Associations between sex work laws and sex workers’ health: A systematic review and meta-analysis of quantitative and qualitative studies

Background Sex workers are at disproportionate risk of violence and sexual and emotional ill health, harms that have been linked to the criminalisation of sex work. We synthesised evidence on the extent to which sex work laws and policing practices affect sex workers’ safety, health, and access to services, and the pathways through which these effects occur. Methods and findings We searched bibliographic databases between 1 January 1990 and 9 May 2018 for qualitative and quantitative research involving sex workers of all genders and terms relating to legislation, police, and health. We operationalised categories of lawful and unlawful police repression of sex workers or their clients, including criminal and administrative penalties. We included quantitative studies that measured associations between policing and outcomes of violence, health, and access to services, and qualitative studies that explored related pathways. We conducted a meta-analysis to estimate the average effect of experiencing sexual/physical violence, HIV or sexually transmitted infections (STIs), and condomless sex, among individuals exposed to repressive policing compared to those unexposed. Qualitative studies were synthesised iteratively, inductively, and thematically. We reviewed 40 quantitative and 94 qualitative studies. Repressive policing of sex workers was associated with increased risk of sexual/physical violence from clients or other parties (odds ratio [OR] 2.99, 95% CI 1.96–4.57), HIV/STI (OR 1.87, 95% CI 1.60–2.19), and condomless sex (OR 1.42, 95% CI 1.03–1.94). The qualitative synthesis identified diverse forms of police violence and abuses of power, including arbitrary arrest, bribery and extortion, physical and sexual violence, failure to provide access to justice, and forced HIV testing. It showed that in contexts of criminalisation, the threat and enactment of police harassment and arrest of sex workers or their clients displaced sex workers into isolated work locations, disrupting peer support networks and service access, and limiting risk reduction opportunities. It discouraged sex workers from carrying condoms and exacerbated existing inequalities experienced by transgender, migrant, and drug-using sex workers. Evidence from decriminalised settings suggests that sex workers in these settings have greater negotiating power with clients and better access to justice. Quantitative findings were limited by high heterogeneity in the meta-analysis for some outcomes and insufficient data to conduct meta-analyses for others, as well as variable sample size and study quality. Few studies reported whether arrest was related to sex work or another offence, limiting our ability to assess the associations between sex work criminalisation and outcomes relative to other penalties or abuses of police power, and all studies were observational, prohibiting any causal inference. Few studies included trans- and cisgender male sex workers, and little evidence related to emotional health and access to healthcare beyond HIV/STI testing. Conclusions Together, the qualitative and quantitative evidence demonstrate the extensive harms associated with criminalisation of sex work, including laws and enforcement targeting the sale and purchase of sex, and activities relating to sex work organisation. There is an urgent need to reform sex-work-related laws and institutional practices so as to reduce harms and barriers to the realisation of health.


Methods and findings
We searched bibliographic databases between 1 January 1990 and 9 May 2018 for qualitative and quantitative research involving sex workers of all genders and terms relating to legislation, police, and health. We operationalised categories of lawful and unlawful police repression of sex workers or their clients, including criminal and administrative penalties. We included quantitative studies that measured associations between policing and outcomes of violence, health, and access to services, and qualitative studies that explored related pathways. We conducted a meta-analysis to estimate the average effect of experiencing sexual/physical violence, HIV or sexually transmitted infections (STIs), and condomless sex, among individuals exposed to repressive policing compared to those unexposed. Qualitative studies were synthesised iteratively, inductively, and thematically. We reviewed 40 quantitative and 94 qualitative studies. Repressive policing of sex workers was associated with increased risk of sexual/physical violence from clients or other parties (odds ratio [OR] 2.99, 95% CI 1. 96-4.57), HIV/STI (OR 1.87, 95% CI 1. 60-2.19), and condomless sex (OR 1.42, 95% CI 1.03-1.94). The qualitative synthesis identified diverse forms of police violence and abuses of power, including arbitrary arrest, bribery and extortion, physical and sexual violence, failure to provide access to justice, and forced HIV testing. It showed that in contexts of criminalisation, the threat and enactment of police harassment and arrest of sex workers or their clients displaced sex workers into isolated work locations, disrupting peer support networks and service access, and limiting risk reduction opportunities. It discouraged sex workers from carrying condoms and exacerbated existing inequalities experienced by transgender, migrant, and drug-using sex workers. Evidence from decriminalised settings suggests that sex workers in these settings have greater negotiating power with clients and better access to justice. Quantitative findings were limited by high heterogeneity in the meta-analysis for some outcomes and insufficient data to conduct meta-analyses for others, as well as variable sample size and study quality. Few studies reported whether arrest was related to sex work or another offence, limiting our ability to assess the associations between sex work criminalisation and outcomes relative to other penalties or abuses of police power, and all studies were observational, prohibiting any causal inference. Few studies included trans-and cisgender male sex workers, and little evidence related to emotional health and access to healthcare beyond HIV/STI testing.

Conclusions
Together, the qualitative and quantitative evidence demonstrate the extensive harms associated with criminalisation of sex work, including laws and enforcement targeting the sale and purchase of sex, and activities relating to sex work organisation. There is an urgent need to reform sex-work-related laws and institutional practices so as to reduce harms and barriers to the realisation of health.

Author summary
Why was this study done?
• To our knowledge there has been no evidence synthesis of qualitative and quantitative literature examining the impacts of criminalisation on sex workers' safety and health, or the pathways that realise these effects.
• This evidence is critical to informing evidenced-based policy-making, and timely given the growing interest in models of decriminalisation of sex work or criminalising the purchase of sex (the latter recently introduced in Canada, France, Northern Ireland, Republic of Ireland, and Serbia).

What did the researchers do and find?
• We undertook a mixed-methods review comprising meta-analyses and qualitative synthesis to measure the magnitude of associations, and related pathways, between criminalisation and sex workers' experience of violence, sexual (including HIV and sexually transmitted infections [STIs]) and emotional health, and access to health and social care services.
• We searched bibliographic databases for qualitative and quantitative research, categorising lawful and unlawful police repression, including criminal and administrative penalties within different legislative models.

Introduction
Sex workers can face multiple interdependent health risks [1,2]. Between 32% and 55% of cisgender (cis) women working mostly in street-based sex work report experience of workplace violence in the past year [3]. Across diverse settings, both cis and transgender (trans) women and men in sex work are at increased risk of experiencing violence and homicide [4][5][6], HIV infection [7][8][9], chlamydia and gonorrhoea [10,11], and poorer mental health than their nonsex-working counterparts [12]. Yet there is considerable variation within sex-working populations [13,14]. The epidemiological context as well as social and structural factors and power relations reproduce inequalities within sex-working populations [2,3,8,9]. For example, cis women working in street-based sex work are more vulnerable to all these outcomes than those working in off-street settings [15,16]. Many vulnerabilities faced by sex workers are multiplicative, closely linked to poverty, substance use, disability, immigration, sexism, racism, transphobia, and homophobia [17].
Qualitative literature demonstrates how social policies and structural factors shape the health and welfare of sex workers. The 'risk environment' concept, developed to understand drug-related harms [18] and adapted to HIV and violence experienced by sex workers [19,20], examines different types (physical, social, economic, and political) and levels of environmental influence (micro and macro), in line with broader efforts to address structural determinants of health [21]. This concept has been used to demonstrate how policing, stigma, and inequalities interplay to shape sex workers' vulnerability to HIV [22], violence [23], and lack of access to healthcare [24] and justice [25,26], and the potential for sex-worker-led interventions to challenge these harms [27]. Epidemiological evidence documents the associations between macrostructural factors (laws, housing and economic insecurity, migration, education, and stigma) and work environment and community factors (policing, work setting and conditions, autonomy, and access to health and peer-led services) and sex workers' risk of violence and HIV transmission [2,3]. Criminalisation and repressive public health approaches to sex work (e.g., mandatory registration and HIV/sexually transmitted infection [STI] testing) have been shown to hinder the prevention of HIV, where the focus of interventions and research has been directed [28][29][30]. Conversely, mathematical modelling has estimated that decriminalisation of sex work could halve the incidence of HIV among sex workers and their clients over a 10-year period [2], and evidence from New Zealand indicates that sex workers in decriminalised settings report improved workplace safety, health and social care access, and emotional health [31,32].
Broadly, there are 5 legislative models used to manage, control, or regulate sex work ( Table 1) [33]. Full criminalisation prohibits all organisational aspects of sex work and selling and buying sex. Partial criminalisation is where some aspects of sex work are penalised (e.g., soliciting sex in public for sex workers and/or clients, advertising services, collective working, or involvement of third parties). In 1999, Sweden criminalised the purchase, but not the sale, of sex, and various other countries have followed [34]. This 'criminalisation of clients' model typically retains laws against 'brothel-keeping', which may in practice also target sex workers working together. Regulatory models make the sale of sex legal in certain settings (e.g., in licensed brothels or managed zones) or under certain conditions (e.g., mandatory registration or HIV/STI testing) but illegal in other settings or for individuals who do not meet registration requirements or eligibility criteria (e.g., migrants, cis men and trans sex workers, or people living with HIV) [35]. Full decriminalisation, implemented in New Zealand in 2003, removes criminal penalties for adult sex work, emphasises enforcing criminal laws prohibiting violence

