Interventions to improve health and the determinants of health among sex workers in high-income countries: a systematic review

Summary Many sex worker populations face high morbidity and mortality, but data are scarce on interventions to improve their health. We did a systematic review of health and social interventions to improve the health and wider determinants of health among adult sex workers in high-income countries. We searched MEDLINE, Embase, PsycINFO, CINAHL, the Cochrane Library, Web of Science, EthOS, OpenGrey, and Social Care Online, as well as the Global Network of Sex Work Projects and the Sex Work Research Hub for studies published between Jan 1, 2005 and Dec 16, 2021 (PROSPERO CRD42019158674). Quantitative studies reporting disaggregated data for sex workers were included and no comparators were specified. We assessed rigour using the Quality Assessment Tool for Quantitative Studies. We summarised studies using vote counting and a narrative synthesis. 20 studies were included. Most reported findings exclusively for female sex workers (n=17) and street-based sex workers (n=11). Intervention components were divided into education and empowerment (n=14), drug treatment (n=4), sexual and reproductive health care (n=7), other health care (n=5), and welfare (n=5). Interventions affected a range of mental health, physical health, and health behaviour outcomes. Multicomponent interventions and interventions that were focused on education and empowerment were of benefit. Interventions that used peer design and peer delivery were effective. An outreach or drop-in component might be beneficial in some contexts. Sex workers who were new to working in an area faced greater challenges accessing services. Data were scarce for male, transgender, and indoor-based sex workers. Co-designed and co-delivered interventions that are either multicomponent or focus on education and empowerment are likely to be effective. Policy makers and health-care providers should improve access to services for all genders of sex workers and those new to an area. Future research should develop interventions for a greater diversity of sex worker populations and for wider health and social needs.


Introduction
Sex work spans a wide range of activities, but is defined in this Review as the provision of sexual services in exchange for money or goods. Sex workers are a heterogeneous population-there is extensive variability in the structural, economic, social, and legal context in which they work and in their health and social needs. 1 Stigma and the hidden-often transient-nature of sex work restrict the availability of accurate data. 1 There are an estimated 1 million sex workers in the USA and 70 000 in the UK. 2,3 There are large research gaps in the understanding of their health needs in different settings. Street-based sex workers are highly marginalised and face disproportionate health inequities and harms related to alcohol and drug use, and sometimes HIV and sexually transmitted infections (STIs), hepatitis B, and hepatitis C. [4][5][6][7][8] Sex workers can encounter high rates of physical, verbal, and sexual violence from intimate partners, perpetrators posing as clients, and the police. 7,[9][10][11][12][13] They frequently have poor mental health, with increased rates of anxiety, depression, loneliness, posttraumatic stress disorder, self-harm, and suicide. 5,7,[14][15][16] There can be severe, complex social needs and structural determinants underlying these health issues, including homelessness or insecure housing, unemployment, adverse childhood experiences, gender and racial inequality, poverty, sex work criminalisation, and the setting of sex work. 5,7,10,14,15,17-21 However, many sex workers do not face this severe marginalisation or these adverse health outcomes and remain largely unrepre sented in academic literature. The legal context in which sex work occurs varies substantially between countries and can either exaggerate or mitigate these harms, with repressive policing practices and criminalisation worsening health outcomes. 22 Many sex workers face large barriers to accessing health and social care. 23 There are few specialist services for this community, 24 and mainstream services are often unaware of sex working and not tailored to sex workers' needs. 17,25 Sex workers are often unaware of available services, 26 and might fear legal implications from being identified as a sex worker. 5 Additionally, past experiences of judgement and stigmatisation while using services could deter them from seeking care again. 5,17 WHO guidelines state the importance of high-quality, integrated services to meet the health needs of sex workers. 27 However, there is little published evidence on effective health and social care interventions for sex workers in high-income countries. 28 There have been three previous systematic reviews, which have focused on psychological interventions for all sex workers, 29 HIV and STI behaviour change interventions for female sex workers in the USA, 10 and interventions for illicit drug use in street-working female sex workers. 30 A com prehensive understanding of interventions tailored to sex workers is Review needed. This study aimed to systematically review the evidence of interventions used to improve health and the wider determinants of health for all sex worker populations living in high-income countries.

