Article Text

Original article
The descriptive epidemiology of male sex workers in Pakistan: a biological and behavioural examination
  1. Souradet Y Shaw1,
  2. Faran Emmanuel2,
  3. Alix Adrien3,
  4. Merydth Holte-Mckenzie2,
  5. Chris P Archibald4,
  6. Paul Sandstrom4,
  7. James F Blanchard1
  1. 1Centre for Global Public Health, Department of Community Health Sciences, University of Manitoba, Winnipeg, Canada
  2. 2Canada–Pakistan HIV/AIDS Surveillance Project, Islamabad, Pakistan
  3. 3Department of Epidemiology, Biostatistics and Occupational Health, McGill University, Montreal, Canada
  4. 4Public Health Agency of Canada, Ottawa, Canada
  1. Correspondence to Mr Souradet Y Shaw, Centre for Global Public Health, Department of Community Health Sciences, University of Manitoba, R070 Med Rehab Building, 771 McDermot Avenue, Winnipeg R3E 0T6, Canada; umshaw{at}


Objectives There is a dearth of published information on the characteristics of sex workers in Pakistan. This study sought to characterise and compare hijra and non-hijra sex workers from eight large cities in Pakistan.

Design χ2 and Kruskal–Wallis tests, and multivariable logistic regression were used where appropriate.

Methods Study respondents were described on demographic, sex-work, and risk behaviour variables using a cross-sectional integrated biological and behavioural quantitative survey.

Results A total of 3350 respondents were surveyed, of which 2694 were included in the study. The average age of respondents was 24.1 years (SD 6.3), and the average duration of sex work was 7.5 years (SD 5.9). Respondents averaged 30.9 (SD 2.7) paid receptive anal sex acts in the month prior to their interview, while 21.5% reported using a condom during their last occurrence of paid anal sex. Of those surveyed, HIV prevalence was 5.4 per 1000; notably, no HIV-positive respondents reported any injection drug use. Finally, intercity heterogeneity was observed on demographic, sex work and risk behaviour characteristics, with almost all characteristics differing at the p<0.01 level.

Conclusions Low levels of education, high volume of sex acts and suboptimal condom use makes for a potentially volatile situation. Information provided by this study can play an important role in designing effective prevention programmes, particularly in capturing heterogeneity in sex work between cities, and as evidence is accumulating that a shift in epidemic phase, as well as affected populations is occurring in Pakistan.

  • Sex workers
  • prevention of sexual transmission
  • sexual behaviour
  • risk factors
  • surveillance
  • epidemiology
  • HIV
  • risk behaviours
  • risk profiles
  • sexual practices

Statistics from


Although the first case of HIV in Pakistan was reported in the mid-1980s,1 HIV was thought to have left Pakistan relatively unscathed, with low-level transmission occurring in isolated subgroups.2–5 However, outbreaks were observed, starting in 2003–2004,2 foreshadowing the potential for a more generalised epidemic. Currently, between 85 000 and 120 000 individuals are estimated to be living with HIV.6 7 A number of factors put Pakistan at high risk for transmission into the general population.3 4 8 In addition to political and cultural circumstances exacerbating the spread of HIV and other pathogens,8–10 researchers have noted the potentially explosive mixture of high rates of utilisation of sex workers by migrant workers,4 11–14 high prevalence of needle-sharing among injection drug users (IDU),5 8 11 15 sexual illiteracy regarding HIV16 and lack of condom availability and use.12 16

With some exceptions,2 4 14 17 18 little is known about the epidemiology of HIV in commercial sex workers (CSW) and, in particular, male sex workers (MSW). Broadly speaking, MSW in Pakistan are males who are paid to have receptive/penetrative anal or oral sex with other men.19 Another important CSW group are hijra sex workers (HSW), these being traditional sex workers who are biologically male but function in society as if female.17 Considering the strong transmission risk posed by receptive anal sex,20 preventive efforts would benefit from knowledge about these risk groups. The natural course of HIV observed in other regions has transmission progress from IDU, to sex workers, and then to their partners.21 22 Thus, there is an urgency for characterising phase-specific HIV transmission dynamics among CSW, in order to better inform the priorities and strategies of programming. This study sought to describe and compare characteristics of HSW and MSW from eight cities in Pakistan, as well as examine the predictors of condom use among these two groups between cities.