Full criminalisation
All aspects of selling and buying sex or organisation of sex work are prohibited. South Africa, Sri Lanka, US $ and coercion, and regulates the sex industry through occupational health and safety standards [36]. All models criminalise coerced sex work and the involvement of minors, and almost all models-including decriminalisation in New Zealand-prohibit migrants without permanent residency from working legally or in a regulated environment. In practice the implementation of these models through bylaws and enforcement practices is complex, and varies between and within countries and even locally within cities [37,38]. The debate around sex work policy and legislation is highly polarised. Some argue that all sex work is itself gendered violence and should be repressed-a notion that underpins the criminalisation of sex workers' clients [39,40]. Others argue that this fails to recognise the diversity of experience and identity in the sex industry and the possibility that financial reimbursement for sex between adults can be consensual [41]. At a time of increasing political interest in legislative reform [42][43][44][45], there is a critical need to bring together this evidence to inform policies that protect sex workers' safety, health, well-being, and broader rights. We conducted a systematic review to synthesise evidence of the extent to which sex work laws and their enforcement affect sex workers' safety, health and access to services, and the processes and pathways through which these effects occur, including in interaction with other macrostructural, community, and work environment factors.

Data extraction and quality assessment
Following a protocol with pre-specified search terms, we searched MEDLINE, CINAHL, Psy-chINFO, Web of Science, and Global Health for public health and social science literature on studies that combined 3 search domains: (1) sex work, AND (2) legislation OR policing, AND (3) health (physical or emotional, including violence/safety) OR access to services (including health, risk reduction, and social care/support). The complete search terms and review protocol are attached (S1 Text). Meta-analyses were not pre-specified, since they were subject to identifying sufficiently homogenous studies in relation to outcomes and definition of criminalisation.
Three authors screened the sources for inclusion, discussing any uncertainties within the team; a second person re-reviewed relevant sources when necessary. Quantitative data were extracted and analysed by LP and JE, and qualitative data synthesised by PG and RM. For qualitative and quantitative studies, we defined quality-related criteria adapted from the Critical Appraisal Skills Programme (CASP) [46] that papers had to fulfil in order to qualify for inclusion: methods and ethics processes described, appropriate study population clearly defined, and conclusions supported by study findings. Quantitative studies were further assessed according to appropriateness of study design, data collection methods, and analyses, using assessment approaches adapted from the Newcastle-Ottawa scale and CASP [46,47]. A full copy of the quality assessment process for the quantitative studies is available (S1 Table). For qualitative evidence, confidence in review findings was assessed according to CERQual guidance, taking account of methodological limitations, coherence, adequacy of data, and relevance of included studies (S2 Text) [48]. Methodological limitations were assessed using CASP guidelines for qualitative evidence.
or otherwise influence the growth, instance or expression of prostitution' [33]. We also included the use of non-criminal penalties to target sex workers, such as fines and displacement orders, including those that do not formally relate to sex work. Within the broad legislative models (Table 1), sex work legislation and policing was operationalised into 8 different categories of police exposure: (1) police repression on an environment in which sex work takes place (workplace raids, zoning restrictions, and displacement from usual working areas), (2) recent (within last year) arrest or prison, (3) past arrest or prison, (4) confiscation of condoms or needles or syringes, (5) extortion (giving police information, money, or goods to avoid arrest), (6) sexual or physical violence from police (negotiated or forced), (7) fear of police repression, and (8) registration as a sex worker at a municipal health authority. Where clear from included papers, we recorded data on gender using the terms 'cis' and 'trans' to refer to people who do and do not identify themselves with the gender they were assigned at birth, respectively. Conscious of cultural diversity in gender identities, we use the term 'transfeminine' to describe feminine-presenting trans populations that do not necessarily describe themselves as female/women [49]. We did not identify any papers that discussed the experiences of people who identify their gender as trans male/masculine or non-binary.

Inclusion criteria
We included quantitative, qualitative, and mixed-methods studies published in English, Russian, or Spanish, and included data specific to the experiences of sex workers. We included papers that measured quantitative associations between criminalisation or decriminalisation of sex work, or repressive policing practices within these contexts, and the following outcomes: threatened or enacted violence, STIs, HIV, hepatitis B/C, overdose, stress, anxiety, depression, risk practices/management (e.g., working with others, reporting violence, condom use, sharing needles/syringes), and access to health/social care services (HIV/STI/hepatitis prevention, testing, and treatment; contraception; abortion; opioid substitution therapy and other drug/alcohol services; mental health and counselling; primary and secondary care; psychosocial support services; housing; and social security). We also included studies that reported qualitative data on the relationships between experiences of criminalisation or decriminalisation and policing and sex workers' experiences of violence, safety, health, risk management, and/or accessing health or social care services, from the perspectives of sex workers themselves.

Data synthesis
We synthesised estimates that adjusted for confounders to assess overall risk of experience of physical or sexual violence, HIV/STI, and condomless sex, stratified by the categories of repressive police activities described above. Where multiple policing practice exposures were presented in the same study, we selected independent estimates in an overall pooled estimate prioritising recent experience of arrest/prison and the most commonly occurring outcomes. Studies including sex workers of different genders were pooled together. We applied random effects models using the DerSimonian and Laird method for all analyses, allowing for heterogeneity between studies and converting all effect estimates into odds ratios (ORs) [50]. We examined heterogeneity with the I 2 statistic. We conducted sub-group analyses to describe differences in experience of violence and condom use by partner type (client versus intimate partner/other) and by type of violence (physical versus sexual or sexual/physical combined). We conducted sensitivity analyses to look at overall associations between policing and our specified outcomes, excluding or pooling studies that did not adjust for confounders or reported only STI outcomes (self-reported and biological) or composite HIV/STI, and altering the priority choice of police exposure (from recent arrest/prison to other). We conducted a narrative synthesis of outcomes that were too heterogeneous to pool, including access to services (both mandatory and voluntary uptake of services), harms related to drug use, and emotional health. Studies that measured associations with registration at the municipal health department were also synthesised separately, since this policy was less comparable with all others that involved direct police action. All analyses were conducted using the metafor package in R version 3.4.1 and RStudio version 1.0.143 [51].
For qualitative studies, data were synthesised inductively, iteratively, and thematically. From the body of eligible papers we first focused on the 'data-rich' papers that contributed substantive or moderate data and analyses relevant to our research questions. Among the body of papers that had a limited focus on the topic, we then purposively sampled studies that reported on an under-represented population, setting, legislative model, or health issue of interest in this review [52] until no new themes emerged (thematic saturation). For the datarich papers, we reviewed and wrote summaries of the results and discussion sections, inductively and iteratively drawing out author-and reviewer-identified themes and sub-themes. We then linked sub-themes and themes to 4 core categories, informed by concepts of structural, symbolic, and everyday violence that argue that mistreatment, stigma, exclusion, and ill health often result from intersecting inequalities that become institutionalised and normalised through policies, practices, and social norms [53]. We paid careful attention to the different levels and forms of environmental influence within risk environments [18]. Finally, we reviewed the less data-rich papers (relative to our research questions) against these emerging categories until they required no further refining. We summarise the core categories narratively with illustrative quotes (Box 1), drawing out findings that help to unpack the quantitative associations and their causal pathways. Within each category, we pay close attention to patterns by legislative model.

