Our findings suggest that the legalisation of some aspects of sex work might help reduce HIV prevalence in this high-risk group, particularly in countries where the judiciary is effective and fair.

Countries that have legalised some aspects of sex work (n=17) have significantly lower HIV prevalence among sex workers than countries that criminalise all aspects of sex work (n=10; β=–2·09, 95% CI −0·80 to −3·37; p=0·003), even after controlling for the level of economic development (β=–1·86; p=0·038) and the proportion of sex workers who are injecting drug users (−1·93; p=0·026). We found that the relation between sex work policy and HIV among sex workers might be partly moderated by the effectiveness and fairness of enforcement, suggesting legalisation of some aspects of sex work could reduce HIV among sex workers to the greatest extent in countries where enforcement is fair and effective.

We estimate cross-sectional, ecological regression models with data from 27 European countries on HIV prevalence among sex workers from the European Centre for Disease Control; sex-work legislation from the US State Department's Country Reports on Human Rights Practices and country-specific legal documents; the rule of law and gross-domestic product per capita, adjusted for purchasing power, from the World Bank; and the prevalence of injecting drug use among sex workers. Although data from two countries include male sex workers, the numbers are so small that the findings here essentially pertain to prevalence in female sex workers.

Sex workers are disproportionately affected by HIV compared with the general population. Most studies of HIV risk among sex workers have focused on individual-level risk factors, with few studies assessing potential structural determinants of HIV risk. In this Article, we examine whether criminal laws around sex work are associated with HIV prevalence among female sex workers.

In this study we, for the first time to our knowledge, take advantage of the diversity of legal frameworks across European countries to test the hypothesis that legalisation of some aspects of sex work is associated with decreased HIV prevalence among sex workers when compared with countries that retain criminal laws. We also investigate whether more effective enforcement of the rule of law mediates the relation between sex work policy and HIV prevalence among sex workers.

Legal structures might only affect health outcomes if they are effectively and fairly implemented, suggesting enforcement can moderate the relation between sex work policy and HIV prevalence.As such, the rule of law establishes whether legal context affects police practices or access to services, which, in turn, can affect condom use and HIV prevalence ( figure ). Importantly, the rule of law can affect sex workers differently depending on sex work policy. In countries that criminalise sex work, we would not anticipate a clear association between the rule of law and HIV prevalence in sex workers because police activity can increase stigmatisation and isolation. Whereas in countries that legalise or decriminalise sex work, we would anticipate a negative association between the rule of law and HIV prevalence among sex workers because sex workers will receive appropriate protections from law enforcement.

Our findings will support future simulation modelling of changes to sex work policy and suggest that legalising sex work could help to reduce HIV prevalence in this high-risk group.

This is the first study to do an ecological analysis of the association between HIV risk among sex workers and legal structures that might affect this risk. We offer cross-national evidence of an association between sex work policy and HIV prevalence among sex workers. We also show that association between sex work policy and HIV among sex workers is moderated by an effective and fair legal system.

We searched PubMed for studies published in English between database inception and Sept 8, 2016, with the terms “HIV”, “sex worker*”, “Europe”, and “structural” and found 17 previous studies. Six of these were studies of specific countries (such as Germany, Armenia, or Russia), which alone are unable to illuminate cross-national patterns in HIV across sex workers. Five were review essays and programmes of action that called for such changes to occur. Two were qualitative studies that did not attempt to investigate cross-national differences in sex work policy and HIV. Two were systematic reviews and meta-analyses: one, which estimated the prevalence of HIV among sex workers worldwide, was not explicitly concerned with legal context; the other did simulation modelling of changes to legal context. The simulation modelling exercises offer an approximation of how sex work policy affects HIV but are necessarily reliant on parameter estimates from previously published studies of the effect of legal context for specific locations, which, so far, are inferred from related outcomes, such as police harassment. Two did cross-national analyses of structural risk factors but neither specifically examined legal structures. This summary of the evidence is consistent with the 2015 review by Shannon and colleagues, which draws attention to important gaps in the scientific literature. So far, studies of the relation between sex work policy and HIV prevalence have been based on country-specific analyses of simulation modelling. These studies have provided important evidence but more detailed analyses investigating structural risk factors for HIV among sex workers are needed. Specifically, previous ecological analyses do not investigate the association between legal structures and HIV risk among sex workers.

