Gender-based violence, defined as ‘any act of violence that results in, or is likely to result in, physical, sexual or mental harm or suffering to women, including threats of such acts, coercion or arbitrary deprivation of liberty, by any person regardless of their relationship to the victim, in any setting’, is a human rights and public health issue associated with complex physical, mental, sexual, reproductive and general health problems (World Health Organization 2017, 2). The consequences of gender-based violence include fatal and non-fatal injuries, depression, post-traumatic stress, sexually transmitted infections (STI), chronic diseases, unintended pregnancies, unsafe abortions and substance abuse (Pico-Alfonso et al. 2006; García-Moreno et al. 2013). Gender-based violence also has social and economic costs that hinder national development goals (Duvvury et al. 2013; United Nations 2015a).

An estimated 35% of women globally have experienced gender-based violence; among women in relationships, 30% have experienced intimate partner physical or sexual violence (García-Moreno et al. 2013; Devries et al. 2013). Gender-based violence prevalence is highest in Africa, where almost half of countries report lifetime prevalence among women over 40% (United Nations 2015b). Few data on gender-based violence are available for Mozambique. The 2011 Mozambican Demographic Health Survey (DHS) found that 46% of married or cohabitating women aged 15–49 years had experienced physical or sexual violence in the past year (Instituto Nacional de Estatística 2013). A study of women who sought forensic services at the Maputo Central Hospital found that, of the 1,433 participants, 55% had experienced severe intimate partner violence and 29% had been sexually coerced in the 12 months prior to seeking services (Zacarias et al. 2012).

Female sex workers are particularly vulnerable to physical and sexual violence (Dunkle and Decker 2013) because of sociocultural and structural factors such as stigma and discrimination, gender-based power disparities, economic inequalities, mobility and migratory lifestyle, trafficking or forced labour experiences, the legal status of sex work, policing practices and patriarchal attitudes (Deering et al. 2014; Ramesh et al. 2012; Okal et al. 2011; Shannon et al. 2015; Renzaho and Pallotta-Chiarolli 2009; Pauw and Brener 2003; Agha and Nchima 2004; Scorgie et al. 2013). A systematic review of workplace violence against female sex workers globally found lifetime prevalence ranging from 45% to 75% and 12-month prevalence from 32%–55% (Deering et al. 2014). Studies in sub-Saharan Africa report lifetime physical and sexual violence ranging from 47% to 82% and, in one study, a 71% prevalence of violence in the past 12 months (Mooney et al. 2013; Alemayehu et al. 2015; Coetzee, Gray and Jewkes 2017; Coetzee, Jewkes and Gray 2017; Muldoon et al. 2017; Lim et al. 2015; Wirtz et al. 2015; Decker et al. 2016; Schwitters et al. 2015; Lyons et al. 2017).

Violence against female sex workers often results from negotiations for payment, condom use and sexual encounter duration (Kiernan, Mishori and Masoda 2016; Schulkind et al. 2016). Negotiations are complicated and often deteriorate in work environments where alcohol use is common (Leddy et al. 2018; Schulkind et al. 2016; Schwitters et al. 2015; Wilson et al. 2016; Panchanadeswaran et al. 2008; Mbonye et al. 2014; Lees et al. 2009). The resulting violence causes women to have decreased condom self-efficacy and consistent condom use with clients, and an increased risk of gonorrhoea, syphilis and HIV infection (Dunkle and Decker 2013; Ramesh et al. 2012; Shannon et al. 2015; Beattie et al. 2015; Reed et al. 2016; Goodman-Meza et al. 2014; Madhivanan et al. 2005). Preventing violence improves health outcomes and has the potential to avert approximately 20% of new HIV infections among female sex workers; modelling also demonstrates that there are cascading prevention benefits for clients and the general population (Shannon et al. 2015; Decker et al. 2013).