Results
From 9,148 papers identified, 134 studies met the inclusion criteria, resulting in 40 papers included in the quantitative synthesis, of which 20 were included in the meta-analysis and 20 in the narrative synthesis. A total of 94 met the inclusion criteria for the qualitative synthesis, of which 46 were included in the thematic analysis, 3 were excluded following quality assessment, and 45 were excluded when thematic saturation had been reached (Fig 1).
The overall effect estimate of repressive policing actions on HIV/STI outcomes was maintained across sensitivity analyses including those focusing on unadjusted estimates (OR 1. 85 Violence. We pooled data from 9 studies that measured the association between repressive policing activities and experience of physical or sexual violence against sex workers by a range of perpetrators, including clients, intimate (sex) partners, and police. Random effects meta-analysis of 9 independent multivariable estimates showed that, overall, repressive policing was associated with substantially higher odds of any kind of violence (5,204 participants, OR 2.99, 95% CI 1. 96-4.57), but with high heterogeneity (I 2 = 83.1%, 95% CI 65.3%-96.0%, p < 0.001). Sub-group analysis suggested that those who had their needles/syringes or condoms confiscated had higher odds of violence than those who did not (1,696 participants, OR 4.67, 95% CI 1.32-16.54, I 2 = 93.9%, 95% CI 76.2%-99.8%, p < 0.01) (Fig 3).
This overall association between police repression and violence increased slightly, but was still associated with substantially higher odds of violence, when all unadjusted estimates were pooled from 6 studies (OR 3.15, 95% CI 1.99-4.99, I 2 = 78.7%, 95% CI 52.5%-97.4%, p < 0.001) (S2 Fig). Odds of experiencing physical or sexual violence by other people (defined as anyone other than paying clients, including the police) was higher for those who had experienced any type of repressive police activity compared to those who had not (OR 3.72, 95% CI 1.74-7.95, I 2 = 84.1%, 95% CI 53.5%-99.0%, p < 0.001). Similarly, physical or sexual violence from clients was higher among those who had been exposed to repressive police activity compared to those who had not (OR 2.71, 95% CI 1.69-4.36, I 2 = 80.4%, 95% CI 45.5%-96.3%, p < 0.001) (S4 Fig). Condom use. Five studies measured the association between repressive policing activities and condom use with both paying and non-paying partners. Meta-analysis of 4 independent Table 2. Summary of quantitative study characteristics and associations between lawful and unlawful police repression and sex workers' experience of violence, condom use and HIV/STI outcomes, access to services, emotional health, and drug and alcohol use.
multivariable estimates (9,447 participants) suggested that on average these practices were associated with increased odds of not using a condom (OR 1.42, 95% CI 1.03-1.94), with moderate heterogeneity across the studies (I 2 = 63.34%, 95% CI 0.0%-98.2%, p = 0.04) (Fig 4). The overall association between repressive policing activities and condom use increased when pooling unadjusted estimates from 2 studies (OR 1.76, 95% CI 1.30-2.38, I 2 = 0.0%, 95% CI 0.0%-0.98%, p = 0.46) (S3 Fig). Sub-group analysis suggested that the odds of condomless sex with clients was higher following policing exposure (OR 1.42, 95% CI 1.03-1.94, I 2 = 63.3%, 95% CI 0.0%-98.2%, p = 0.04) or when additional money was offered (OR 1.54, 95% CI 1.10-2.15, I 2 = 66.7%, 0.0%-97.8%, p = 0.03). There was no difference in the odds of condomless sex with non-paying partners after police exposure (OR 1.0, 95% CI 0.80-1.24, I 2 = 0.0%, 95% CI 0.0%-17.7, p = 0.97) (S4 Fig). Access to services and mandatory testing. Five studies looked at the association between repressive policing activities and access to health and social care services. One study in India found that arrest in the last year was associated with increased odds of attendance at an STI Recent prison is presented as the outcome in the original analysis but as temporal associations were not measured the outcomes and exposure variables have been inverted for the review in order to facilitate comparison. � Unmet health need defined as sometimes, occasionally, or never getting healthcare services when you need them versus always or usually getting them [61]. # Subotnik is defined as sex demanded by police in exchange for leniency towards pimps and female sex workers in past 3 months [73].
\ Physical violence defined as ever having been violently pushed, shoved, slapped, hit, kicked, choked, or otherwise physically hurt. Sexual violence defined as ever having experienced forced sex through physical force, coercion, or penetration with an object against one's will [68].^H IV prevention package included condom distribution, community-based methadone maintenance treatment and/or needle and syringe programme, and peer HIV/ AIDS education [80]. + Emotional ill health defined as reported diagnosis of depression, post-traumatic stress disorder, anxiety, schizophrenia, borderline personality, attention deficit, or bipolar disorder within last 6 months [63].
https://doi.org/10.1371/journal.pmed.1002680.t002 clinic (OR 1.74, 95% CI 1.02-2.98, p = 0.04) [71]. Confiscation of needles/syringes in Mexico by the police was associated with increased odds of having an HIV test among sex workers who inject drugs (OR 1.49, 95% CI 1.09-2.05, p-value not reported) [63]. In Canada, fear of Health impact of sex work legislation police and police harassment, including arrests, was associated with avoiding healthcare services among street-based cis women [85] and cis and trans women [61]. Geospatial analyses among the same population showed that a higher density of police enforcement practices Health impact of sex work legislation (including displacement, legal restrictions of sex work areas, and police harassment) was associated with disrupted HIV treatment [56]. In Uganda, rushed negotiations with clients due to police presence was associated with less frequent dual contraceptive use (OR 0.65, 95% CI 0.42-1.00, p = 0.05) [83]. In a study in China, where HIV testing is mandatory following detention, history of arrest was associated with increased odds of having an HIV test or taking up HIV prevention interventions, but fear of arrest was associated with decreased odds of both HIV testing (OR 0.78, 95% CI 0.55-1.12, p = 0.18) and accessing prevention interventions (OR 0.39, 95% CI 0.22-0.68, p < 0.001) [80].
Emotional ill health. Three studies looked at indicators of emotional ill health. In India, cis female sex workers mostly working on the street who had been arrested had increased odds of major depression (defined through Patient Health Questionnaire-2) (OR 1.6, 95% CI 1.1-2.3, p = 0.05) compared to those who had not been arrested [88]. In Canada, recent incarceration was associated with poor emotional health outcomes among both cis and trans female sex workers in a univariable analysis (OR 1.55, 95% CI 1.12-2.14, p < 0.10) [60]. Among the same population, individuals who reported that the police had affected where they worked had increased work stress compared to those who did not report this [55]. Drug and alcohol use. Five studies examined the association between repressive policing practices and drug use including injecting drug use [60,66,86,87], the use of non-prescription opioids [27], and excessive alcohol drinking [60,66]. All of these studies showed a positive association between exposure to repressive policing practices and drug/alcohol use. One study among cis female sex workers in Mexico who inject drugs found a positive association between police confiscation of needles/syringes and injecting in public places (linked to increased risk of skin and soft tissue injuries but reduced risk of overdose) (OR 1.6, 95% CI 1.1-2.4, p-value not reported), as well as injecting in the groin area (linked to increased risk of overdose) (OR 1.9, 95% CI 1.2-2.9, p-value not reported), but reduced odds of injecting with clients (potentially linked to sharing needles/syringes but reduced risk of overdose) (OR 0.64, 95% CI 0.44-0.94, p-value not reported) [63]. Another study with the same population found that confiscation of needles/syringes was associated with lower safe injection self-efficacy at 8 months (−0.51, SE 0.25, p = 0.04) [62]. Recent history of incarceration was associated with use of crystal methamphetamine among cis female sex workers in Iran [86].
Registration at a municipal health service. Four studies reported associations between mandatory registration at a city health service in Tijuana, Mexico and health outcomes [79,81,89,90]. One study suggested that registered sex workers had reduced odds of working in a sex work venue with high prevalence of HIV or syphilis and testing positive for HIV or an STI (syphilis, gonorrhoea, or chlamydia) univariably. These associations became insignificant after adjusting for injecting risk behaviours, age, and time in sex work [79]. Of note, sex workers who test positive for HIV in this system have their registration revoked, and sex workers already living with HIV cannot work in the regulated sector; therefore, sex workers who know or suspect they are living with HIV are unlikely to register. Registered sex workers had reduced odds of ever injecting drugs and higher odds of being tested for HIV [81]. A final study suggested that lack of registration was associated with increased odds of unprotected sex (OR 2.1, 95% CI 1.2-3.5, p-value not reported) [90].
Evaluation of sex work policies. Two studies in Canada evaluated a new policing guideline that prioritised enforcement of laws against clients and third parties over arrest of sex workers introduced in Vancouver in 2013. These studies found that there was no decrease in physical and sexual violence (OR 1.09, 95% CI 0.59-2.04, p = 0.78) among participants surveyed after 2013 compared to those surveyed before, but there was increased report of rushed negotiations with clients due to police presence (OR 1.73, 95% CI 1.03-2.90, p-value not reported) [57,92]. The introduction of an anti-trafficking policy in South Korea, accompanied by brothel closures, in 2010 was associated with a decrease in prevalence of gonorrhoea and antibodies to Treponema pallidum (indicating current or past infection), but also changes in the demographic profile of sex workers. Sex workers were younger in surveys conducted after the act compared to before, which may contribute to the lower prevalence of infection, although sex workers reported receiving more clients [77].