One commonly invoked structural risk factor is sex work policy, particularly the existence of laws that criminalise buying, selling, or procuring sex ( figure ).Such laws can exacerbate stigma and exclusion faced by sex workers, abrogating access to essential health services and increasing risks of exploitation and violence;the resulting precarious working environments and greater poverty increase HIV risks.Such insecurity can make it more difficult for sex workers to negotiate condom use because they might feel compelled to engage in condomless sex (because condoms can be used as evidence in judicial proceedings) or work in hidden areas away from possible police detection or harassment.Moreover, criminalisation of sex work can force sex workers to work in isolation, deprived of any protection police might offer. As such, criminalisation can infringe on the rights of sex workers, increasing their marginalisation, reducing their access to HIV services, and ultimately increasing their risk for HIV infection. For example, 2015 modelling estimatesproject that decriminalising sex work in Canada could avert 39% of infections in female sex workers over a 10 year period.

Dashed boxes contain unrepresented and unobserved factors along the pathway between sex work policy and HIV prevalence among sex workers.

Structural and environmental barriers to condom use negotiation with clients among female sex workers: implications for HIV-prevention strategies and policy.

Structural and environmental barriers to condom use negotiation with clients among female sex workers: implications for HIV-prevention strategies and policy.

The effect of structural determinants of HIV risk on sex workers is unclear, partly because most existing studies have not explicitly investigated structural drivers, instead focusing on individual-level predictors of HIV risk, such as knowledge about HIV transmission.A 2015 reviewof the global epidemiology of HIV in female sex workers concluded, “an understanding of how structural factors (eg, contextual factors external to the individual) shape HIV acquisition and transition risks has only just begun to emerge”.

Sex workers (people who exchange sex for money) are disproportionately affected by HIV compared with the general population.Increased risk occurs because sex workers have multiple sexual partners who might request condomless sex.Sex work can be accompanied by other high-risk behaviours, including injecting drug useand substance use.Sex workers are also often exposed to physical and sexual violence, commonly have limited access to (and use of) health-care services,and might face exclusion or social stigma.Although HIV prevalence among sex workers seems to be substantially higher than in the wider population in some contexts, prevalence varies greatly across countries in Europe.For example, HIV prevalence among sex workers seems to be quite low in the Czech Republic (0·1%), but high in Latvia (22%), suggesting that structural factors affect HIV prevalence among people who exchange sex for money.

The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.

Data on the prevalence of sex workers were unavailable for five of the 27 countries included in this data; additionally, data on the proportion of HIV cases provided with ART were missing for three countries. GDP=gross domestic product. PPP=purchasing power parity. IDU=injecting drug user. ART=antiretroviral therapy.

In a final stage, we test whether enforcement of sex work policy moderates the relation between laws and HIV outcomes ( figure ). To measure enforcement, we use an indicator of the rule of law taken from the World Bank's Governance Indicators,which is measured on a scale of −1·4 to 1·9 in our data (with higher scores representing a better rule of law) and captures the confidence people have in effectiveness and fairness of the judiciary, including the police force. To test the moderating effect of the rule of law on HIV risk in countries that criminalise or legalise some aspects of sex work, we stratify the analysis by sex work policy and investigate whether the rule of law is associated with reduced HIV prevalence in both sets of countries. Descriptive statistics are listed in table 2 (raw data are listed in appendix p 3 ). All models were estimated with STATA, version 13.0.

We also investigate the relation between HIV prevalence among sex workers and other forms of sex work policy. First, we compare countries that have legalised buying and selling sex but not procurement (coded as 1) with those that have legalised buying, selling, and procurement (coded as 2). Second, we compare countries implementing the Nordic model with countries that permit the market for sex (buying and selling) but not procurement, and with countries that permit both the market for sex and procurement through brothels. These tests analyse in more detail some of the variation in sex work policy across countries masked by our primary (dichotomous) indicator.

Next, we estimate adjusted models, which include each of our additional controls sequentially and independently based on previous predictors of HIV risk among sex workers identified in the scientific literature ( figure ). These predictors include gross-domestic product (GDP) per capita in 2005, adjusted for inflation and purchasing power (World Bank), and the proportion of sex workers who are also injecting drug users.

Our regression equation and all other regression models are weighted according to the sample size of the country-specific data. To reduce the possible effect of measurement error, these models place greater weight on HIV prevalence estimates coming from large samples, while recognising that such sample sizes are still potentially biased. Our assumption is that larger samples of this hard-to-reach population are more representative than smaller samples.

