Until 2003 there was little evidence of a widespread HIV epidemic among injecting drug users (IDUs) in Pakistan. Studies of IDUs recruited in harm reduction programmes indicated a prevalence of <1% among a sample of IDUs in Karachi (n = 161) and no cases among IDUs recruited in Lahore (n = 255) and Quetta (n = 96).1 2 The same studies reported an extremely high prevalence of hepatitis C virus (HCV) of about 88.5%, suggesting that injecting risk behaviours conducive to transmission of HIV are occurring. More recent reports from second generation surveillance studies suggest that outbreaks of HIV among IDUs have occurred at a city level countrywide, increasing between 2004 and 2006 and reaching a prevalence of 30% in Karachi and Hyderabad, 13% in Faisalabad, 51% in Sargodha and 10% in Peshawar.3

Evidence suggests a recent increase in the injection of pharmaceutical opiates, and associated injection risk behaviours has occurred in preference to inhaling heroin following the Afghanistan war as a result of declining supplies and increasing cost of good quality heroin combined with increasing availability of low-cost injectable opiates at chemists.4^–6 Behavioural surveillance suggest that drug users in Pakistan tend to be male and have many vulnerable characteristics alongside their drug use, such as low levels of education, income and high levels of homelessness.2 5 7 8 Behavioural surveillance also points to potential transmission routes of HIV, HCV and sexually transmitted infection (STIs) from drug users to other populations. For example, the donation of blood to donor banks is often used as a method of income generation for many drug users, and there is some evidence of sexual networks between drug users and male and female non-drug-using sex partners presenting transmission routes should HIV enter the IDU populations.2 8^–10

We undertook a survey of IDUs to measure the prevalence of HIV and HCV and associated risk behaviours and the prevalence of Neisseria gonorrhoeae, Chlamydia trachomatis, Treponema pallidum and herpes simplex virus 2 (HSV2) among IDUs in Rawalpindi and Abbottabad. The findings presented in this paper compare injecting and sexual risk behaviours of IDUs between the two cities and report on risk factors associated with HCV and HIV with a view to making recommendations for interventions to reduce injecting and sexual risk behaviours among IDUs in Pakistan. The findings of the study were also used to parameterise the model reported in the paper by Vickerman et al which appears elsewhere in this supplement.11

METHODS

A detailed description of the recruitment process, collection of clinical samples and the statistical analysis used in the conceptual framework is included in the paper by Hawkes et al which appears elsewhere in this supplement.12 Between August and September 2007 we conducted a cross-sectional survey of IDUs recruited from non-treatment settings in Rawalpindi and Abbottabad. Rawalpindi is an industrial city (population 1.8 million) located in Punjab near Pakistan’s capital of Islamabad and home to the Pakistan military headquarters. Abbottabad (population 140 000) is in North West Frontier Province at 1250 m above sea level and is renowned for its education institutions and tourist-driven economy.

Sampling employed a “respondent-driven sampling” (RDS) approach.13 14 Incomplete collection of the RDS data prevented the sample proportions from being adjusted for network sizes and recruitment biases to obtain population estimates of the IDU characteristics. Since an RDS approach was used, data on numbers of participants who refused to participate were not recorded as recruitment occurred through the widespread distribution of coupons by participants. Attempts to maximise the representativeness of the sample were made by recruiting respondents from multiple sites around the city and identifying a diverse selection of initial recruits.

The inclusion criteria were injection in the last 4 weeks, consent to clinical examination and collection of clinical samples. Interviews took place at fixed sites located in a rented house (Rawalpindi) and a community education centre (Abbottabad) using hand-held palm pilots (Hewlett Packard) administered by peer interviewers or people with extensive experience of working with IDUs.

In brief, clinical samples (20 ml urine, 5 ml venous blood and finger prick blood for point-of-care syphilis testing) were collected from all participants. Determine syphilis testing (Abbott Diagnostics) was carried out on site, and those who were positive were offered immediate treatment with benzathine penicillin. All participants were offered presumptive treatment for gonorrhoea and chlamydial infection. Urine samples were tested for N gonorrhoeae and C trachomatis infections with Amplicor PCR (Roche Diagnostics, Mannheim, Germany). Blood samples were tested for HSV2 (HerpeSelect 2 ELISA IgG, Focus Diagnostics, California, USA; Quest Diagnostics) HIV (AsSYM, Abbott Diagnostics, Wiesbaden, Germany (screening); Vironostika HIV Uni-Form II plus O EIA, Biomerieux, Boxtel, The Netherlands (for confirmation), quantitative Rapid Plasma Reagin (SYPHILIS (SYP-RPR), Randox, Co Antrim, UK) and T pallidum haemagglutination assay (TPHA) (SYPHILIS (SYP-TPHA), Randox), HCV version 3.0 (AXSYM, Abbott Diagnostics). The RPR test was performed in the laboratory on samples from all participants and RPR positive sera were titrated (1:2–1:32).