Qualitative synthesis
Included qualitative studies. From the 94 eligible papers including qualitative data, we generated 4 core analytical categories over 37 unique analyses (papers) in different legislative frameworks and geographical settings, refining these through the inclusion of a further 9 purposively sampled papers (S3 Text). Studies were undertaken in a range of legislative models: Full criminalisation models were represented in 3 papers in the US; 2 papers each in Cambodia, Kenya, Serbia, South Africa, and Sri Lanka; and 1 paper each in Australia, China, Nepal, Pakistan, Uganda, and Zimbabwe. Partial criminalisation models were represented in analyses from 5 papers in Canada and 1 paper each in Hong Kong, India, Nigeria, Thailand, and the UK. Five papers focused on Canada following the introduction of criminalisation of clients, and 1 on Sweden, where that model is in place. Regulatory models-which criminalise those non-compliant with regulations including tolerance zones, regulated venues, and/or mandatory registration at a health care facility-were represented by 2 papers each from Australia, Guatemala, Mexico, and the US and 1 from Turkey. Four papers related to New Zealand, where sex work has been decriminalised. In total, interviews with 2,199 sex workers were analysed, representing a range of sex work locations (including street settings, truck stops, brothels, massage parlours, bars, night clubs, hotels, lodges, and homes) and means of meeting clients (including organised in person, via phone or online, independently, and via third parties). Most studies focused on cis women exclusively (n = 25), with a minority including subsamples of trans women or transfeminine people (n = 18) or cis men (n = 9). Just 2 papers focused exclusively on the experiences of trans sex workers, and 1 on male sex workers. Ten studies included interviews with other actors associated with sex work, including clients, venue managers/owners, police, and outreach workers, but our analyses focused on data from sex workers themselves. Characteristics of included studies (data-rich and purposively sampled) [22,26,[34][35][36]49, are summarised in Table 3, indicating which papers were purposively selected. A list of the other papers that were identified but not included is available (S3 Text).
Core analytical categories identified include disrupted workspaces and safety strategies; institutionalised violence, coercion, and extortion, and restricted access to justice; reproduction of multiple stigmas and inequalities; and restricted access to health and social care and support (S4 Text). Illustrative quotes from the core categories are summarised in Box 1.
Core category 1: Disrupted workspaces and safety strategies. In contexts of full or partial criminalisation, laws against soliciting or communication in public places for the purpose of prostitution-and feared or actual arrest-compromised street-based sex workers' safety by rushing or displacing client screening and negotiations to secluded places, resulting in greater vulnerability to violence and theft by clients and others (Quote 1) [22,98,121,122,125,130]. For sex workers operating indoors, these laws impeded direct negotiations with clients and communication between peers about safety and sexual health [121]. This pattern persisted in contexts where clients were criminalised. Since it was in clients' and sex workers' mutual interest to avoid police detection, and because increased police presence and reduced number of clients led to the need to work longer hours [34,114], sex workers limited, rushed, or forewent usual client screening and negotiation, and were displaced to more isolated areas, increasing their exposure to violence and sexual health risks (Quotes 2, 3, 4a, and 4b) [34,114]. In Canada, cis and trans female sex workers continued to be displaced by police in areas undergoing gentrification, and, even when they were not targeted, some still experienced police presence as harassment [26,114]. Across diverse contexts, experience of possession of condoms being used as evidence of sex work, and experience of police raids where condoms had been confiscated, led to sex workers not carrying, using, or accessing condoms consistently [93,98,106,109] and venues restricting or not providing them [93,98,109,118]. In South Australia, sex workers attributed the latter to increased raids, closures, and the recent arrest of a venue owner [98].
Laws against brothel-keeping and bawdy houses left sex workers in the UK [123] and Canada [102,121] having to choose between working safely with other sex workers and/or third parties (e.g., security guards and drivers) and avoiding arrest by working in isolation (Quote 5), and deterred venue managers from providing sexual health training and supplies [93,121]. A lack of legal protection left sex workers vulnerable to exploitation by venue managers who could restrict access to information on their working and legal rights [121,123].
Anti-trafficking policies in Cambodia and attempts to 'eliminate' sex work in China resulted in police crackdowns on brothels, which displaced sex workers to unfamiliar and     Observation and interviews. Thematic analysis.
Effects of police practices following the 2010 crackdown and strategies used in response.
(Continued )    sometimes isolated locations (e.g., the street, bars, massage parlours, and private accommodations) where, working alone, they had less protection and control over negotiations with clients, lacked peer support to establish collective norms on condom use (Quote 6a and 6b), and were more vulnerable to sexual and other violence both from police and perpetrators posing as clients [110,117,118]. In Guatemala, some venue managers warned sex workers about raids, Experience and negotiation of sexwork-related stigma.
� Legislation and policing refers to at the time of the research. † Papers purposively selected to reflect populations, settings, legislative models, and/or health issues under-reflected in the synthesis.
but, in common with experiences in Sri Lanka [120], others encouraged them to provide officers free sexual services to avoid their prosecution [132]. In India, some brothel owners paid police to avoid raids, or allowed pre-selected sex workers to be arrested [99]. Police harassment, raids [35,110,120], undercover operations, entrapment, and pressure to act as informants [97,128] generated fear, anxiety, and stress, with media sometimes publicising sex workers' faces during raids [120]. Conversely, where certain indoor work places were informally approved by police in a wider landscape of criminalisation, as occurred in low-barrier housing for women in Canada, the removed threat of criminal penalties fostered venue-level safety strategies, in which sex workers could refuse unprotected sex or call the police in the event of a client becoming violent (Quote 7) [113]. Similarly, in the context of decriminalisation in New Zealand, cis female sex workers working on the street reported greater police presence contributing to their protection as well as increased time for screening clients (Quotes 8 and 9) [36,[94][95][96]. Sex workers across sectors reported being able to negotiate services more directly and refuse clients [36]. Police became more focused on sharing information with women about violent incidents or individuals, and when their presence was off-putting to clients, women could request that they left [96]. Sex workers working outdoors no longer needed to move to isolated areas [94], although they continued to experience verbal and physical abuse by passers-by [95]. Although sex worker organisations objected to mandatory condom use within this model, some sex workers felt that it helped them insist on condom use [36].
In contexts of regulation in Australia, Mexico, and the US, venue-level systems such as alarms, fixed prices, intercoms, and condom use [100,124], as well as being able to work in close proximity with other sex workers and third parties [35,100,101,124], improved control and sense of safety for those able to work in regulated venues. Yet, in the US, some women criticised such systems as a veiled means of surveillance and as protecting management and clients' interests above their own safety [100]. Across these settings, those unable to conceal venue-prohibited substance use were excluded from these premises and left as the authors note with 'no choice but to work on the streets' [124] or in the minority of venues where management overlooked these regulations [35,100,101]. In Canada, the cost of business licenses and the ineligibility of those with criminal records restricted access to and mobility between regulated venues [93,121]. In Mexico, only well-networked, resident, HIV-negative, cis female sex workers gained access to tolerance zones and regulated venues, which offered fewer physical risks than unregulated indoor and outdoor settings but were often overcrowded, making income less stable [35,101]. In Australia, Guatemala, and Mexico, the ineligibility of minors to work in regulated venues meant that they had to work on the street [35,124,126]. In Australia and Sri Lanka, sex workers operating in unregulated venues had less control over negotiations with clients, and some owners encouraged women to provide sex without a condom [124,120].
Core category 2: Institutionalised violence, coercion, and extortion, and restricted access to justice. Studies showed that policing practices in contexts of criminalisation and regulation institutionalised violence against sex workers, both directly through police inflicting physical or sexual violence or demanding fines in lieu of arrest, and indirectly by restricting access to justice and thus creating an environment of impunity for perpetrators of violence [97,102,122,125,[127][128][129][130].
Violence and abuses of power by police were reported across all genders and diverse political and economic contexts, including Cambodia, Canada, the Democratic Republic of the Congo, India, Kenya, Nepal, Nigeria, Pakistan, Serbia, South Africa, Sri Lanka, Thailand, Uganda, the US, and Zimbabwe [49,97,99,104,106,111,112,118,119,122,125,127,128]. This took the form of arbitrary arrest and detention, verbal harassment, intimidation, humiliating and derogatory treatment, extortion, forcible displacement, physical violence, gang rape, and other forms of sexual violence during raids and in police custody [49,97,99,103,104,106,111,112,118,122,127,128]. In Kenya, Mexico, Nepal, Pakistan, Serbia, Sri Lanka, and the US, sex workers experienced extortion (unofficial 'fines', payments, or bribes) or provided sexual services enforced through physical or sexual violence or under threat of detention, arrest, transfer to rehabilitation centres, or forced registration (Quotes 10 and 11) [49,101,103,110,119,122,[128][129][130], with limited or no opportunity to negotiate condom use [128]. Similar extortion and/or arbitrary fines were reported in China, India, Thailand, and Turkey (Quote 12) [99,107,110,125]. In Nepal, cis female sex workers, including those hired as peer educators, reported being arrested, beaten, and robbed by police upon being found in possession of condoms [106].
Reporting violence could result in sex workers' being further criminalised [49,97,[120][121][122]127,128]. Sex workers were reluctant to report violence and theft to the police [98,125] for fear of the following: arrest for prostitution-related activities, unrelated petty offences, or nonpayment of previous fines [97,98,116,120,124,131]; being accused of crimes they had not committed [49,103]; harsh treatment or moral judgement [97,120]; further extortion or violence [35,101,112]; disclosure in court [97]; prohibitive costs [112]; or because no action would be taken to address the crime [97,111,112,114,116]. Long-standing discrimination, and the sense that police viewed them as criminals, made sex workers doubt the police would take complaints seriously [114,115,128]. When reports were submitted to police, sex workers' accounts were dismissed as implausible, with police simultaneously blaming sex workers for the violence they had experienced [49,120,125], discrediting them as victims (Quote 13) [97,103,121,127,128], and sometimes further attacking or extorting them [49]. Cis and trans women in Canada and the US reported police questioning whether it is possible for a sex worker to be raped [97,128]. (Quote 14). Similarly, in Kenya, one cis woman reported being asked by an officer 'how a prostitute like me could be raped as I was used to all sizes', discouraging her from going to the police in future: 'Never will I again go to report a case' [127]. This produces an environment of impunity, where further violence, extortion, and theft from police and others operate unchecked [98,103,120,121,125,127], perceived to be a major contributor in normalising violence against sex workers [26,125].
Reluctance to report violence occurred even in contexts where the purchase but not the sale of sex was criminalised, due to fears that information about where sex work takes place could be used to target clients and harass sex workers (Quote 15) [34,114]. While some cis and trans women in Canada felt that police were now more concerned for their safety [26,114], others felt that officers continued to view them as 'trash', blame them for the violence they experienced, and deprioritise their safety [97], despite laws and police guidelines constructing them as victims [26]. In contexts of regulation, registered sex workers in Guatemala viewed their health cards (recording compliance with mandatory testing) as protective against police and immigration harassment [126,132], and registered sex workers in Mexico had better access to police protection but rarely reported violence [35]. In Senegal, registered workers still experienced being disbelieved when reporting physical or economic violence to police and so were reluctant to report it as a result (Quote 16) [105]. Concerns about being exposed to family and friends were paramount [35,105] and deterred some from registering [126]. Relationships with police were precarious, conditional on maintaining registered status, which can vary each month depending on compliance with mandatory screening requirements-with those whose registration has (temporarily) lapsed facing arrest, detention, and/or fines (Quote 17) [35,126]. Those who were not registered were afraid they would be sent to jail or fined for working illegally, or for active drug use [35], and were more heavily targeted by police for fines, arrest, detention, extortion, and sometimes sexual violence [35,101,124]. In India, marked reductions in police raids and violence were achieved through a peer-based intervention that facilitated access to justice and challenged power relations between sex workers and police, although some officers cited lengthy procedures to dissuade reporting [99]. In Canada, Mexico, Thailand, and the US, some sex workers described certain officers' concern for their safety and support, but such concern was the exception [35,97,103,125].
Since decriminalisation in New Zealand, sex workers describe having better relationships with the police, and greater access to justice which-despite some prevailing mistrust in police -makes them feel safer and more confident with clients [36,95,96] and more deserving of respect (Quote 18) [36]. The removal of threat of arrest-which reduced police power and afforded sex workers rights-gave sex workers, and particularly young people [95], greater confidence to report violent incidents, exploitation by managers, and disputes with clients [36,96]. However, some officers treated disputes with clients as breaches of contract rather than crimes [96]. While there were still some reports of abuses of police power, there were also examples of offending officers being prosecuted as a result, helping to challenge environments of impunity [36,94,96].
Core category 3: Reproduction of multiple stigmas and inequalities. Findings show that repressive police treatment reinforced inequalities and entrenched marginalisation of sex workers, as well as creating disparities within sex-working communities, with police targeting specific settings or populations. In the context of full criminalisation in Sri Lanka, sex workers reported experiencing harsher punishment than their clients or managers: both sex workers and clients might be fined, but clients were not arrested or charged in the way that sex workers were [49], nor were managers of flats arrested during police raids [120]. Across settings, arrests, fines, extortion, and theft by police particularly targeted street-based sex workers [101,103,120,128], resulting in loss of income and increased economic vulnerabilities (Quote 19) [49,99,103,118,125,127,129,130]. Findings from Canada, Sri Lanka, and the US also show how criminalisation and police enforcement restricted freedom of movement, as sex workers were targeted arbitrarily by police during and outside of sex work hours and environments [49,97,103,120,128], and outed as sex workers by officers [97].
Studies showed how police targeting and mistreatment of sex workers, and inaccessibility to justice, reproduced inequalities and discrimination against sexual and gender minorities [26,49,116,119,127,129,130], people who use drugs [22,103,128,133], women, people of colour, and migrants [26,34,97,98,128,129,132]. In Serbia, Roma trans sex workers were treated with 'contempt' both by police enacting 'extreme violence' against them and by clients who expected cis women (Quote 20) [129]. In sub-Saharan Africa, male and trans sex workers described the 'double stigma' they faced, which could result in humiliation, ostracisation, eviction, and lack of access to micro-finance schemes, and this was worse in settings where homosexuality is also criminalised (Quote 21) [127]. In Sri Lanka, where both sex work and homosexuality are criminalised, trans sex workers were less likely to be charged than cis women but they experienced extensive extortion, humiliation, false accusations of crime, and verbal, physical, and sexual violence by officers targeting their gender expression (Quote 22) [49,120]. Similar experiences were reported among feminine-presenting male and trans sex workers in Pakistan and among trans women and sex workers of colour in Canada and the US [26,119,128]. In Canada, trans sex workers attributed officers' lack of response to their reports of violence to the stigma and discrimination surrounding their gender, sex work, and drug use, reinforcing their self-blame [116].
Long-standing racial discrimination and community mistrust reinforced black and indigenous sex workers' doubts that the police would take their complaints of violence seriously [26,128], and drug use was used to undermine sex workers' testimony against their attackers (Quote 23) [128]. In the US, one woman described what police said to an ex-boyfriend who had beaten her up: 'You can't go hitting her, even though I'd hit her for being a junkie' [128].
In Canada, a cis female independent sex worker described a police officer calling her 'just a fat. . .native whore' [97], while some white male independent sex workers attributed their lack of police attention to their race and social and economic privilege [102].
In criminalised and regulated settings, the precarious legal status of undocumented or unregistered migrant sex workers was used by clients [127] and venue owners [132] to refuse payment, and by landlords to charge inflated rents for substandard rooms [107]. Migrant sex workers did not report violence and other crimes to the police due to fear of deportation [35,131,132] or language barriers [98]. In Guatemala, police officers sometimes rounded up migrant sex workers whether or not they were registered [126], and in Turkey, police targeted 'foreign-looking' women presumed to be migrant sex workers [107]. In Sweden, immigration legislation and anti-trafficking policies have been used to deport migrant sex workers, despite their characterisation in national prostitution law as victims of violence, as a way of reducing sex work [34].
Core category 4: Restricted access to health and social care and support. Research demonstrates how criminalisation and police enforcement restrict sex workers' access to health and social care. In Cambodia and various sub-Saharan African countries, crackdowns on brothels have reduced access to health services by disrupting peer networks and displacing sex workers from usual places of work, making it difficult for outreach services to find people, and hindering collective organisation (Quote 24) [118,127]. In China, sex workers were reluctant to accept condoms from health services after police crackdowns, for fear of their use as evidence [110]. In Sweden, the mandate to reduce sex work acted as a barrier to services, as sex workers' access became conditional on leaving the sex trade and conforming to a victim discourse, and health services no longer distributed condoms through outreach [34]. Based on ethnographic observations, authors noted multiple difficulties experienced by sex workers as a result of laws against renting property used for sex work, including problems with eviction as well as with immigration, child custody, and tax authorities [34]. In Canada, some sex workers had received referrals from supportive police to health, counselling, and legal aid services [97], but indoor venue managers remained reluctant to allow outreach visits for fear of prosecution, restricting access to sexual and broader healthcare-particularly disadvantaging migrant sex workers who relied on outreach [93]. Trans sex workers in Canada [116] and sex workers of all genders in South Australia [98] were fearful of accessing clinics [116], sex-worker-led outreach services, and peer information and resources [98], for fear of being reported to the police.
Studies showed how registration and mandatory testing necessitated more frequent contact with healthcare systems [100,108,115,132] and were viewed positively by authors in Nevada, US, as a way of maintaining a low level of STIs [100] and by some sex workers as a form of self-responsibility for health [108,126]. However, in Guatemala the decision to comply with testing requirements was mostly motivated by fear of police harassment and detention rather than health considerations [126,132]. In Turkey, unregistered migrant sex workers were forcibly tested upon arrest [107], and in Australia, some sex workers experienced judgement and were refused testing by health professionals [108]. Mandatory testing of sex workers is considered a rights violation by the UN Refugee Agency and the Joint United Nations Programme on HIV/AIDS that can create barriers to sex workers accessing voluntary services and can facilitate discrimination against sex workers living with HIV. In Nevada, sex workers who test HIV positive can face up to 10 years in prison if they are found selling sex in a licensed or an unlicensed environment [100]. Discrimination against sex workers in general was often reinforced, and mandatory registration was not only time-consuming but could lead to public disclosure of sex work, adversely affecting individuals' credit rating and ability to obtain a loan (Quotes 25-28) [108,115,127]. Regulation systems also restricted migrants' access to sexual health services [35], and those with undocumented status in Turkey lacked broader access to healthcare and banking services, leaving them vulnerable to theft [107]. In Canada, sex workers' fear of becoming known to the authorities left them dependent on cash and unable to access loans [107,121].