In this equation, “i” is the country. “HIV” is the logged prevalence of HIV among sex workers drawn from the latest available data. The data are logged to adjust for positive skew in the distribution of HIV prevalence. “Legal” is a binary exposure variable, for which countries that retain criminal laws are coded as 0, and countries that have reformed some or all aspects of sex work are coded as 1. The dichotomous measure of sex work policy is used in all regression models except where otherwise stated.

We use ecological linear regression models to test whether countries that have legalised some aspects of sex work have a decreased proportion of sex workers living with HIV compared with countries that criminalise sex work. We first present unadjusted models of the association between sex work policy and HIV prevalence; for example, with the equation:

The laws governing sex work vary across European countries ( table 1 ).We collected data on the laws governing sex work in Europe from Country Reports on Human Rights published by the US State Department, a widely used source for laws related to sex work.Where necessary, we have supplemented these data with other sources, such as country-specific statute documents ( table 1 ).Some jurisdictions, such as Lithuania, criminalise selling, buying, and procuring sex. Other countries have legalised selling sex but have criminalised purchasing sex. Sweden, Norway, and Iceland are prominent European examples of this approach (commonly known as the Nordic model), although France and Northern Ireland have recently implemented similar laws. Despite being rooted in feminist legal scholarship,this approach has been associated with similar harms to the criminalisation model, such as violence, police repression, and increased risk of HIV.A third set of countries (including the UK) have legalised the buying and selling of sex but have criminalised procurement through brothels or pimping. A final group of countries legalised most forms of organised and unorganised sex work (eg, street workers and brothels), albeit regulating the industry. In line with previous publications in this area, in this study we label sex work as illegal in countries that criminalise all aspects of sex work and as legal in countries that have legalised some aspects of sex work.Among legal countries, we also distinguish countries that have legalised procurement and from countries that have implemented the Nordic model of sex work policy (ie, legal to sell only).

No end in sight: why the ‘end demand’ movement is the wrong focus for efforts to eliminate human trafficking.

All data on the legal status of sex work come from the US State Department's Country Reports on Human Rights Practices.In some instances, these data have been supplemented with data from other sources, indicated by additional references.

Sex work can involve many forms of erotic labour but in this ecological regression analysis we focus on sex workers who exchange sex for money.We collected data on the proportion of sex workers living with HIV from the European Centre for Disease Prevention and Control (ECDC)'s latest estimates for 27 European countries ( table 1 ).These data are provided through the Dublin Declaration as part of the Global Aids Response Progress Reporting indicators, one of which is HIV prevalence. Survey strategies differ across countries, in terms of methods and sample size ( appendix p 2 ). Although data from two countries include male sex workers, the numbers are so small that the findings here essentially pertain to prevalence in female sex workers. Some high-income countries, including the Netherlands, were not included in this analysis because data were not included in the ECDC's report.

To ensure our results are not attributable to our modelling choices, we did a series of sensitivity tests: removing outliers ( appendix p 6 ); including other possible confounders, such as antiretroviral therapy coverage and the population prevalence of HIV ( appendix p 7 ); and adjusting for sample composition and data collection ( appendix p 8 ). We also re-estimated the model with a bootstrap procedure to test whether our results are attributable to any single country ( appendix p 9 ). In each case, we found that our results did not qualitatively change.

We found no bivariate correlation between HIV prevalence among sex workers and the rule of law ( appendix p 5 ); similarly, the rule of law has a non-significant but positive association with HIV prevalence among sex workers in countries where sex work is criminalised ( table 5 ). Contrastingly, the rule of law does have a significant negative association with HIV risk for sex workers in countries that have legalised some aspects of sex work ( table 5 ). These findings suggest that effective and fair enforcement mediates the association between legalisation and HIV but only in countries that legalised some aspects of sex work.

In an unadjusted model, no clear difference existed between the HIV prevalence among sex workers in countries legalising the selling but not the buying of sex (the Nordic model: Sweden and Norway; coded as 0; n=2) and the countries that legalise buying and selling sex but not procurement (coded as 1; n=14). However, when we compared Sweden and Norway with Germany (a country that also legalises procurement), we found that HIV prevalence is lower in Germany than in the Nordic countries; although again the number of observations is small.

In countries where the selling of sex is legal but brothels are not (n=16), HIV prevalence continues to be lower than in those where sex work is criminalised (n=10; table 4 ). This association is even stronger when looking at countries that legalise profiting from all forms of sex work (including brothels; n=1). However, only one country in our analytic sample (Germany) permits these forms of sex work and therefore these results should be treated with some caution.