Recruitment methods

Data were from a larger cross-sectional behavioural and biological survey of key subpopulations of HIV risk. MSW and HSW from eight cities (Karachi, Lahore, Faisalabad, Multan, Hyderabad, Sukkur, Peshawar and Quetta) were recruited and interviewed individually by trained peer workers. Data were collected during 2005–2006. Sampling strategy has been described previously.19 23 Briefly, respondent-driven sampling was used to create a representative sample of MSW in each city, while HSW were recruited through a multistage cluster sampling technique.

MSW were defined as ‘any male who undertakes sexual activity with a man in return for money or other financial benefits.’ HSW were defined similarly, with ‘transvestite/transsexual’ used in place of ‘male.’ Exclusion criteria included being under 15 years or over 45 years; unwilling to provide informed consent; judged to be incapable of understanding the information provided; and participating in the current round of data gathering.

Behavioural data were gathered using a structured questionnaire. Informed consent was obtained prior to conducting interviews. Biological data were gathered using the capillary ‘dried blood spot’ methodology. A debriefing session was held with participants after interview completion. The institutional review board of Health Canada approved the study.

In total, 3350 respondents were interviewed. Of these, 656 (19.6%) were missing information on the number of clients with whom they had anal sex. Thus, analyses were restricted to respondents with complete information on this variable (n=2694). For analyses examining predictors of condom use, we included only those individuals reporting anal sex with a paying client in the month prior to interview date (n=2608).


Respondents were compared on demographic, sex-work-related risk behaviour and HIV/STI knowledge variables. Age, sexual debut, total and sex work income, education and marital status were used as demographic variables. Duration in sex work, HSW versus MSW status, solicitation method and number of paying (and non-paying) clients in the month previous were used as sex-work-related variables. Receptive and penetrative sex were not differentiated in the survey item. Condom use at last anal sex with a paying client, and whether the respondent had, in the previous 6 months: used alcohol or charas (a type of hashish), used injection drugs, had sex with a known IDU, or sold blood were used as risk behaviour variables. Lastly, respondents were asked if they had heard of HIV, if condoms could protect against HIV and whether diseases could be transmitted through sexual contact.

Statistical analysis

Four sets of analyses were performed. First, individuals were compared using ANOVA, Kruskal–Wallis and χ2 tests, where appropriate. Second, HSW and MSW were compared on the same variables as the first analysis, with the Mann–Whitney U replacing the Kruskal–Wallis test. Results from continuity-corrected χ2 tests are reported, where appropriate. For the third set of analyses, multivariable logistic regression was used to determine predictors of condom use at last anal sex. Given the smaller sample sizes in some cities, HSW and MSW were combined, and selected predictors used, with age at sexual debut kept as a continuous variable. For both the second and third analyses, separate analyses were performed for each city. Results of intercity analyses are presented as supplementary tables in this article's electronic version. All regression analyses were unweighted. Finally, HIV-positive and HIV-negative respondents were compared on the same set of variables as in the first analysis. Because of the small number of HIV-positive respondents (n=18), HSW and MSW were combined. The Fisher exact test was used to test for bivariate associations between HIV status and selected variables. Analyses were performed using SPSS 15.0 (SPSS, Chicago, Illinois) and STATA 9.2 (Stata, College Station, Texas).24 25


Intercity differences were significant at the p<0.05 level for all variables (table 1). The mean age was 24.1 years (SD 6.3), with sex work duration averaging 7.5 (SD 5.9) years. The mean age of sexual debut was 15.4 (SD 2.7) years. Education level was low, with 40.9% reporting no formal education. Most respondents were not currently married, with Peshawar having the highest proportion of married men (24.8%). Approximately 41% identified as HSW. Respondents averaged anal sex with 30.9 (SD 2.7) paying customers in a month. Condom use at last paid anal sex was generally very low, at 21.5%.