Box 1. Quotes
Core category 1: Disrupted work spaces and safety strategies Quote 1: 'They couldn't have designed a law better to make it less safe, even if they sat for years! It's like you have to hide out, you can't talk to a guy, and there's no discussion about what you're willing to do and for how much. The negotiation has to take place afterwards, which is always so much scarier. And you're in a parking lot somewhere with some dude and all of a sudden he decides he doesn't want to pay that, or pay anything at all and what are you going to do about it? So, yeah, it's designed to set it up to be dangerous. I don't think it was the original intention, but that's what it does.'-cis woman, sector and age unspecified, Canada [121] Quote 2: 'Twenty seconds, one minute, two minutes, you have to decide if you should go into this person's car. . .now I guess if I'm standing there, and the guy, he will be really scared to pick me up, and he will wave with his hand "Come here, we can go here round the corner, and make up the arrangement", and that would be much more dangerous.'cis woman, internet escort/street, age unspecified, Canada [34] Quote 3: 'While they're going around chasing johns away from pulling up beside you, I have to stay out for longer.. . .Whereas if we weren't harassed we would be able to be more choosy as to where we get in, who we get in with you know what I mean? Because of being so cold and being harassed I got into a car where I normally wouldn't have. The guy didn't look at my face right away. And I just hopped in cause I was cold and tired of standing out there. And you know, he put something to my throat. And I had to do it for nothing. Whereas I woulda made sure he looked at me, if I hadn't been waiting out there so long.'-cis woman, street, age unspecified, Canada [114] Quote 4a: 'Sometimes the guy will drive up and just sort of wave or point to go down the alley or something like that somewhere else where he can pick me up. [How does that affect your safety?] You never know who it is, right? And you can't really see his face, can't really see anything they could have a gun in their hand or. You know what I mean they could be a little drunk or something if you can't really see them very clearly, you know. And you don't you can't say hi or whatever before you get in. You have to just hurry up before the cops come.'-cis woman, street, age unspecified, Canada [114] Quote 4b: 'Clients are worried about police. To avoid police they wanna move to a different area. I don't want to go out of my zone right.. . .Once you get out there, like you know their turf so it's harder for me cause it's their comfort zone so they act differently, you know what I mean. Yeah it never ends up good'-cis woman, street, age unspecified, Canada [114] Quote 5: 'The ideal situation is where you. . .have a separate premises where you can work from, and share those premises. . .Because then you've got companionship, added security, there's someone to interact with. Because of the legal situation you have to be very, very careful. Because obviously it's running a brothel, which has. . .really dangerous consequences these days.'-cis man, independent, age unspecified, UK [123] Quote 6a: 'In the past, we just stay in the brothel and no one dared to hurt us or beat us because we are there in the brothel. But now [since police crackdowns] we cannot know where they take us to. Such as taking us to Prek Ho [a village 15 km from Phnom Penh] and hurt us. We don't know in advance. There is no one to control us. So it is not safe for us.'-cis woman, formerly brothel-based, age 26 years, Cambodia [118] Quote 6b: 'Now some clients may force us not to use condoms but when we lived in the brothel we had more rights than clients and they dared not to force us because they come into our house.'-cis woman, formerly brothel-based, age 22 years, Cambodia [118] Quote 7: 'One of the staff caught one [a violent client]. He was a visitor in the house, and he came in as a date, and they called the police, and he got arrested.'-cis woman, indoor, age unspecified, Canada [113] Quote 8: 'And the police weren't around as much (before decriminalization). But when it got legalised the police were everywhere. We always have police coming up and down the street every night, and we'd even have them coming over to make sure that we were all right and making sure our minders, that we've got minders and that they were taking registration plates and the identity of the clients. So it was, it changed the whole street, it's changed everything.'-cis woman, street, age unspecified, New Zealand [36] Quote 9: 'You stand outside the car and talk. Don't get in the car and talk-it's best to just get them to wind the window down, stand there, talk to them and judge them. Yeah.'-cis woman, street, age unspecified, New Zealand [94] Core category 2: Institutionalised violence, coercion, and extortion, and restricted access to justice Quote 10: 'There was this time when I was arrested by six policemen. They afterwards demanded sex from me. One of them threatened to stab me if I refused. I ended up having sex with all of them and the experience was so painful.'-cis man, sector unspecified, age 26 years, Kenya [127] Quote 11: 'It's really pathetic taking money from us. I don't know how they don't understand I struggled for that. I sold my body. I worked. The man, for instance, pardon me, fucked me and everything, for the money. And they take the money. Why? I don't know, but so they say it goes into some fund, what do I know?'-cis woman, street, age not specified, Serbia [129] Quote 12: 'Does the law limit how much they [police] charge [when fining sex workers]? Today, 500, tomorrow 300. Why the law does not limit. . .the charges for this amount? For gambling, 1000 charged, prostitution 500, isn't there a limit? We don't understand. I feel like the charges just depend on their [police] mood.'-cis woman, focus group, sector and age not specified, Thailand [125] Quote 13: [In a case where a participant reported being attacked by a client and the case going to court.] 'He ended up getting off even though I had photos of the bruises. This is likely related to the institutional attitude that women who sell sex deserve what they get from taking on a dangerous occupation-it's such bullshit but so common! Also, I feared prosecution myself as a prostitute so I was unable to be completely truthful in court and my abuser was let off-even with the evidence'-cis woman, independent off street, age not specified, Canada [121] Quote 14: 'The police don't look at us as victims when we're raped and when we're beaten and stuff like that. If we get into a physical altercation and we have to fight for our lives, we're most likely to be jailed because of it.'-cis woman, sector not specified, age 40 years, US [128] Quote 15: 'They come to my door and, you know, ask for my ID and so forth so it's like harassment. . .The third time it's like, "We know what you're doing, I mean, what you're about. We're going to go after your clients". . .I make a living out of this, so I was really paranoid for a very long time after.'-cis woman, internet escort, age not specified, Sweden [34] Quote 16: 'One night a client went off with a girl, and after their encounter he beat her. The next day she recognised him in the bar and told the bar owner who told her to go to the police. When she got to the police station the officers didn't believe her-they said she didn't have any proof. The police don't give us any help at all.'-cis women, working in a bar with registration, age not specified, Senegal [105] Quote 17: 'Once, I forgot to return [to the city clinic] for a health stamp. The police threatened to take me and nine other girls to jail, but they let us go with a warning and a 2,000 pesos fine [$220].'-cis woman, sector not specified, age 19 years, Mexico [35] Quote 18: 'Well it definitely makes me feel like, if anything were to go wrong, then it's much more easier for me to get my voice heard. And I also, I also feel like it's some kind of hope that there's slowly going to be more tolerance perhaps of you know, what it is to be a sex worker. And it affects my work, I think. . .when I'm in a room with a client. . .I feel like I am deserving of more respect because I'm not doing something that's illegal. So I guess it gives me a lot more confidence with a client because, you know, I'm doing something that's legal, and there's no way that they can, you know, dispute that. And you know, I feel like if I'm in a room with a client, then it's safer, because, you know, maybe if it wasn't legal, then, you know, he could use that against me or threaten me with something, or you know. But now that it's legal, they can't do that.'-cis woman, sector and age not specified, New Zealand [36] Core category 3: Reproduction of multiple stigmas and inequalities Quote 20: 'They [police] started going wild, only on us transvestites. They let the girls go. They just pick us up, and go to the woods, and go wild on us. . .First, they beat us in the woods, and then they take us to the station. And then they tell us at the station "Hey, freshen up," and they beat us up in the bathroom'-transvestite [author's term], street, age unspecified, Serbia [129] Quote 21: 'Sometimes a man will take you and after fucking, he says, "You are gay, where can you report me? I'm not paying you and you can do nothing about it."'-cis man, focus group, sector and age unspecified, Uganda [127] Quote 22: [After reporting being jailed on charges of prostitution and describing an incident with police involving forced gender behaviour] 'I'm very scared of policemen of