Methods
We have adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. 31 Our review protocol was registered with PROSPERO in November, 2019 (CRD42019158674). Our team included authors with lived experience, and authors who had worked with and continue to work with sex workers, to ensure the Review's relevance and contextual insight in interpretation of the data.

Search strategy and selection criteria
We conducted a systematic literature search in six databases (MEDLINE, Embase, PsycINFO, CINAHL, the Cochrane Library, and Web of Science). We used a combination of subject headings and keyword searching related to sex work and health interventions (appendix pp 2-3). Grey literature was also searched using EthOS, OpenGrey, and Social Care Online, the Global Network of Sex Work Projects, the Sex Work Research Hub, and by contacting academic experts and people with lived experience of sex working. Further studies were identified through searching reference lists and citations of included studies. Studies were restricted to those published in English between Jan 1, 2005 and Dec 16, 2021.

Eligibility criteria
Eligibility was defined using population, intervention, control, and outcomes criteria. The included population were current sex workers, which we defined as people who had exchanged sex for money, drugs, or other goods within the past 12 months. Trafficking and indirect sex work (in which there is no physical contact of any kind with the client) were not included. We included studies with sex workers aged 18 years and older in high-income countries, as defined by The World Bank. 32 Any intervention with data specifically for sex workers was included. Studies with populations that did not entirely consist of sex workers, and for which-following contact with the authors-disaggregated sex-worker-specific data were not available, were excluded. If the majority of a study population was older than 18 years, and the data was specific to sex workers, the study was still included even if disaggregated data was not available following author contact. Any intervention that studied outcomes related to health or the wider determinants of health (eg, housing and welfare support) was included. Studies of sex work laws were excluded as these were investigated in a systematic review in 2018. 22 Control groups were not specified a priori.
The review included all quantitative study designs to summarise study effectiveness: randomised controlled trials, quasi-experimental studies (ie, uncontrolled or controlled before-and-after studies), observational studies (ie, cohort, case-control, time series, and crosssectional), and mixed-methods studies with a quantitative component.

Data extraction and quality assessment
Titles and abstracts were single-screened for inclusion by one of two reviewers (MB or BM). Remaining articles were double screened at full-text review by two independent reviewers (LCP and BM). Discrepancies were resolved through discussion.
Data extraction was done by one of three reviewers (LJ, LW, or HB) with accuracy checked by a second reviewer (LJ, KW, LW, or HB). Discrepancies were resolved through discussion or decided by a third reviewer (LCP) when they could not be resolved. A spreadsheet was used to extract a standard set of data on study and population characteristics, design, intervention, control, outcome, and results.
Rigour was assessed using the Effective Public Health Practice Project's Quality Assessment Tool for Quantitative Studies, 33 chosen due to its comprehensive assessment of both observational and experimental studies, and showed reliability and validity. 34 Criteria assessed include selection bias, study design, confounders, blinding, data collection methods, withdrawals, intervention integrity, and analyses.