Table 1

Comparison of sex workers in Pakistan, by city (N=2694)*

Alcohol/charas was used by about half of respondents (45.1%). Reported injection drug use was 5.3%, with Hyderabad being highest (9.4%). Eighteen respondents were HIV-positive, for a prevalence of 5.4 per 1000 in the entire sample (18/3350). Prevalence was 6.7 per 1000 among those without missing responses. The majority of HIV-positive individuals (56%, 10/18) were from Karachi.


Table 2 reports comparisons between HSW and MSW. MSW were significantly younger (p<0.001) and were significantly older at sexual debut (p<0.001). HSW were more likely to be uneducated (p<0.001) and were less likely to report condom use at last anal sex (p<0.001). HIV prevalence for HSW was 0.6%, while for MSW it was 0.4% (p=0.34). As shown in the supplementary tables A and B, compared with MSW from the same city, condom use among HSW was lower in Hyderabad (8.1% vs 28.6%, p<0.001) and Multan (8.9% vs 27.1%, p<0.001) but higher in Sukkur (38.1% vs 11.0%, p<0.001). HSW in Lahore were more likely to report sex with an IDU, compared with their MSW counterparts (23.2% vs 6.1%, p<0.001); HSW in Multan were more likely to report an STI in the last 6 months (13.3% vs 5.7%, p=0.020).

Table 2

Comparison of hijra and non-hijra sex workers (N=2694)*

Condom use

Table 3 reports the results from regression models examining the predictors of condom use at last anal sex, adjusted for city. All things being equal, a positive relationship between use of condom at last anal sex and increasing age of sexual debut was observed (p=0.017), with each yearly increase of sexual debut positively associated with condom use (OR: 1.1; 95% CI 1.0 to 1.1). Married sex workers more likely to report condom use (OR: 1.4; 95% CI 1.0 to 1.9; p=0.037). Greater numbers of anal sex clients were negatively associated with condom use (p<0.001). For example, compared with those reporting less than 10 clients in the last month, sex workers reporting 30 or more clients were at 0.5 times the odds (95% CI 0.3 to 0.7) of condom use. Finally, reported sex with a known IDU was positively associated with condom use (OR: 1.5; 95% CI 1.1 to 2.1, p=0.017). The results of stratified analyses can be seen in Supplementary Table C. Notably, MSW were more likely to report condom use in Multan (OR:3.6; 95% CI 1.8 to 7.1, p<0.001) but less likely in Sukkur (OR: 0.3; 95% CI 0.1 to 0.7, p<0.001).

Table 3

Adjusted odds ratios and 95% CI from multivariable logistic regression analysis, predictors of condom use at last anal sex*

HIV-positive respondents

As seen in table 4, although not statistically significant, HIV-positive individuals were older (25.6 vs 24.1), with earlier sexual debut (15.0 vs 15.4) and were involved in sex work longer (8.5 vs 7.5 years). Relative to HIV-negative respondents, positive respondents reported anal sex with more clients (50.2 (IQR: 20 to 90) vs 29.7 (IQR: 10 to 40)). Fewer HIV-positive respondents used condoms at last anal sex (16.7% vs 21.5%). Positive cases were far more likely to have used alcohol (66.7% vs 45.0%), and none reported injection drug use. HIV and STI awareness was higher among HIV-positive MSW.

Table 4

Comparison of HIV-positive and HIV-negative hijra and non-hijra sex workers on selected characteristics, Pakistan* (N=2694)


Using a cross-sectional sample from eight large Pakistani cities, we have demonstrated important demographic and risk behaviour differences between HSW and MSW respondents. We found, in a country with suboptimal condom use, that older age at sexual debut, sex with an IDU and being married were predictive of condom use. Our results suggest that HIV transmission has moved into sexual networks of sex workers, as all HIV-positive individuals in this study denied recent injection drug use.

Our study possessed a number of strengths, including the integration of biological and behavioural components. We had a relatively large sample size from several cities. This was no small task, considering the extreme marginalisation of HSW and MSW in a very socially conservative country.26 Importantly, our results are similar to those from other studies.17 18 The median age of our sample in Karachi and Lahore was similar, as was HIV prevalence to Bokhari et al., although in our study prevalence in Karachi was slightly lower but higher in Lahore.17 The proportion of respondents that injected drugs in Karachi in our study was slightly lower at 3.4% (vs 3.7%), while our findings were slightly higher in Lahore (4.9% vs 3.9%).

