Over the past 15 years, rapid transmission of HIV has occurred among injecting drug users (IDU) in many developing and transitional countries, with HIV prevalence rising to levels of 40–90% . High HIV seroprevalence epidemics among IDU often have a strong self-perpetuating component. With large numbers of potential transmitters and large numbers of susceptible individuals (including new initiates into drug injection) even modest amounts of risk behavior can lead to unacceptably high rates of new HIV infections . In addition, because most IDU are sexually active, high HIV prevalence among IDU often leads to the transmission of HIV to non-injecting sexual partners, with the possibility that this initial sexual transmission from IDU will then lead to self-sustained heterosexual transmission in the general population .
Some interventions are known to reduce HIV risk behavior among IDU, including community outreach, needle and syringe access programmes and treatment for drug addiction . Large-scale implementation of such programmes has brought HIV epidemics under control in a number of industrialized countries, including parts of the United States , France , Spain , and Italy .
There is, however, very little information on whether the large-scale implementation of HIV prevention programmes can bring HIV epidemics under control in developing and transitional countries . This is primarily because of the limited scale of the implementation of the programmes in most of these countries  and the difficulties in utilizing HIV infection as an outcome for prevention programmes . Large-scale implementation of HIV prevention programmes requires at least modest financial and personnel resources, and IDU have little political support in many developing countries. If political leaders in developing countries are to be convinced of the value of large-scale HIV prevention programmes for IDU, it will be critical to develop evidence that such programming can bring HIV epidemics among IDU under control.
Detailed descriptions of the Cross Border Project and the research methods have been presented elsewhere , so only a brief overview will be presented here. The project focuses on HIV prevention for IDU and has been implemented in five sites in Lang Son Province, Vietnam, and four sites in Ning Ming County, Guangxi Province, China. The intervention follows a peer outreach model developed in the United States . The peer educators regularly contact other IDU in the community and provide them with information on reducing drug use and sexual risk behaviors. They distribute sterile needles and syringes, ampoules of sterile water for injection, condoms, and no-cost vouchers that can be redeemed for sterile injection equipment and condoms in participating local pharmacies. Over the course of the project, an average of 10 000–15 000 new needles/syringes have been provided per month in each country . The peer educators also collect and safely dispose of used needles/syringes directly from drug injectors at injecting sites in the community.
Project staff also work with law enforcement, government leaders and various community members to create understanding of and support for the project. The project can thus be considered a structural intervention .
With some variation in the IDU recruitment strategies, data collection methods were parallel in Ning Ming County and Lang Son Province.
To be eligible to participate in the study, a person had to have injected illicit drugs in the previous 6 months. (Heroin is the predominant drug injected in both study areas.)
In Ning Ming County, a modified ‘snowball/peer recruitment’ technique was used. The project peer educators sent letters to IDU they knew personally, inviting them to come to participate in the survey. The IDU who came to project centers for interviews were encouraged to recruit two to three additional participants. The research participants received 20 Chinese yuan (approximately US$2.50) for the interview, 5 yuan for each additional male respondent recruited, and 10 yuan for each additional woman respondent recruited.
In Vietnam, approximately half of the sample was based on lists of IDU registered with the government in each project site and the other half was based on individuals recruited from IDU present at drug injecting places mapped by the study team. Using probability proportionate to size, IDU were picked at random from the official register lists. They were contacted and then referred others until the quota for the commune was reached. For the portion of the sample selected initially at IDU injecting places, quotas were determined by probability proportionate to size based on the numbers of individuals observed at these places during the mapping phase. The interview team then visited the selected places and chose four individuals at random from among those present. The quotas for each place were then filled by the snowball method. The Vietnamese participants were paid 30 000 dong (approximately US$2) for participating.
Cross-sectional surveys were conducted at baseline (before any implementation), 6 months (while the project was still in start up), 12 months (by which time the project had reached full implementation), 18 months, 24 months and 36 months post-baseline. Individual subjects were permitted to participate in each different survey wave.
Informed consent was obtained and a structured questionnaire was administered by trained interviewers, primarily staff of the local health departments.
Participants were given pre-test counselling and post-test counselling at local health centers. Blood was drawn by trained phlebotomists from local health departments.
In China, testing was by double enzyme-linked immunosorbent assay (Vironostika HIV-uni-form; Organon, Oss, the Netherlands) with confirmation of initial HIV-positive results by Western blot methodology (Genelabs Diagnostics, Singapore). All testing was conducted at the laboratory of the Guangxi Center for HIV/AIDS Prevention and Control in Nanning. In Vietnam, testing was performed at the laboratory of the Lang Son Provincial Health Services by Serodia SFD screening test (BioRad, Marnes le Coquette, France) and double enzyme-linked immunosorbent assay (Genescreen, BioRad; Vironostika, Organon). This is the official protocol of the Ministry of Health in Vietnam.
Estimation of HIV incidence among new injectors
The questionnaire included questions on current age and age at first drug injection. We defined ‘new injectors’ in each survey wave as individuals who reported their first injection as occurring within 3 years of their current age.