Data synthesis
Due to heterogeneity in method, interventions, and outcomes, we used descriptive vote counting 35  HIV prevention resources distributed (eg, condoms, lubricant, and syringes) and referrals made to health and social support, and drug treatment services Women within the cohort who did not access the MAP van during the 18-month follow-up Over the 18-month study period, 42% of reports from sex workers stated use of the van; those using the van were more likely to be working with ten or more clients per week, working in isolated public spaces, and using the WISH drop-in centre service (linked to the van); those who used the programme were more likely to have used inpatient addiction treatment services in the past 6 months; there was no significant relationship between use of the van and accessing outpatient drug treatment; youth aged ≤24 years were significantly less likely to access the van Women not reporting exchanging sex for money or drugs at baseline 12 months after discharge from treatment, sex workers had a lower likelihood of abstinence from drugs and alcohol than nonsex workers using the service; of all women in the programme, there was a reduction of 30% in those reporting sex work, and a reduction of sex work transactions in those who continued sex work; women with a longer duration of treatment and those who received more mental health and psychological services were more likely to have stopped sex work; cessation of sex work was predictive of less frequent drug and alcohol use and fewer mental health symptoms at follow-up Weak ( Review a narrative synthesis 36 to summarise findings, following guidance from the Cochrane Handbook for Systematic Reviews of Interventions. 35 For the narrative synthesis, categories of intervention were developed based on the included papers, and interventions with multiple components were allocated to as many categories as relevant. Intervention components identified were education and empower ment, drug treatment, sexual and reproductive health care, other health care (eg, vaccination, screening, and primary care), and welfare. We summarised the papers in each intervention category according to four main areas: the nature of the interventions, outcomes reported, what was effective, and what was ineffective. We report outcomes as described in the studies but recognise that outcomes relating to cessation or reduction of sex working might not be wanted or important for many sex workers.
To quantitatively analyse results, we used vote counting, which can be used when outcomes are measured heterogeneously between studies. 35 Vote counting compares the number of studies in which a particular outcome improved with the number of studies in which that outcome did not improve, based only on the direction of effect and therefore with no measure of the magnitude of effect. All studies that measured outcomes before and after an intervention were included. For randomised controlled trials, both the intervention and control groups were included separately if enough information was available. We did this as most controls were well designed interventions that contributed important results to the Review. Intervention categories mirrored the narrative synthesis; the exception to this was that multicomponent interventions were categorised separately both to prevent double counting and because their effectiveness relies on the entirety of the intervention. Outcomes were grouped together into categories. Only outcome categories measured in two or more different interventions were included. If multiple outcomes were reported within one category for a particular intervention, only the primary outcome was used. If no primary outcome was identified and the results were not all in a single direction, the intervention was labelled as having mixed results for that outcome. No intervention had an outcome (or group of outcomes) that deteriorated within an outcome category. We display these data within a harvest plot, which provides a visual summary of the vote counting. 35 Additionally, we produced a standard binary metric (benefit or mixed results), which we used to calculate a proportion, 95% CI (binomial exact calculation), and p value (binomial probability test) to show the evidence for each intervention category's effectiveness across all outcome measures.  Most studies were from North America. 11 (55%) focused on street-based sex workers, and nearly all exclusively studied female sex workers. Eight (40%) of the inter ventions were multicomponent. 18 (90%) of the inter ventions (90%) took place in a context where sex work was fully or partly criminalised at the time of study. Interventions were primarily based in static locations, although seven (35%) studies included outreach compo nents. The most common outcomes measured related to drug use and drug harm reduction, sexual risk behaviours, and mental health and wellbeing. No harms associated with the interventions were reported. All studies presented limitations in sampling strategy. Most used convenience or snowball sampling. A few used repeated time-space sampling of mapped sex worker districts to improve systematicity. 26,38,41,42,49 Four (20%) studies were randomised controlled trials, but all had limitations including non-systematic recruitment strategies, 37,38,40,41 an absence of information on the randomisation process, 38,40 and no data on loss to follow-up. 40 Only one was reported using Consolidated Standards of Reporting Trials guidelines. 37 The harvest plot (figure 2) summarises evidence for effectiveness within each intervention category across Data are presented as n or n (%). *Legal context categorisations from Platt and colleagues (2018). 22 Full criminalisation prohibits all aspects of sex work and selling and buying sex; partial criminalisation criminalises only some aspects; in criminalisation of purchase of sex models, the sale of sex is legal but clients are criminalised; and regulatory models allow the sale of sex in some settings or conditions. Full decriminalisation removes all criminality of sex work while still prohibiting violence and coercion of sex workers. †Can be in more than one category. ‡Study categorisations used are derived from the Effective Public Health Practice Project's Quality Assessment Tool for Quantitative Studies. 33 §One study evaluated both outreach and static interventions.  44 which only included a peer-design element, all other interventions involving peers included both a design and delivery element and showed potential benefit. Three outreach interventions were included, of which one showed potential benefits across outcomes, 40 and two showed mixed results. 43,44 We analysed the number of positive outcomes (potential benefit) to the number of total outcomes reported per intervention category using the binomial exact calculation and binomial probability test (table 3). Education and empowerment and multicomponent interventions showed a greater proportion of positive outcomes than would have been expected by chance, suggesting their potential effectiveness, whereas the little evidence for drug treatment and other health-care interventions precludes clear insight.