This study also had several limitations. First, most responses were self-reported. Second, data were cross-sectional. Third, differential recruitment was difficult to control for; this, in combination with heavy societal sanctions against men having sex with men, may have added some bias to the recruited respondents. Fourth, the questionnaire did not distinguish between penetrative and receptive anal sex. Similarly, types of HSW were not differentiated.18 26 Fifth, some variables, such as those on income, had poor response rates. Finally, because the sample was subsetted to include only those men who reported anal sex, and weights were not used in regression analyses, the generalisability of our results to the cities from which samples were drawn was limited.

This last section will discuss the importance of characterising HSW and MSW, followed by a discussion of the predictors of condom use and HIV-positive individuals. To conclude, challenges surrounding prevention and intervention programming for HSW and MSW in Pakistan will be presented.

Hijra and non-hijra

The results demonstrated that HSW were significantly older and had been involved in the sex trade for a longer period than MSW, suggestive of HSW being a distinct subgroup. Importantly, distinctiveness translated into higher levels of risk behaviours associated with pathogen transmission,18 such as decreased condom use and increased use of psychoactive substances during sex work.18 27 Unlike others, however, we did not find any evidence (at an aggregate level) of increased levels of STI or HIV infection among HSW,18 although stratified analyses revealed that HSW in Multan were more likely to be positive for STIs than MSW. Similarly, at an aggregate level, we did not observe any association between HSWs and sex with a known IDU.27 Upon stratification, however, HSW in Lahore reported substantially higher levels of sexual mixing with IDU.

Different socio-demographic characteristics and risk profiles suggest that intervention strategies specific to HSW would be of benefit,18 26 with consideration given to regional differences. For example, integrated prevention approaches among HSW and IDU may be especially useful in cities where there is evidence of overlapping sexual networks (ie, Lahore).27

Condom use

Aggregate condom use was low, consistent with previous studies.17 18 27 However, upon stratification, Quetta respondents reported over a twofold likelihood of condom use. Respondents in Quetta were least likely to be uneducated (15%) and were most likely to report that condoms protected against HIV (50%). Sex workers from Quetta were also most likely to report knowledge of government-run sexual health programmes (data not shown). Somewhat paradoxically, Peshawar respondents reported the lowest condom usage (16%) while reporting the highest rates of HIV knowledge (95%). Perhaps most importantly, Peshawar also had the highest proportion of sex workers who reported paying for a condom (vs receiving it for free) the last time they obtained a condom (66%, data not shown). When subsetting the cohort to only those reporting anal sex clients in the last month, the number of missing responses fell from 6% to 3% for this variable (data not shown).

Access and availability of condoms remain critical issues in Pakistan, where sex outside marriage is taboo.4 17 28 In a subanalysis, of those who answered, over half stated condoms were purchased at a pharmacy the last time condoms were obtained (58%, 138/236; data not shown), with less than 1% indicating they were given condoms by non-governmental organisation (NGO) workers. Considering sex workers reportedly charge less than US$1 per sex act,26 financial barriers may impede consistency. Pakistan's national government has recognised men who have sex with men (MSM) as a vulnerable population,29 with prevention and control programmes targeting MSM in Karachi, Hyderabad and Sukkur.30 However, critical gaps remain,31 as programming is mostly delivered by NGOs and dependent on international funders, leading to fragmented and incomplete coverage of vulnerable populations.28 Given the positive impact of condom use demonstrated in this population,18 priority should be given to condom distribution.

Multivariable analyses revealed that older age at sexual debut, being married and reporting sex with known IDUs were positively associated with condom use, while increased numbers of clients had the opposite effect. Early sexual debut and high-risk behaviour have been documented in other populations.32 33 Here, early sexual debut may be a marker for lower levels of education, decreased exposure to sexual health knowledge and early entry into the sex trade.26 Our results differ from that of Hawkes et al, who examined predictors of current STI and HIV infection among HSW in Pakistan.18 The authors found a positive association between current infection and older age at sexual debut, as well as a negative association with number of paying clients.18 Recognising that condom use in our study is not an exact corollary to current infection, possible reasons for the differences observed include samples being drawn from different sites, and HSW and MSW being combined in our analyses.