Discussion
We estimate that, collectively, lawful or unlawful repressive policing practices linked to sex work criminalisation (partial or full) are associated with increased risk of infection with HIV or STIs, sexual or physical violence from clients or intimate partners, and condomless sex. The qualitative synthesis clearly shows pathways through which these policing practices and health risks are associated: enacted or feared police enforcement-targeting sex workers, clients, or third parties organising sex work-displaces sex workers into isolated and dangerous work locations and disrupts risk reduction strategies, such as screening and negotiating with clients, carrying condoms, and working with others. Specific policing practices, including confiscation of condoms or needles/syringes, are associated with increased odds of HIV, STIs, and violence course.. . .They straight away tell.. . ."Go sing a song! sweep!" Talk to us like dogs.'trans woman, street, age unspecified, Sri Lanka [49] Quote 23: 'Because it wasn't a trial of rape, it was a trial of me being a heroin addict, me being on methadone. It got thrown out of court. . ..'-cis woman, street, age unspecified, Canada [22] Core category 4: Restricted access to health and social care and support Quote 24: 'Since the new law was passed, fewer women access health care and prevention services because we live at different places nowadays and NGOs could not find us. In the past, women live in one place at the brothel. We also want to contact NGOs but we don't know the location of the NGOs. . .So we could not access to prevention services. . .Since the brothel was closed I have never contacted it again.'-cis woman, brothel, age 22 years, Cambodia [118] Quote 25: 'Because the policemen crack down often we cannot earn money. We are sleepless, so we sleep at day time, so I am lazy to go to check my health. I have no feeling to go.'-cis woman, brothel, age 22 years, Cambodia [118] Quote 26: 'I think every month is stupid. It has to be every three months at least. Because it's a pain for owners, it's a pain for girls, for everyone, because like you can't go to your family doctor and say, "Listen I need a certificate". You have to go to a sexual health clinic and wait all day to see a doctor.'-cis woman, brothel and escort, age unspecified, Australia [108] Quote 27: '[For] any insurance one of the questions is, "Have you been a prostitute?" Whatever, now if they pulled your health records and they saw how many tests you'd had, you can't lie about that one and I think it should be totally illegal [insurance companies asking about sex work]. And I would like to see them do a bit of a study on girls in the sex industry who have worked, that aren't on drugs and how many diseases they actually have, to see if this kind of discrimination is warranted, because it's not.'-cis woman, sector and age unspecified, US [108] Quote 28: 'I worked in a legal prostitution setting in Nevada. I did that for a couple of weeks to see what it was like. The amount of controls and the lack of freedom was horrendous. You know, I don't want someone else telling me how to work. And I don't think it is necessary really. Yeah, I think decriminalization gives us the most freedom.'cis woman, independent in-call and out-call, age 39 years, US [115] by a range of actors. Repressive police practices frequently constitute basic violations of human rights, including unlawful arrest and detention, extortion, physical and sexual violence by law enforcement, lack of recourse to justice, and forced HIV testing-violations inextricably linked to increased unprotected sex, transmission of HIV and STIs, increased violence from all actors, and poorer access to health services [3,29,134]. The qualitative synthesis shows how violence and stigma against sex workers are institutionalised, legitimised, and rendered invisible [26,35] in contexts of any criminalisation and regulation [26,35], as sex workers across settings consistently report being further criminalised, blamed, or ignored when they report crimes against them. This structural, symbolic, and everyday violence fosters climates of impunity and under-reporting, and failure to recognise sex workers as citizens deserving protection, care, and support [26]. Targeting and exclusion of the most marginalised sex workers reinforces and obscures the injustices they face.
Our findings build on previous reviews documenting the extent to which and how social and structural factors influence sex workers' safety and vulnerability to HIV. They do so by showing how these factors interplay with criminalisation to further marginalise sex workers and deprive them of civil, labour, and social rights [134][135][136][137]. Fear of prosecution and moral judgement, due to laws against homosexuality and transgenderism [138] and drug use [135], and, in the case of migrant workers [139], fear of deportation, further reduce willingness to report violence and exploitation to the police. Other evidence has shown how evictions based on landlords' fears of brothel-keeping charges increase vulnerability to homelessness for sex workers and their families, while arrest and criminal records or simply being identified as a sex worker can lead to sex workers' children being placed in institutional care [135,140].
Despite including search terms relating to broader health outcomes, the majority of epidemiological literature focused on sexual health outcomes and, in more recent evidence, violence. We found few studies that focused on emotional health, but these show detrimental associations with repressive policing and criminalisation. Qualitative and quantitative studies demonstrate that police enforcement and its threat is a major source of anxiety [103,141], whereas working in indoor, decriminalised environments is associated with improved mental health outcomes [32,142]. A recent critical literature review demonstrates that criminalisation, stigma, poor working conditions, isolation from peer and social networks, and financial insecurity have negative repercussions for sex workers' mental health [13]. Only 1 quantitative study reported on the associations between policing and violence from intimate or other partners, and further research is needed to understand the mechanisms of this relationship [58]. It is clear that criminalisation and stigma interact to reproduce sex workers' exposure to physical and sexual violence, and limit possibilities to resist or challenge it, and interventions are urgently needed to address violence against sex workers from all perpetrators. Successful sexworker-led approaches to improving access to justice and challenging institutional stigma in South India offer important examples of what can be achieved with sustained funding and support [99].
Findings clearly show that criminally enforced regulatory models create major disparities within sex worker communities, possibly enabling access to safer conditions for some but excluding the large majority who remain under a system of criminalisation, including trans women, cis men, people who use drugs, migrant populations, and often sex workers operating in outdoor environments, who are at increased risk of HIV in many settings [81,90,126]. In contexts of mandatory HIV testing following arrest, fear of enforcement can hinder voluntary uptake of HIV testing and interventions [71,80], showing how this punitive approach to public health ultimately reduces access to health services. More recent research from Senegal has shown that while registration was associated with better physical health, the stigma attached to being registered has a detrimental effect on well-being; only a minority of sex workers are registered, and those who test HIV positive are excluded [143]. As the qualitative synthesis demonstrates, in New Zealand, following decriminalisation, sex workers reported being better able to refuse clients and insist on condom use, amid improved relationships with police and managers [36,144,145]. Other research in this setting indicates that decriminalisation has the potential not only to reduce discrimination, denials of justice, denigration, and verbal abuse but also to improve sex workers' emotional well-being [31]. This concords with existing modelling data that suggest a positive effect of decriminalisation on incidence of HIV [2].
We were unable to examine the effects of different legislative models in the quantitative synthesis due to limited data, particularly for the models of decriminalisation and the criminalisation of the purchase of sex. Evidence included in our qualitative synthesis clearly shows that criminalisation of clients does not facilitate access to services, nor minimise violence. This is supported by the epidemiological evidence from Vancouver that showed that sex workers who were stopped, searched, or arrested were at increased risk of client violence despite the introduction of more severe laws against the purchase of sex introduced in 2014 (alongside fewer sanctions for sex workers working together and modelled on the Swedish law) [57]. In addition, the practice of rushing negotiations due to police presence increased and was associated with increased client-perpetrated violence [92]. Findings from our qualitative synthesis suggest that enforcement strategies that seek to reduce the numbers of sex workers [118] or clients [114] are unlikely to achieve these effects, since the economic needs of sex workers remain unchanged, resulting in sex workers having to work longer hours, accept greater risks, and deprioritise health. There is no reliable evidence from Sweden that the numbers of sex workers have decreased since the law changed in 1999 [34].