Education and empowerment
Seven interventions (six single component, [37][38][39][40][41]44 and one multi component 47 ) focused on education and empowerment, and four multicomponent interventions had a small educational component, but did not detail what was provided. 45,46,51,52 Of the seven, three focused on street-based sex workers, 38,41,47 one on street-based and indoor-based sex workers, 44 and in the other three the authors did not state the sex worker population that the intervention was targeting. 37,39,40 Three were of moderate quality, [37][38][39] and four were of weak quality. 40,41,44,47 A few studies used health behaviour models that recognise structural and environmental vulnerabilities contributing to HIV and sexual health risk. 37,47 Structural determinants were addressed through enhancing sex worker self-efficacy and condom negotiation skills, 37,47 as well as teaching strategies to minimise risk of violence. 38,44 Several used psychological therapiesfive were individually administered, [37][38][39]41,44 and one used family therapy between mothers who were sex workers and their children. 40 Two were developed and delivered in collaboration with peer sex workers. 38,41 Key study outcomes for these interventions related to sexual risk behaviours, 37,38,41,47 drug use and drug harm reduction, 38,41,47 mental health and wellbeing, 37,40,44 use of other health-care and support services, 37,41 and criminal activity. 39 All interventions showed a level of effectiveness, but most only measured outcomes at 3 months after intervention. 37,44,47 A brief intervention that provided Review information on strategies to improve sex worker safety and reduce the risk of violence affected safety behaviours and use of relevant support programmes 12 weeks later. 44 A six-session resilience-promoting programme showed improvements in resilience, self-esteem, and condom use 3 months later. 37 A 12-session family therapy programme showed greater reductions in drug use and depressive symptoms than a psychoeducational programme with only sex workers. 40 One programme helped sex workers to develop negotiation skills with different sexual partner types alongside teaching jewellery-making skills. 47 3 months post-intervention, there were reductions in transactional and total sex partners, as well as injection and non-injection drug use. In another intervention, female sex workers in prison were provided two brief motivational interviews to help identify and problem solve their greatest concerns for post-release. 39 This intervention led to a reduction in the number of arrests in the 12-month period after release. Two studies including peer sex workers in the development and delivery of an intervention showed mixed results. 38,41 Both were randomised controlled trials with interventions showing similarly positive outcomes to control groups, which were high quality. One found that a strengths-based programme did not show additional effectiveness when incorporating a peer facilitator over a case manager alone. 38 The other study showed that a sex worker-focused HIV risk education programme, developed and delivered in collaboration with sex workers, led to a significantly greater reduction in unprotected oral sex and episodes of sexual violence than the US National Institute on Drug Abuse standard intervention at 6-month follow-up. However, other HIV risk outcomes were similar to the standard intervention. 41