Risk behaviours and characteristics were compared by city. Pearson’s χ² test was used for categorical variables and t tests with equal variance for continuous variables. Associations between antibodies to HCV and HIV (the outcomes of interest) and covariables were explored univariably and by multiple logistic regression for both cities using a conceptual framework approach.15 Variables were classified into five groups: demographic characteristics, history of drug use, injecting risk behaviours, sexual risk behaviours and environmental indicators. Analyses were conducted using Stata 10 (Stata Corp, College Station, Texas, USA). Interaction terms between variables in the final model and city were investigated to assess whether predictors for the outcome variables differed by city. If the p value from the likelihood ratio test for the interaction term was <0.01, separate models for each city were explored.

RESULTS

A total of 302 IDUs were recruited in Rawalpindi and 102 in Abbottabad. HIV prevalence was 2.6% (95% CI 0.83% to 4.5%) in Rawalpindi and no HIV cases were found in Abbottabad. The prevalence of HCV was higher in Rawalpindi (17.3% (95% CI 13.0% to 21.6%)) than in Abbottabad (8% (95% CI 2.6% to 13.4%)). The prevalence of antibodies to HSV2 was no different in Abbottabad and Rawalpindi (6% (95% CI 0.1% to 10.7%) vs 11% (95% CI 7.4% to 14.5%)). The prevalence of syphilis was 7.6% (95% CI 4.6% to 10.6%) in Rawalpindi and 3.9% (95% CI 0.01% to 7.7%) in Abbottabad and the prevalence of N gonorrhoeae was 1.3% (95% CI 0.02% to 2.6%) and of C trachomatis was 0.7% (95% CI −0.2% to 1.6%) in Rawalpindi. These results are summarised in table 1.

Comparison of risk behaviours by city

Table 2 summarises the characteristics of IDUs by city. The mean age of the respondents was 32 years in Rawalpindi and 30 years in Abbottabad (p = 0.04) and almost all (98%) were male (data not shown). Proportionally more IDUs in Rawalpindi reported sleeping on the street in the previous 4 weeks (21%) than in their own or parents’ accommodation (67%) or other accommodation (12%, defined as with their sex partners other than their wives, at work or at a shrine). In comparison, in Abbottabad only 11% of IDUs reported sleeping on the street, 75% in their own or parents’ accommodation and 14% in other accommodation (p = 0.08).

The mean duration of injection among IDUs was 6.0 years in Rawalpindi and 7.6 years in Abbottabad (p = 0.05). Almost all participants in Rawalpindi reported injecting heroin more recently, with 94% reporting injection in the last week in Rawalpindi compared with 67% in Abbottabad (p<0.001). Almost all IDUs in both cities reported their main source of new needles/syringes to be from pharmacies and shops or a health service; however, 19% of IDUs in Rawalpindi reported using informal sources (eg, finding needles/syringes or obtaining them from friends or dealers) as their main supply. A far higher proportion of participants reported ever injecting with a used needle/syringe in Rawalpindi than in Abbottabad (40% vs 14%, p<0.001). The majority of IDUs in both cities correctly answered a question on whether HIV was transmitted via needle/syringe sharing; proportionally more IDUs in Abbottabad answered this question correctly than in Rawalpindi (data not shown).

Reported cleaning of needles/syringes was more common in Rawalpindi than in Abbottabad (59% vs 29%, p<0.01), but more IDUs in Rawalpindi reported using the same needle/syringe for injecting twice or more than in Abbottabad (67% vs 31%, p<0.001). In both cities a substantial proportion reported sharpening their needles/syringes in order to use them longer, with more reporting this practice in Rawalpindi than in Abbottabad (10% vs 2%, p = 0.01). A quarter of injectors in Rawalpindi reported ever having injected with a used needle/syringe of unknown origin that they had found compared with only 7% in Abbottabad (p<0.001).