Research on female sex workers in Mozambique is limited, and no specific data on gender-based violence or vulnerability to violence existed prior to this survey. Although sex work is now legal in the country, acts defined as contrary to decency and public morals are excluded (Republica de Moçambique 2014). This leaves sex workers vulnerable to harassment and prosecution. Female sex workers also face great health risks in Mozambique, where the HIV prevalence among this group of women is high (17.8%–31.2%) (Augusto et al. 2016). We present study results describing violence and related risk factors with the goal of strengthening advocacy to reduce violence against female sex workers and informing HIV prevention programmes in Mozambique.

Our analysis reveals that female sex workers in Mozambique experienced high levels of physical and sexual violence, which notably is associated with condomless sex with clients and STI symptoms. This finding is consistent with those in other settings and has important public health implications (Dunkle and Decker 2013; Ramesh et al. 2012; Shannon et al. 2015; Beattie et al. 2015; Reed et al. 2016). During a violent encounter, a woman’s ability to negotiate the circumstances of sex (e.g. where the sexual exchange occurs and what type of sex transpires) is compromised, and HIV and STI prevention is often sacrificed in the face of implied pressure or overt coercion (Dunkle and Decker 2013; Lees et al. 2009). A mediating factor for violence and aggression is often a client’s desire for condomless sex and the woman’s resistance (Dunkle and Decker 2013; Okal et al. 2011; Kiernan, Mishori and Masoda 2016; Schulkind et al. 2016).

Our study also showed that women aged 15–24 years were almost twice as likely to have experienced violence as female sex workers older than 25 years. Consistent with our findings, other research shows that adolescent age at sex work entry and less sex work experience is associated with greater vulnerability to physical and sexual violence (Alemayehu et al. 2015; Silverman 2011). In contrast, older women with more experience may mitigate threats by identifying potentially violent clients, securing safer work locations and reducing police conflict (Okal et al. 2011).

Another striking study finding was the small proportion of female sex workers who sought legal recourse or medical care after experiencing physical or sexual assault. Reporting violence to the police, in particular, may be perceived as a risk because police in many settings use the law to harass, threaten, beat and sexually coerce female sex workers (Decker et al. 2015; Arnott and Crago 2009; Schulkind et al. 2016; Pauw and Brener 2003; Agha and Nchima 2004; Onyango et al. 2015). This was likely a concern of participants in Mozambique during our study implementation when a penal code from 1886 provided the legal framework (Republica Portuguesa 1919) and stated that the ‘prostitute’ may be subject to security measures in cases of ‘… public scandal or continuously disobey[ing] the police prescription’. This code, which permitted subjective police interpretation, was replaced with an updated code in 2014 that still leaves female sex workers in a precarious position by penalising behaviour against decency and public morals (República de Moçambique 2014). In medical settings, the Ministry of Health encourages providers to report violence and refer survivors to legal services (Romão et al. 2012); however, stigma experienced in some clinics, combined with uncertain legal standing, could deter female sex workers from seeking medical care.

The regional variation in the prevalence of violence experienced by female sex workers (10.0% in Maputo, 22.8% in Beira and 25.6% in Nampula) may highlight cultural, religious and societal factors in Mozambique (Audet et al. 2010) that require ethnographic research to be better understood. For example, the cities have differing urbanity, educational access and migration to and from neighbouring countries (Raimundo 2009; van der Berg, da Maia and Burger 2017) that may influence violence. Similarly, further exploration is needed on the observed differences between our data and violence experienced nationally and by female sex workers regionally. The prevalence of violence against participants in our study, although unacceptably high, was lower than the prevalence of violence experienced by Mozambican women in the general population and female sex workers in other sub-Saharan African countries (Instituto Nacional de Estatística 2013; Berger et al. 2018; Misganaw and Worku 2013; Renzaho and Pallotta-Chiarolli 2009; Muldoon et al. 2017; Wilson et al. 2016; Fawole and Dagunduro 2014; Alemayehu et al. 2015; Micheni et al. 2015; Sherwood et al. 2015; Mooney et al. 2013; Schwitters et al. 2015).

