As in many parts of the world, especially in Eastern Europe and Asia, the HIV epidemics in Vietnam and in Southern China are driven by injection drug use. There have been various small-scale interventions for injection drug users (IDUs) launched in Vietnam and China since the late 1990s and the governments of both countries have been generally supportive of harm reduction approaches such as peer outreach, needle/syringe programs (NSP), and opioid substitution treatment (OST), the effectiveness of which have been well-established by research from around the world (Institute of Medicine, 2007). Nevertheless, Vietnam has not yet scaled up NSP and most such interventions in the country are still supported by international donors, a situation that cannot be sustained as donor support is phased out when Vietnam achieves middle-income status. China does have a substantial government-supported network of needle/syringe programs but overall coverage remains low.
Since 2002, the Cross-Border Project has implemented high-coverage HIV prevention interventions , for IDUs, including peer outreach and NSP in Lang Son Province, Vietnam and Ning Ming County (Guangxi), China, and since 2003 in Ha Giang Province, Vietnam. While few such harm reduction interventions in Vietnam and China have been systematically evaluated, our Cross-Border Project has a strong evaluation component whose results add to the already strong evidence base.
This report is addressed to HIV/AIDS policy makers, program implementers and researchers.
Figure 1 shows the geographic setting and sites of the Cross-Border Project.
Figure 1: Cross-Border Project Sites
The interventions involve peer outreach and needle/syringe distribution, directly and through redemption of pharmacy vouchers. This was the first time identical interventions have been implemented on both sides of an international border.
The project’s peer educators are recruited and supervised by local health departments. The interventions directly reached 60%-70% of IDUs in the sites but are considered structural in that even greater numbers were positively affected and the government and general community were very supportive. Evaluation of the Cross-Border interventions employs serial cross-sectional surveys of IDUs (behavioral interviews and HIV testing) with sampling by modified “snowball” method and 250-300 respondents per wave, resulting in one of the longest data series on IDUs in Asia. This is an appropriate method for structural interventions. Primary outcomes were HIV risk behaviors, HIV prevalence and incidence. We used 2 methods to estimate HIV incidence: 1) HIV+ new injectors (LTE 3 years) from prevalence data, assuming HIV- at start of injection history and seroconversion halfway from start of injection to survey date; 2) BED testing of HIV+ samples (only 36-84 month available for Lang Son).
Analysis of demographic characteristics of respondents in the IDU surveys (data not shown) revealed that in all sites they were overwhelmingly male, largely between 21-30 years old (although there was some tendency for the samples to age over time), and about two-thirds were single. Ethnic minorities contributed substantial percentages to these IDU samples: usually 70%-80% in Ning Ming, 45%-60% in Lang Son, and 40%-60% in Ha Giang.
We found significant declines to low levels in drug-related risk behaviors (receptive and distributive sharing of needles/syringes, sharing of drug solution, and sharing of any injection equipment). Figure 2 shows reductions in HIV prevalence among IDUs (Lang Son: 46%-23% [p< 0.001], Ning Ming:17%-11%[p=0.003], and Ha Giang : 51%-15% [p.0.001]).
Figure 2: HIV Prevalence Trends
New injector analysis revealed significant declines to low levels in HIV incidence through 36-48 month follow up surveys, but then some apparent rebound. BED-based estimates revealed significant reductions in incidence through 84 months. 95% confidence intervals were narrower for the BED-based estimates than for the new injector-based estimates. Figure 3 shows the incidence trends by the 2 estimation methods.
The consistent downward trend in behavioral and biological outcome measures suggests a positive effect of this structural intervention. The apparent rebound in new-injector based incidence estimates in later surveys may reflect infections through sexual contact prior to initiating injection. BED-based estimates may exaggerate incidence because of false positive results among individuals with long-term or late stage HIV disease but surrogate survey responses on duration of infection appear to confirm the BED results in our surveys. In addition, the 95% confidence intervals are wider for the new injector-based than for the BED-based incidence estimates. The apparent increase in the BED-based incidence estimate for Ha Giang from 60 months to 72 months is within the 95% confidence band and remains lower than the new-injector based estimates.
In sum, the combined trends from the 2 incidence estimation methods show sharp declines in incidence to low levels, indicating that the Cross-Border interventions have played an important role in bringing the HIV epidemics among IDUs under control in our sites and offer a model of HIV prevention for IDUs that is worthy of large-scale replication.