In both cities less than half of participants reported using a condom the last time they had sex, with fewer reporting condom use in Rawalpindi than in Abbottabad (45% vs 32%, p = 0.09). In both cities over half of participants reported having symptoms of an STI in the previous year; this was higher in Rawalpindi than in Abbottabad (68% vs 55%, p = 0.07). In both cities almost a quarter of participants reported having sex with a female sex worker in the previous year (p = 0.52). More IDUs in Rawalpindi reported having sex with a transgendered person in the previous year than in Abbottabad (14% vs 2%, p<0.001).

A far greater proportion of IDUs in Abbottabad than in Rawalpindi reported ever having been in prison (40% vs 14%, p<0.001). Similar proportions in both cities reported a history of drug treatment (approximately 22%, p = 0.15). A small proportion of IDUs in both cities reported ever being tested for HCV (5%) or HIV (3%, data not shown).

Risk factors

No evidence of interactions was found between HCV, city and other covariables, suggesting that predictors for anti-HCV did not differ across the two cities. We therefore present the adjusted model for both cities combined for HCV and in Rawalpindi only for anti-HIV as no cases of HIV were reported in Abbottabad (table 3).

Risk associated with anti-HCV

In the univariate analysis, increased risk of anti-HCV was associated with reporting the main source of new needles/syringes to be informal via friends or found (OR 2.2, 95% CI 1.13 to 4.34 ) or via health services/hospital (OR 2.2, 95% CI 0.75 to 6.26) than through pharmacies. IDUs who reported injecting with a used needle/syringe found on the street had twice the odds of testing positive for HCV (OR 2.1, 95% CI 1.16 to 3.9) as those who reported experience of drug treatment (OR 1.9, 95% CI 0.04 to 1.03). Decreased odds of anti-HCV was associated with living in Abbottabad (OR 0.4) and those who reported having symptoms of an STI in the last year (OR 0.6).

Following adjustment, only four variables remained associated with testing positive to anti-HCV: IDUs living in Abbottabad had reduced odds of anti-HCV (OR 0.4, 95% CI 0.14 to 0.97) compared with those who lived in Rawalpindi, as did those who reported symptoms of an STI in the last year (OR 0.4, 95% CI 0.18 to 0.74). IDUs reporting their main source of needles/syringes to be from a health worker or a hospital (OR 3.2, 95% CI 0.88 to 11.6) or reported using informal supplies for obtaining needles/syringes (OR 2.8, 95% CI 1.25 to 6.01) had approximately three times the odds of anti-HCV compared with IDUs who reported using pharmacies and shops. IDUs reporting a history of drug treatment had almost four times the odds of anti-HCV (OR 3.7, 95% CI 1.66 to 8.28). Neither age nor duration of injecting were associated with odds of HCV.

Risk associated with anti-HIV

Only one risk factor was associated with anti-HIV among IDUs in Rawalpindi in both the univariate and multivariate analysis. IDUs who reported having symptoms of an STI in the last year had decreased odds of anti-HIV compared with those reporting no symptoms (OR 0.2, 95% CI 0.03 to 0.94).

DISCUSSION

Overall, a low prevalence of HIV was found in Rawalpindi and no cases among IDUs in Abbottabad. The prevalence of HCV was also found to be low in both cities relative to the prevalence reported in other cities; its prevalence was higher in Rawalpindi and increased injecting risk behaviours were reported by IDUs there. There was no association between HCV and duration of injecting or age in either city, or with HIV (Rawalpindi only). This suggests that transmission may have occurred recently. The prevalence of syphilis was relatively high in both cities, pointing to the possible risk of sexual transmission of HIV among IDUs. This is supported by the behavioural data which indicate reported condom use at last sex to be low and highlights the sexual links between IDUs and transgendered population, particularly in Rawalpindi.

The prevalence of HIV among IDUs in Rawalpindi was consistent with findings from other studies conducted in the city between 2004 and 2006 which indicated a prevalence of <0.5%.3 The prevalence of HCV found in the two cities was lower than that reported in other cities; the prevalence of HCV in Abbottabad is more comparable to population estimates from cities such as Lahore where the HCV prevalence was found to be 6.9% in the general population and 88% among IDUs.4