Drug treatment
Drug treatment was provided by four interventions (two single component 14,43 and two multicomponent 45,48 ). Three specifically targeted sex worker populations-a drug treatment clinic for street-based female sex workers, 45 a one-off harm reduction intervention for street-based female sex workers, 43 and a one day per week clinic offering an array of primary care and harm reduction services to an unspecified sex worker population. 48 The fourth study compared outcomes between sex workers (no subpopulation identified) and non-sex workers using US-Government-funded drug treatment programmes across 71 facilities. 14 Three interventions were at static locations, 14,45,48 and all studies were low quality.
The primary care clinic studied by Stewart and colleagues 48 found that 31 (62%) of 50 women seen had opioids in their urine. Of these, nine (29%) of 31 started opioid substitution therapy (OST) and three (10%) were already in OST programmes. In the harm reduction inter vention, women were given naloxone, harm reduction advice, and self-administered tests for detecting the presence of fentanyl in drugs. 43 Fentanyl has a higher risk of overdose and death compared with heroin. 1 month after intervention, opioid and injection drug use, as well as solitary drug use, had reduced. However, fentanyl detection in drugs did not lead to changes in harm reduction behaviours for most people. The other two studies on drug treatment found a significant decrease in drug use at the end of drug programmes (one focused on heroin, 45 the other included various drugs 14 ), and a reduction in the number of women still engaging in sex work. 14, 45 Burnette and colleagues 14 found that those still involved in sex work were doing significantly less sex work than they had before. Both interventions provided physical and mental health services alongside OST, which led to improvements in mental health and wellbeing. Burnette and colleagues 14 found higher use of mental health services was associated with increased probability of cessation of sex work at follow-up, which in turn was associated with lower drug use, higher abstinence rates, and fewer mental health symptoms. 14

Sexual and reproductive health care
Two interventions provided STI screening, STI treatment, and HIV pre-exposure prophylaxis through sexual health outreach clinics in brothels, 51 and a 1 day per week, multicomponent primary care intervention for an unspecified sex worker population. 48 Two multicomponent welfare services for street-based sex workers provided free condoms and lubricants. 26,42,50 One study described the management of a syphilis outbreak in street-based sex workers in east London. 53 Through partnership with a charity providing outreach to sex workers, women with suspected syphilis were invited to the charity's drop-in centre, from where they were driven to a nearby genitourinary medicine clinic. Reproductive health-care service components included pregnancy testing, 48 contraceptive prescriptions and advice, 45,52 and post-coital contraception. 51 No details of service provision, uptake, or acceptability were provided and no study focused on reproductive health care. For each intervention category, a p value was calculated using the binomial probability test to determine the chance that the true proportion of potentially beneficial outcomes of total outcomes was 0·50. Accompanying exact 95% binomial CIs are also displayed.

Review
Studies were observational and low quality. Two provided information on STI treatment. 51,53 Sturrock and colleagues 51 invited sex workers with positive results back for treatment. 42 (62%) of 68 participants returned for their results and seven (17%) of 42 returning sex workers received treatment. In the syphilis outbreak, epi demiological treatment (ie, treatment based on probable exposure) was provided to all sex workers. 53 Most individuals declined intramuscular penicillinthe best available treatment-and many instead chose oral antibiotics, which are a suboptimal alternative. 13 (93%) of 14 sex workers were followed up.

Other health care
Other health-care interventions included a trauma-based psychoeducational therapy group for street-based sex workers, 9 a vaccination programme for multiple sex worker populations, 49 a multicomponent clinic focused on women's health for street-based female sex workers, 52 and multicomponent primary care clinics in the proximity of welfare drop-in centres. 26,48 The vaccination programme was a nationally run programme in the Netherlands. The programme provided free hepatitis B vaccinations to sex workers through local community health services, working alongside existing sex worker outreach services, and by community health-service staff periodically visiting various sex work locations, including brothels and streets over several years. All other interventions were at static locations, with the therapy group provided at both a community-based residential centre and a moderate-security prison. All studies were low quality. Several interventions offered referral to other health or social services, but no study gave information on the uptake of this offer. 26,38,40,42,50,52 The psychotherapy group participants' trauma scores decreased significantly in six of ten parameters at the end of the 12-week intervention. 9 The decrease was more profound in the prison group than in the residential centre, possibly as their baseline trauma scores were higher. Baars and colleagues 49 provided evidence of the effectiveness of the Netherlands' hepatitis B vaccination programme. Through a cross-sectional survey of 259 sex workers working in various settings across three cities, they found that 2 years after programme initiation, 205 (79%) of 259 were aware of the programme and 163 (63%) of 257 had received at least one dose-134 (82%) of 163 through the programme. Of those who started the vaccine programme, 74 (79%) of 94 received all three vaccinations. Those who had been vaccinated were more likely to have worked in an area for longer and 75% reported receiving their first vaccination at an outreach location. Wong evaluated a well-women clinic's cervical cancer screening intervention for streetbased female sex workers in Hong Kong. 52 208 (88%) of 236 tested women returned for their smear results, and 13 (45%) of 29 women with atypical smear results were uncontactable. Nine (31%) of the 29 with atypical results were given referral letters to attend a gynaecologist, but it is not known whether they were seen.

