These data revealed that drug users in St. Petersburg who inject only stimulants and live in certain neighborhoods appear to constitute a unique population in terms of HIV exposure risk, even though their proportion in the sample is small. Almost all (97%) drug users in the cohort injected heroin (either alone or in combination with or in addition to stimulants) while only a small number injected stimulants only. Stimulant users did not differ demographically from the heroin users as a whole, but the risk behaviors associated with HIV infection did when considered in the context of neighborhood of residence. Whereas front or back-loading and sharing non-syringe equipment were significantly associated with HIV infection among heroin users (either those who injected only heroin or both heroin and stimulants), receptive syringe sharing was significant among the stimulant only users.
Studies conducted both in the former Soviet Union and elsewhere in the world have reported differences in risk behavior between stimulant injectors and opioid injections [22
]. A study of risk behaviors by type of drug used in Ukraine found front and back-loading was more common among opiate injectors while reusing a syringe was more common among stimulant users [25
]. While heroin has been associated with passivity, regular injection and decreased sexual activity, stimulant use has been associated with aggression, frequent, binge injection, needle sharing, increased sexual activity and young age [26
]. It has been suggested that contrasting physiologic responses to opiates versus stimulant drugs result in different risk profiles for HIV [11
]. For St. Petersburg, this is supported by the observation that frequent stimulant use is the primary factor associated with seroconversion [7
]. The present analysis reveals that even though stimulant users share demographic and behavioral characteristics with heroin users, their behavior distinguishes them in terms of HIV risk as ascertained by prevalence rates.
Our data are subject to limitations. First, the number of injectors who used only stimulants was quite small. Given that injection of stimulants only is unusual among drug users in Russia as a rule, increasing the sample size is unlikely to yield many additional such individuals [1
]. Second, associations between risk and prevalent HIV infection were revealed only when correlation by residence was included in the analysis, which suggests that geographic differences in risk may be as important as differences in the type of drug injected. Further research will be needed to determine if the choice of drug remains a significant factor in predicting transmission of HIV among drug injectors in St. Petersburg. Third, while our sample appears to be representative of drug users in St. Petersburg and is distributed randomly across the city [8
], the results may not be generalizable to populations outside of St. Petersburg. However, if characteristic effects and preparation processes for the different drugs explain some of our observed behavioral differences then the differences could occur among drug users in other settings. Fourth, our data analysis permitted us to identify associations that link prevalent, but not incident cases of HIV to drug type, injection risks, and geography. It must be noted however, that when we followed 520 seronegative individuals in our sample for an additional year, we found that psychostimulant use was strongly associated with incident infections, with hazard ratios of 8.1 for individuals who made three or more psychostimulant injections weekly versus those who made none [7
] and 5.5 for psychostimulant only injectors versus heroin only injectors. However, no injection practices were associated with incidence, a consequence of the low power provided by the smaller number of psychostimulant injectors in the seronegative cohort. These findings lead to the hypothesis that there is an association between receptive syringe sharing, which was more common among the psychostimulant only injectors, and HIV transmission, but more research would be needed to test this hypothesis. Finally, needle and syringe sharing is a widely recognized risk factor for parenteral infections and may be more socially unacceptable than sharing other drug preparation equipment. This could result in a socially desirable response bias leading to under-reporting of needle and syringe sharing compared with other equipment sharing behaviors. However, this under-reporting would not account for the association of stimulant injection, receptive syringe sharing, and HIV prevalence in one small subset of the population while failing to find such an association in a larger subset given the statistical power of the larger subset.
The role of geography was evident in our findings, but its exact impact was hard to determine. Since psychostimulant only injection was associated with certain city districts, with receptive syringe sharing, and with subsequent seroincidence [7
], it seems likely that the interaction of a risky injection practice with districts in which HIV prevalent cases were already clustered [8
] is sufficient to explain the increased likelihood of HIV transmission among psychostimulant injectors. The one district with both incident infections and psychostimulant injection was (and remains) a fairly typical residential district of apartment blocks connected to the rest of the city by bus, metro, and rail. Since little neighborhood ethnography has been conducted to study local variations in the drug scene across districts in St. Petersburg, it is hard to speculate on neighborhood contextual factors that might have further enhanced risk for injectors who resided there.
In conclusion, differences in drug preparation and distribution practices for opioid versus stimulant injection may account for some differences in risk and exposure to HIV and other bloodborne viruses [4
]. Some of these differences may be reflected in the spatial component of our findings – that neighborhood of residence is an important covariate when studying the relationship between HIV prevalence and risk behaviors. In designing targeted interventions, it becomes important to address both the drug type and the neighborhood differences since they result in distinct routes of infection. More generally, intervention programs to reduce HIV among this population should identify and focus on risk behaviors specific to the type of drug used and the social context in which is it used [32
]. Across the board, IDUs could profit from prevention education that emphasizes both access to clean syringes and preparing and apportioning drug with these clean syringes, but slight differences in emphasis and attention to neighborhood differences might improve the intervention impact for injectors who favor different drugs.