We recognise that a large proportion of the survey sample was composed of younger women, and age, an important confounding variable, may limit generalisability of our findings to the total female sex worker population of each city. Nevertheless, coupon return rates were similar to other studies, and weighting accounts for population age distribution. Another limitation derives from the secondary nature of this analysis, which uses data collected primarily to measure HIV prevalence. This resulted in a limited number of violence-related questions and insufficient power to analyse sexual and physical violence separately. Additionally, the definition used in the survey included only sex work-related physical violence and, as such, does not capture other experiences. The lack of consistent definitions of violence and reference time periods across studies obfuscates differences and makes comparisons difficult. For example, physical violence in our survey was defined as being hit or beaten in the prior six months specifically for being a female sex worker, whereas the DHS defines physical violence more broadly with a 12-month period of exposure. The reliance on self-reported violence also may have biased results. Participants may have been reluctant to disclose their experience with violence and/or may have misclassified violence, associating it with terms of sex work. It is well-documented that many female sex workers view violence as ‘part of the job’ and forced sex as repayment for food and drinks consumed (Okal et al. 2011; Decker et al. 2015; Leddy et al. 2018). Finally, the exclusion of alcohol use from analysis because of missing responses is a limitation as it is an important factor in violence against female sex workers (Leddy et al. 2018; Schulkind et al. 2016; Schwitters et al. 2015; Wilson et al. 2016; Panchanadeswaran et al. 2008; Mbonye et al. 2014).

Despite these limitations, our analysis highlights female sex worker vulnerability in Mozambique and the need for comprehensive services to address the complexity of violence and resulting risks for survivors. At the individual level, female sex workers could benefit from integrated programmes that combine HIV and STI interventions with gender-based violence prevention (Bekker et al. 2015; Schulkind et al. 2016). Model programming has demonstrated reduced violence and STI prevalence using strategies such as peer outreach, condom promotion and provision and HIV and STI testing and treatment (Beattie et al. 2015). Another potential component of comprehensive services for female sex workers in Mozambique is cash transfers. This may limit early sex work debut, thereby reducing the risk of violence for younger women, and reduce dependence on violent sexual relationships for monetary gain (Fulu, Kerr-Wilson and Lang 2014). Given the increased risk of sexual infection from condomless penetration, the provision of pre-exposure prophylaxis (PrEP) for HIV should also be standard to keep uninfected women healthy. Following an assault, presumptive STI treatment, emergency contraception and post-exposure prophylaxis (PEP) also should be routinely administered. The inclusion of mental health interventions is equally important, as female sex workers who are sexually assaulted have a high burden of post-traumatic stress disorder (Rossler et al. 2010).

At the community and structural levels, efforts to mitigate violence against female sex workers in Mozambique should include strengthening networks, providing advocacy and legal support and promoting safer work environments. Peer support, galvanised by the formal or informal creation of female sex worker associations, can facilitate ‘bad date’ warning systems as an additional method of protection from violence (Goldenberg, Duff and Krusi 2015; Lim et al. 2015; Schulkind et al. 2016; Leddy et al. 2018; Cange et al. 2017). This strategy, along with guidance from experienced female sex workers (Stadler and Delany 2006), may benefit younger women who were more vulnerable in our study. Rapid violence response through 24-hour crisis management teams has also proven to be effective elsewhere (Beattie et al. 2015). Simple structural strategies to promote safer work environments include making changes to venues (e.g. improving the lighting), enacting supportive managerial practices (e.g. restricting alcohol sales to inebriated clients), providing accessible HIV/STI prevention and improving relations with police (Shannon et al. 2015; Goldenberg, Duff and Krusi 2015). As previously noted, the police who interpret laws governing sex work contribute to an environment in which violence against female sex workers is tolerated and easily perpetrated (Decker et al. 2015; Pauw and Brener 2003; Agha and Nchima 2004; Schulkind et al. 2016). In the absence of clear legislation that decriminalises sex work in Mozambique, there is a need to clarify how the current code is operationalised and interpreted by the judicial system and police in an effort to better protect the rights and health of female sex workers.

Comprehensive services and policy strategies constitute an appropriate public health response to violence perpetrated against female sex workers in Mozambique that will benefit the communities where they work and reduce new HIV infections attributable to sex work. Therefore, violence prevention for female sex workers should be a priority for the country’s HIV care and treatment strategy.