Limitations
There are a number of limitations to this review. Findings from our pooled meta-analyses examining condom use and violence were limited by high heterogeneity, although effect estimates remained consistent across sensitivity analyses, suggesting we can be confident in their robustness. By limiting the search to literature written in English, Russian, and Spanish, we may have missed key studies. There was a lack of comparable quantitative data on outcomes such as access to services, drug-related harms, and emotional ill health, which precluded the use of meta-analysis. Similarly, few qualitative studies explored the emotional health effects of criminalisation and enforcement, and its effects on access to health and broader services received less attention relative to safety and health risks, within the rich body of evidence reviewed. Methodologically, some studies did not provide sufficient detail on sampling and analysis methods, and few included reflexive discussions on the position of the researcher. Although a growing number involve sex workers as researchers or advisors, few included discussion of the challenges and benefits of participatory approaches. We found few eligible studies that included trans female or cis male sex workers, who experience particular inequalities in relation to HIV, access to services, and-as the qualitative synthesis shows-police targeting and violence, limiting our ability to generalise findings to these populations. It is also possible that some studies may not have differentiated between trans women and cis men [146], or between cis and trans participants within samples of female and male sex workers, and few disaggregated experiences or outcomes by gender. This is an important area of future research given the specific vulnerabilities experienced by these populations, in contexts where gender and sexual minorities are criminalised, inadequately protected against hate crimes, and, in the case of trans people, not legally recognised. There is particular need for research with trans women, who experience intense violence, discrimination, and exclusion from education and employment, and whose health needs have been obscured by their conflation with 'men who have sex with men' [146].
Our review focuses on the implementation of enforcement practices linked to 5 broad legislative models. While it is clear that sex work laws and enforcement practices are inextricably integrated and it is key to link practice to legal frameworks to inform policy-making and advocacy, our findings reinforce previous evidence [37,38] that shows wide variation in how laws are enforced, which vary with sex work setting [126], visibility of sex work, sex workers' and managers' relationships with individual officers [99,101], and political and media attention [110,125], or arbitrarily by city [121]. We report on recent and past history of arrest or prison based on the information available to us, but few studies reported whether the arrest was related to sex work, was related to another offence, or had to do with social, gender, or racial profiling. Assessing the extent to which the enforcement practice was lawful or unlawful is beyond the scope of this review, but in some cases unlawful activities are clearly evidenced (e.g., police violence) while in others they are less visible or evidenced. This limits our ability to assess the specific contribution of sex work penalties to the health and safety of sex workers, relative to the use of other penalties and abuses of police powers against sex workers in contexts of criminalisation. Lack of clarity on the lawfulness of police enforcement practices also reflects the difficulties in measuring stigma and its interaction with criminalisation, and the need for mixed-methods approaches to unpack these complexities in context. We found few data on the interplay between criminalisation, collective organisation, and health outcomes. Evidence from India has shown how tackling social injustice and mistreatment by the police as part of a sex-worker-led HIV prevention intervention has resulted in fewer arrests, more explanation of reasons for arrest, and fairer treatment by the police, as well as decreased violence against sex workers [84,99]. However, most evaluations of community-led health interventions have been limited to HIV prevention and have been implemented in India, Dominican Republic, and Brazil [147,148]. Although there are numerous examples of active sex worker organisations advocating for sex worker rights and evidence-based policy internationally, as well as developing guidelines for rights-based HIV programming with, for, and by sex workers [149], the voices of sex workers continue to be dismissed and silenced in policy debates in many settings as well as in the design and evaluation of public health interventions.