The positive association between marital status, sex with an IDU and condom use was somewhat surprising. Although more research is needed, the wider use of condoms as a means of contraception in the former case,18 and the relatively higher (although far from optimal) rates of condom use among IDU in the latter,17 28 34 may partially explain our observations.

HIV-positive sex workers

Injection drug use has been noted as contributing to Pakistan's initial HIV epidemic1 5 8 11 15; however, our observation that no HIV-positive respondents reported recent injection drug use is consistent with HIV spreading through sexual networks in Pakistan.17 Since respondents did not know their HIV status, and given the low rates of injection drug use among sex workers,31 the sexual spread of HIV is a likely possibility. An earlier study of MSW from Karachi and Lahore hypothesised transmission of HIV through receptive anal sex with HIV-positive IDU.17 That we observed this in cities other than Karachi and Lahore, in light of evidence of risk-group mixing,31 is notable.

Programmatic challenges

In addition to difficulties commonly observed in developing countries,10 28 the challenge in designing and implementing strategies specific to MSW and HSW may be an especially difficult one. Homosexuality and sex work remain illegal in Pakistan and are highly stigmatised.26 31 35 In analysing potential HIV programmes in Pakistan, Buse et al note strategies targeting MSM were thought to hold the least amount of ‘resonance’ among stakeholders, with opposition from ideological and structural institutions an especially prominent barrier.28 Additionally, and along with stigma, HSW and MSW interventions are constrained by the lower priority placed on male sex work by international funding bodies, the heavy reliance on foreign donors and relatively lower levels of social organisation among community members in Pakistan.28

In terms of longer-term challenges, political disruptions in Pakistan (and in neighbouring Afghanistan) had a dramatic impact on early HIV spread, by disturbing the supply and quality of opium, and by diverting funds (and support) away from social programmes.5 8 9 Given the volatility in this region, although longer-term structural interventions are desirable, these interventions may not be sufficient to stem the immediate public health priorities of high-risk groups.36 37 Because of the competition for scarce human and financial resources, structural interventions may be beyond the scope and at times may be antagonistic to public health needs.16 36–38

Future directions

Large-scale mobilisation of peer-led prevention efforts targeting core groups in neighbouring India have led to reductions in projected HIV incidence and prevalence.37–40 To emulate India's success, knowledge and characterisation of core groups vital to the transmission of HIV are of critical importance.18 Thus, studies are needed to fully understand transmission risk in various areas and subpopulations. The differential associations of condom use with age, marital status and sex with an IDU should be further explored. Condom availability and use are sporadic in Pakistan19; along with information on clients and locations where sex occurs, circumstances and situations promoting condom use should be examined to inform strategic prevention and intervention initiatives.26 36 41

In conclusion, this study presented a comprehensive description of male sex work in Pakistan. Consistent with previous research in South Asia, our results imply a remarkable degree of diversity within this high-risk group. Our findings highlight the importance of recognising aggregate characteristics of sex workers, while simultaneously considering the impact of regional heterogeneity in the design and implementation of prevention and intervention programming.

Key messages

  • All HIV-positive sex workers denied recent injection drug use.

  • Hijra respondents were distinct from non-hijra respondents. Despite the higher-risk, HSW did not have higher rates of STIs and HIV, with the exception of one city.

  • Older age at sexual debut, sex with an injection drug user and being married were associated with increased condom use.

  • Although aggregate characteristics of sex workers are pivotal for programme design, consideration should also be given to regional differences.


Supplementary materials

  • Web Only Data sti.2009.041335

    Files in this Data Supplement:

  • Web Only Data sti.2009.041335

    Files in this Data Supplement:

  • Web Only Data sti.2009.041335

    Files in this Data Supplement:


  • Funding Support for this study was provided by the Canadian International Development Agency.

  • Competing interests None.

  • Patient consent Obtained.

  • Ethics approval Ethics approval was provided by Health Canada.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.