Welfare
All interventions that addressed welfare were multicomponent and focused on meeting basic needs through providing food and drink, washing facilities, clothing, and a safe space. 26,42,46,50 The intervention by Sherman and colleagues 47 was the only exception which, alongside teaching better condom negotiation skills, taught female sex workers jewellery-making skills over six sessions. These women then had the opportunity to sell their handmade items at a stand within a hospital. The intervention was designed to address structural determinants preventing these women earning a sustainable, alternative income. 3 months after com pletion, there were significant reductions in transactional and total sex partners, as well as injection and non-injection drug use. 47 Women who earned more money through market sales had a significantly decreased number of transactional sex partners at follow-up. All welfare interventions were oriented towards street-based sex workers and studies were of low quality.
Three studies focused on two linked interventions in Vancouver, BC, Canada. 26,42,50 The inter ventions, both designed for female sex workers, were the Women's Information Safe Haven (WISH) drop-in centre and a peer-led, van-based outreach programme called the Mobile Access Project (MAP). Those with greater numbers of clients and working in isolated areas were more likely to use the MAP van, 42 reflecting the outreach approach used. The studies showed that both services were associated with accessing other health servicesinpatient addiction services for the MAP van, 42 and sexual and reproductive services for WISH. 26 However, the temporality of both relationships is unclear. Both the MAP van and WISH were less likely to be used by younger sex workers compared with older sex workers. Of those who used the MAP van, 94% felt safer when the van was present, 16% recalled a time it had prevented physical assault, and 10% a time it had prevented sexual assault.