Conclusion
The public health evidence clearly shows the harms associated with all forms of sex work criminalisation, including regulatory systems, which effectively leave the most marginalised, and typically the majority of, sex workers outside of the law. These legislative models deprioritise sex workers' safety, health, and rights and hinder access to due process of law. The evidence available suggests that decriminalisation can improve relationships between sex workers and the police, increasing ability to report incidences of violence and facilitate access to services [36,95,96]. Considering these findings within a human rights framework, they highlight the urgency of reforming policies and laws shown to increase health harms and act as barriers to the realisation of health, removing laws and enforcement against sex workers and clients, and building in health and safety protections [134]. It is clear that while legislative change is key, it is not enough on its own. Law reform needs to be accompanied by policies and political commitment to reducing structural inequalities, stigma, and exclusion-including introducing anti-discrimination and hate crime laws that protect sex workers and sexual, gender, racial, and ethnic minorities. Mixed-methods, interdisciplinary, and participatory research is needed to document the context-specific ways in which criminalisation or decriminalisation interacts with other structural factors and policies related to stigma, poverty, migration, housing, and sex worker collective organising, to inform locally relevant interventions alongside legal reform. This research must go alongside efforts to examine concerns surrounding decriminalisation of sex work within institutions and communities, which influence policy and practice, and sex workers must be involved in decision-making over any such research and reforms [121,150]. Opponents of decriminalisation of sex work often voice concerns that decriminalisation normalises violence and gender inequalities, but what is clear from our review is that criminalisation does just this by restricting sex workers' access to justice and reinforcing the marginalisation of already-marginalised women and sexual and gender minorities. The recognition of sex work as an occupation is an important step towards conferring social, labour, and civil rights on all sex workers, and this must be accompanied by concerted efforts to challenge and redress cultures of discrimination and violence against people who sell sex. While such reforms and related institutional shifts are likely to be achieved only in the long term, immediate interventions are needed to support sex workers, including the funding and scale-up of specialist and sex-worker-led services that can address the multiple and linked health and social care needs that sex workers may face.