After adjusting for GDP, HIV prevalence among sex workers remains lower in countries that legalise some aspects of sex work (β=–1·86; p=0·038; table 3 ) than in countries that do not. This result is maintained after adjusting for the prevalence of injecting drug use among sex workers, despite reducing the sample size (−1·93; p=0·026; table 3 ). Neither GDP nor the measure of the proportion of sex workers who are injecting drug users are associated with HIV prevalence after controlling for sex work policy.

In countries where sex work is partly legalised, HIV prevalence is lower than in countries where sex work is criminalised (β=–2·09, 95% CI −0·80 to −3·37; p=0·003; table 3 ). For example, mean prevalence of HIV infection among sex workers in countries where the practice is illegal was 4·02% (ten counties), but mean prevalence was only 0·50% in countries where some aspects of sex work have been legalised (17 countries).

Discussion

5 Shannon K

Strathdee SA

Goldenberg SM

et al. Global epidemiology of HIV among female sex workers: influence of structural determinants. Countries that have legalised some aspects of sex work have fewer sex workers living with HIV.In fact, countries that permit buying, selling, and procuring sex seem to have the lowest HIV prevalence among sex workers. This association is seen even when we adjust for the level of economic development and the proportion of sex workers estimated to be injecting drug users.

3 Shannon K

Montaner JS The politics and policies of HIV prevention in sex work. , 5 Shannon K

Strathdee SA

Goldenberg SM

et al. Global epidemiology of HIV among female sex workers: influence of structural determinants. 7 Rusakova M

Rakhmetova A

Strathdee SA Why are sex workers who use substances at risk for HIV?. , 12 Shannon K

Strathdee SA

Shoveller J

Rusch M

Kerr T

Tyndall MW Structural and environmental barriers to condom use negotiation with clients among female sex workers: implications for HIV-prevention strategies and policy. , 13 Strathdee SA

Crago AL

Butler J

Bekker LG

Beyrer C Dispelling myths about sex workers and HIV. , 33 Platt L

Grenfell P

Bonell C

et al. Risk of sexually transmitted infections and violence among indoor-working female sex workers in London: the effect of migration from Eastern Europe. , 34 Decker MR

Wirtz AL

Baral SD

et al. Injection drug use, sexual risk, violence and STI/HIV among Moscow female sex workers. , 35 Markosyan K

Lang DL

Darbinyan N

Diclemente RJ

Salazar LF Prevalence and correlates of inconsistent condom use among female sex workers in Armenia. 21 Republic of Latvia Regulations regarding restriction of prostitution. , 22 Outshoorn J These findings are consistent with other studies that reported that the removal of criminal laws protects sex workers, allowing them to enter the formal economy, to benefit from social insurance, and to receive protection from law enforcement.The removal of criminal laws might also allow sex workers to exit the industry more easily, allowing them to pursue other avenues of employment. Such legal configurations also increase access to condoms, reduce the risk of violence, and potentially empower women in condom negotiation.Our results are also consistent with evidence that suggests the Nordic model might not reduce HIV risk among sex workers and stresses lingering questions about the expansion of this legal approach to sex work across the world.

9 Shannon K

Csete J Violence, condom negotiation, and HIV/STI risk among sex workers. , 34 Decker MR

Wirtz AL

Baral SD

et al. Injection drug use, sexual risk, violence and STI/HIV among Moscow female sex workers. 12 Shannon K

Strathdee SA

Shoveller J

Rusch M

Kerr T

Tyndall MW Structural and environmental barriers to condom use negotiation with clients among female sex workers: implications for HIV-prevention strategies and policy. , 26 Chu S

Glass R Sex work law reform in canada: considering problems with the Nordic model. 18 Weitzer R New directions in research on prostitution. Effective and fair enforcement of these laws seems to modify the association between sex work policy and HIV prevalence, suggesting HIV prevalence among sex workers is lowest in countries that both legalise sex work and have an effective and fair judiciary. HIV risk for sex workers will be high in countries that criminalise sex work and have an ineffective and unfair judiciary because they will be marginalised within society, potentially exposed to violence from clients and police, and face the risk of prosecution.However, countries with an effective and fair rule of law in contexts where sex work is criminalised might not necessarily improve HIV risk. An effective judiciary might reduce police violence but might also increase marginalisation by increasing fear of arrest or prosecution.Among countries that legalise sex work, the relation to rule of law will be different. Legalisation of sex work, even in countries with an ineffective or unfair judiciary, might still reduce HIV risk because legalisation will enable sex workers to access necessary services, organise their work, and work in safer environments.Furthermore, countries with an effective and fair rule of law in contexts where some aspects of sex work are legalised or decriminalised might lead to even further reductions in HIV risk because of improvement in sex worker's protection against violence and access to services.

