 in the PBS was significantly lower than either of the preceding years across 12 of the 13 districts surveyed. Discussions with FSWs suggest that violence may have been under-reported 2 , particularly in the baseline surveys, not least because at this time many sex workers did not consider sexual or physical violence by their intimate partners as being untoward events. Furthermore, the definition of violence used in this study was quite narrow, and thus may not have captured all violence, leading to further under-reporting. Baseline surveys were conducted 8-16 months after program initiation, by which time levels of violence may have already started to reduce, and thus the reductions in violence may have been underestimated; unfortunately, it is difficult to undertake cross-sectional surveys such as these among sex worker communities until prevention programs are underway and trust with the community is well established.

Embedding a violence intervention program such as this one within a broader HIV prevention framework has several challenges. Dealing with violence directly may conflict with power structures which are responsible for violence, and can perhaps, in the short term, put FSWs at increased risk of harm. The relationships among violence, FSWs and the perpetrators of violence are complex, with perpetrators (such as police, lovers and pimps) having the potential to act as protectors, as well as purveyors of violence 17 20 . The routine transfer of police personnel and the highly mobile nature of FSW populations suggest that in order for reductions in violence to be sustained, training of police personnel and legal empowerment training need to be ongoing.

The data presented here highlight the potential for violence to undermine HIV prevention efforts, and underline the importance of addressing violence as part of wider HIV prevention programming. Such an approach can reduce the levels of violence faced by this vulnerable population, and is key to protecting the basic right of sex workers to live and work in an environment free from physical and sexual violence and abuse.

The authors declare that they have no competing interests.

TSHB performed the analyses and wrote the first draft of the paper. VG, PB, JA and HLM designed and oversaw local implementation of the intervention program. BMR and SI designed and supervised the IBBA and PBS surveys. AR, TW, JB and JFB were involved in the study design, and contributed to writing the paper. SM designed the study and contributed to writing the paper. All authors have read and approved the final manuscript.