The fact that neither age nor duration of injecting was associated with HIV or HCV may point to recent infection and explain the current low prevalence, particularly in Rawalpindi where IDUs report injecting for a shorter duration. We found increased odds of HCV among IDUs associated with homelessness. Only a small proportion of our sample in Abbottabad were homeless, which may contribute to the current low prevalence of HCV in that city compared with Rawalpindi.2 It is possible that the lower prevalence of HCV was an artefact of sampling bias and the use of RDS leading to tapping into a network of IDUs in which HCV prevalence is low. A major limitation of the study is the lack of complete RDS data collected during sampling, which limited the ability to assess the sampling biases that may have entered the recruitment process. Furthermore, social desirability bias through the use of self-reported data may have led to under-reporting of some risk behaviours. Attempts to reduce this were made through the use of interviewers from local organisations with established links with IDUs.16 17

Injecting risk behaviours were comparable to or lower than those reported in a behavioural survey conducted among 200 IDUs between 2004 and 2005, suggesting that our findings may, if anything, underestimate the risk.18 The implementation of future community-recruited cross-sectional surveys as part of a wider surveillance programme would substantiate the validity of the observed differences in the prevalence of HIV/HCV here relative to other cities as well as corroborate the behavioural data.

Comparisons of risk behaviour across the two cities suggest that IDUs in Rawalpindi are engaging in higher risk injecting practices including injecting more frequently in the last week, injecting for less time and using heroin via injection and inhalation, further helping to explain the higher prevalence of HIV and HCV there. More potential for onward transmission of HIV/HCV was found in Rawalpindi through sexual risk behaviours where participants were more likely not to use a condom during last sexual intercourse and report having sex with a transgendered person. They were also more likely to report selling blood, suggesting more opportunities for onward transmission via blood donation.

It may be that the sociocultural context of both cities contributes to the creation of different risk environments. This has not been explored in depth in this study and further research would be needed to understand how city contexts may mediate HCV and HIV risk.19 20

In both cities the majority of participants report that, on average, they inject with the same needle/syringe twice or more, suggesting that access to clean needles/syringes is inadequate. In addition, approximately 10% of the samples in both cities report the practice of sharpening their needle to aid reuse, a practice that has been associated with injecting drugs in prisons where access to clean needles is scarce.21^–24 Research has shown that the practice of reusing needles is strongly associated with an increased risk of bacterial infections, further contributing to high morbidity in this population.22 25 Recent outbreaks of HIV observed in prisons3 alongside the practice of sharpening needles suggests the urgent need for HIV interventions in prisons, particularly in Abbottabad where 40% of the sample reported a history of being in prison.

Our findings suggest that increased odds of HCV infection were associated with using an informal source for obtaining needles/syringes, pointing to the need for formal needle/syringe provision. Counterintuitively, increased odds of HCV infection were associated with obtaining needles/syringes from a health worker or hospital compared with pharmacies. Findings from a linked qualitative study suggest that obtaining needles/syringes from health services may not constitute a harm reduction intervention that promotes safer injecting practices (see paper by Mayhew et al elsewhere in this supplement26). In practice, health workers are unlikely to provide clean needles/syringes to IDUs. There is evidence to suggest that needles/syringe may have been stolen from the hospital by injectors,26 and other evidence from Karachi indicates that hospital staff often inadvertently use unsterilised recycled needles/syringes as a result of inadequate disposal systems.27 The prevalence of HIV in Karachi was found to be 30% in one study, highlighting the potential for a similar spread of HIV in Rawalpindi should this practice of reusing needle/syringe be occurring.3 Further research on how needles/syringes are provided by healthcare workers is urgently needed, particularly in Rawalpindi.

More positively, our findings suggest some opportunities for HIV interventions such as through drug treatment services, which had been used by almost one-fifth of the sample. Moreover, there were high levels of awareness of HIV transmission routes via needle/syringe sharing among IDUs, particularly in Abbottabad. The increased odds of HCV associated with a history of drug treatment suggest that drug treatment services may be more successful at attracting high-risk IDUs, a population which our modelling studies have indicated need to be particularly targeted by interventions if HIV prevention programmes are to be successful (see paper by Vickerman et al elsewhere in this supplement11).

Harm reduction interventions and needle/syringe provision is underway in Pakistan, but coverage is still limited. Projects have been initiated by international donors and subsequently taken over by the Pakistani Government.28 Despite commitment to scale-up interventions to all major cities, success in achieving this has been patchy and regular coverage of IDUs remains low,28 although still higher than among other most at-risk populations in Pakistan. Our findings support the need for urgently expanding the coverage of harm reduction services to IDUs, including improving the access to clean needle and syringe provision, increasing the uptake of HIV and HCV testing, and promoting safer sexual practices. Such changes are vital to prevent further transmission of HIV among IDUs as well as onward sexual transmission to non-drug-using populations.