36 Pierson P Increasing returns, path dependence, and the study of politics. 5 Shannon K

Strathdee SA

Goldenberg SM

et al. Global epidemiology of HIV among female sex workers: influence of structural determinants. , 8 Pando MA

Coloccini RS

Reynaga E

et al. Violence as a barrier for HIV prevention among female sex workers in Argentina. 26 Chu S

Glass R Sex work law reform in canada: considering problems with the Nordic model. , 37 Krüsi A

Kerr T

Taylor C

Rhodes T

Shannon K ‘They won't change it back in their heads that we're trash’: the intersection of sex work-related stigma and evolving policing strategies. , 38 Weitzer R Sociology of sex work. However, the causal mechanisms linking legalisation with HIV among sex workers remain unclear because the timing and sequence of reforms to sex work policy might also shape how legalisation affects HIV prevalence.Simply because a factor is causally associated with an increase in HIV prevalence, does not mean that the removal of this factor would necessarily lead to a reduction in HIV prevalence. For example, sex workers are less likely to forego condoms if they have ever been the victim of violence, suggesting that reducing the future risk of violence might only have a limited effect on reducing HIV prevalence through increased condom usage.Moreover, laws might change statutory rights but “stigma does not necessarily erode simply because sex work is legal”.Further research is needed to understand how legalisation (and decriminalisation) might contribute toward reducing HIV prevalence among female sex workers.

2 European Centre for Disease Prevention and Control 3 Shannon K

Montaner JS The politics and policies of HIV prevention in sex work. This study has several important limitations. First, the use of population-level data can result in ecological fallacies. Ideally, we could use multilevel data to compare situations across countries, including individual-level data, but these do not currently exist. Second, and importantly, the HIV prevalence estimates are based on unrepresentative samples of hard to reach groups, creating several methodological challenges. One challenge is the accuracy of estimates of HIV prevalence among sex workers, which might vary across contexts because of differences in the willingness of sex workers to engage with researchers and the lower number of clinics that are likely to be available in countries where sex work is criminalised. Furthermore, data collection procedures differ between countries, meaning that cross-national comparisons can be unreliable. As such, although we use the best data available to test our research question, these differences between countries are not trivial and lead to significant measurement error, which can lead to conservatively biased estimates of associations and can create a downward bias in assessment of the effect of criminal provisions on HIV prevalence.In view of these limitations, our findings are suggestive and point toward the need for improved data collection and monitoring as a means of providing more accurate estimates of HIV prevalence in specific high-risk groups.

39 Wolffers I

van Beelen N Public health and the human rights of sex workers. Third, measuring sex work policy is difficult because two countries might differ in how strongly they enforce specific laws or, if they both permit sex work, because of subtle differences in how they have legalised it.In short, our binary or even tripartite distinctions might mask important variation between and within these categories. To the extent that this imprecision in the categorisation of sex work policy is randomly distributed across categories, this variation would lead to conservative estimates of the association. Fourth, the reform of sex work policy has occurred at different times in different countries and our analysis does not account for these temporal differences. Again, this limitation is likely to lead to conservative estimates because we are assuming no differences exist between exposure to treatment, even though large differences exist between countries. These differences should be studied independently in future research.

The causal effect of legalising sex work on HIV prevalence among sex workers remains an important topic of future research. Studies examining this question might be able to use natural experiment research designs that exploit policy change within a particular context to test how these policies affect health (including HIV risk) among sex workers more broadly. For example, under the assumption that appropriate data were available, the recent adoption of the Nordic model in France (2016) and Northern Ireland (2015) could be used to examine the short-term and long-term health effects of these policies.

Our research has important policy implications. The legalisation of aspects of sex work could reduce HIV prevalence among sex workers. Furthermore, our evidence suggests that the greatest gains appear when countries have legalised buying, selling, and procuring sex. However, the legalisation of sex work alone might be insufficient if these laws are not administered effectively and fairly. Crucially, legal reform might not solve these issues on its own, but our data suggest that legalisation could be one important part of a robust and effective response to HIV among sex workers.