Discussion
We identified 20 studies, with intervention components divided into education and empowerment, drug treatment, sexual and reproductive health care, other health care, and welfare. 12 interventions were single component and eight were multicomponent. Considering the diversity of sex worker populations and their corresponding needs, this was a very small number of studies. There was promising evidence for interventions that focused on education and empowerment and those that were multicomponent. Sherman and colleagues' 47 jewellery skills and sexual negotiation strategy workshops were particularly innovative as a multicomponent intervention combining empowerment and a focus on the Review structural determinants of health. Evidence across studies also showed that designing and delivering interventions alongside sex workers was effective. Importantly, only six interventions used co-design or co-delivery. The harvest plot provided unclear results as to the effectiveness of outreach. However, two interventions that involved outreach, but could not be included in the plot because they were cross-sectional studies and did not follow up participants, showed evidence of possible benefit. Both the Netherlands' hepatitis B vaccination programme 49 and the management of a syphilis outbreak in east London 53 relied on collaboration with existing outreach services and showed good uptake and retention. Few interventions incorporated reproductive health care, 48 and there was no evidence for interventions treating chronic diseases. One intervention provided cervical cancer screening, but many people with atypical results could not be contacted and informed, and it is unclear whether those who were contacted were followed up. 52 Previous systematic reviews analysing health interventions for sex workers in high-income countries underscored the need for flexible services which are nonjudgemental, built on respect and trust, traumainformed, and targeted at specific sex worker needs. 10,29,30 Our Review also found outreach might be important in ensuring high levels of engagement in some contexts. However, outreach was not always enough to ensure continuity of care. Similar to other systematic reviews, 10,29 we found many studies highlighted low levels of follow-up. Three studies were exceptions to this. One involved OST, 45 possibly showing the perceived value of this intervention. The other two, 49,53 as discussed earlier, collaborated with established outreach services, which might have improved trust and provided a channel by which to follow up individuals.
To the best of our knowledge, this Review is the first compre hensive overview of evidence on sex worker interventions aiming to improve health and wider determinants of health outcomes in high-income countries. Academic databases and grey literature were searched, and both academic experts and people with lived experience of sex work were contacted to ensure we identified all relevant literature. Importantly, we have included authors with lived experience, and authors who have worked with and continue to work with sex workers, from the study's inception-to develop the search strategy, ensuring relevant grey literature channels were searched and experts in the field contacted, and to ensure findings were relevant, correctly interpreted, and presented with appropriate language and without stigma. This Review has some limitations. Where stated, the majority of interventions were either primarily or exclusively targeted at street-based sex workers, 26,38,[41][42][43][45][46][47]50,52,53 probably because they are more easily identified by service providers and researchers; are more exposed to structural determinants such as homelessness, poverty, and violence; 18,54,55 and typically have worse health outcomes. 56 Therefore, general isability of this Review's findings to other sex worker populations is limited. People engaged in street-based sex work often have a range of different health and social issues, including homelessness, 4,18 drug use, 10,14 and history of imprison ment, 10 emphasising the need for a wider inclusion health approach to service provision and research that addresses multiple, overlapping risk factors and vulnerabilities. 28 We reviewed English language studies since 2005 as a pragmatic choice and because an initial scoping search suggested most studies relevant to this Review met these criteria. We did not include qualitative studies that might provide insight into differences in results between studies. Outcomes were highly heterogeneous, often self-reported, and might not be the outcomes that are important for all sex workers. The development of a core outcome set in collaboration with sex workers would help future researchers to ensure that outcomes measured are relevant. 57 Methods used by the included studies also represent an important limitation, with only three studies 37-39 rated moderate in our quality assessment, and all other studies rated weak (table 2). One common reason for low quality was study designthe most common design was a single group, pre-post cohort study (often referred to as quasi-experimental studies). Additionally, due to the nature of recruiting marginalised populations, all studies presented limitations in sampling strategy and most used either convenience or snowball sampling. Finally, the complexity and dynamic nature of the legal sex working context in which the interventions took place could not be accounted for in the Review's findings and is likely to be an explanatory factor for study heterogeneity.
There is scarce investment both in services and research, particularly for sex workers who are not street based. However, a range of interventions are likely to be effective. Services should be developed and delivered in collaboration with sex workers. Interventions that are focused on education and empowerment or those that are multicomponent are likely to be effective, and an outreach or drop-in component could be of benefit in some contexts. 58 Future interventions should incorporate components related to chronic diseases given they are an important contributor to sex worker mortality. 58 Within the identified studies, almost all interventions were designed exclusively for female sex workers-the only exceptions being two that included transgender women sex workers, 26,42 and one that included male sex workers. 51 Sex worker services and future research should take a gender-sensitive and inclusive approach. Several studies highlighted that sex workers who were new to working in an area were less likely to access services than those who had been working in an area for longer. 26,42,49 Effective information dissemination and outreach could help ensure accessibility. Crucially, repressive policing practices and the criminalisation of www.thelancet.com/public-health Vol 8 February 2023 e153 Review sex work have already been shown to adversely affect access to health and social services and sex worker health outcomes. 22 Therefore, the effectiveness of any service will always be restricted in settings where sex work is criminalised.

Contributors
LCP and SAL conceptualised the study. LCP, MB, BM, LJ, LW, KW, and HB were involved in screening and data extraction. All authors were involved in interpretation of the findings. LJ and SAL wrote the first draft. All authors were involved in draft revisions and approving the final draft for submission. All authors had full access to all systematic review data and accept responsibility for the decision to submit for publication.

Declaration of interests
SAL and LCP are Pathway Fellows. Pathway is a charity that provides health care to homeless and inclusion health patients, including sex workers. All other authors declare no competing interests.