We then estimated HIV incidence among new injectors using the following assumptions and calculations: (i) All new injectors were assumed to be HIV seronegative when they began injecting; (ii) The number of incident HIV cases among a group of new injectors is thus simply the number of HIV-seropositive individuals; (iii) HIV infection among the seropositive new injectors was assumed to have occurred midway between the time of first injection and the time of the interview, so that the time at risk among HIV-seropositive individuals is half the time from the first injection to the time of interview; (iv) The time at risk for HIV-seronegative new injectors is the total time from the first injection to the time of the interview. The estimated HIV incidence among new injectors in a specific survey wave is thus the number of HIV-seropositive new injectors in that wave divided by the sum of one half of the total years injecting among the HIV-seropositive plus the total years injecting of the HIV-seronegative new injectors in the survey wave.
Chi-square and chi-square tests for trends were the primary statistical tests used. Multiple logistic regression was used to assess the relationship between survey wave and being HIV seropositive while controlling for demographic characteristics. To test the significance of the trend in estimated HIV incidence, we used a generalized estimating equation with a normal distribution and identity link . In the generalized estimating equation, the incidence was the dependent variable, the project site and survey wave were independent variables, and we allowed clustering when an individual appeared in more than one survey wave. The test of significance was the z-statistic reported for the survey wave. The SAS statistical programmes (SAS Institute Inc., Cary, North Carolina, USA) were used for the analyses.
The study was reviewed and approved by the institutional review boards of the following institutions: Guangxi Center for HIV/AIDS Prevention and Control, the National AIDS Standing Bureau of Vietnam, Abt Associates Inc., and Beth Israel Medical Center.
Programme implementation began in July 2002 in Lang Son and in October 2002 in Ning Ming. This was one month after the baseline survey in Lang Son and 2 months after the baseline survey in Ning Ming. Implementation occurred in two phases in Lang Son, with an average of approximately 5000 syringes and pharmacy vouchers distributed for months 2 to 8 after baseline, and then reaching 10 000 syringes and vouchers distributed from month 9 onwards. In Ning Ming, there was a 2-month delay after the baseline survey before implementation was started, and then a rapid build up to 10 000 syringes and vouchers per month in month 4 after the baseline survey. We considered 10 000 syringes and vouchers to be ‘public health scale’ implementation in which community-level effects on HIV transmission might occur.
We have previously reported on reductions in self-reported risk behaviors and trends in HIV prevalence among IDU in the Cross Border sites through the 24-month survey wave , so that this report will focus on HIV prevalence and estimated incidence among new injectors as an outcome measure.
HIV infection among injection drug user subjects
Table 1 presents the demographic characteristics and HIV serostatus of the subjects in Ning Ming County and Lang Son Province, in addition to the 36-month percentage of men. There was a statistically significant variation in the percentages of female participants in Lang Son. This was partly the result of an increased effort to recruit women. The mean age of the subjects significantly increased in both Ning Ming County and Lang Son Province, although these were not large increases. There were significant declines in HIV prevalence in Ning Ming County and in Lang Son Province, although the Ning Ming situation might best be described as stabilization (a full analysis of the trends in HIV seroprevalence is in preparation.)
Table 2 presents the numbers of new injectors, the percentage of each survey wave that were new injectors, the HIV prevalence among new injectors, the mean years injecting among new injectors and the estimated HIV incidence among new injectors. The percentage of subjects who were new injectors declined significantly over the survey waves in both areas, although the decline was less uniform in Lang Son. There was relatively little variation in the mean years injecting in both areas, and there were statistically significant declines in HIV prevalence and estimated HIV incidence among new injectors in both areas. (The declines in prevalence and estimated incidence were also statistically significant at the 24-month survey.)
There are particularly sharp drops in the HIV prevalence and estimated HIV incidence between the 24 and 36-month surveys. Because of the modest sample sizes of new injectors in the 36-month surveys, the differences in HIV prevalence and estimated incidence among new injectors between the 24 and 36-month surveys were not statistically significant for the individual areas, but were highly significant (P < 0.01) when data from two areas were combined for comparisons between the 24 and 36-month surveys.
Figure 1 shows the estimated HIV incidence among new injectors in each study area in relation to when the project reached full implementation. There is no indication that the estimated incidence began to decline before full implementation in each study area, and the declines in incidence are in the proper temporal order for the interventions to act as causal factors.
We also conducted a multivariate logistic regression analysis of factors associated with being HIV seropositive among the new injectors. In both areas, new injectors in later survey waves were significantly less likely to be HIV seropositive after controlling for demographic characteristics. (Data not presented, available from the first author.)
Finally, there were strong similarities in the associations of project implementation and the reductions in HIV prevalence and estimated incidence among new injectors in each area. Comparisons of the pre-implementation situations (baseline and 6-month surveys) with the 24-month surveys show reductions of approximately one half, and with the 36-month surveys show reductions of approximately three-quarters for prevalence and estimated incidence among new injectors in both areas.
In the first two cross-sectional surveys in this study we observed very high HIV prevalence and very high estimated HIV incidence among IDU in Ning Ming County, Guangxi Province, China, and in Lang Son Province, Vietnam. These high levels of HIV infection are consistent with the rapid spread of HIV among IDU in many industrialized and developing countries . We also observed very substantial decreases in prevalence and estimated incidence among new injectors after the implementation of large-scale peer outreach and syringe access programmes, without any evidence of an increase in the rate of individuals initiating drug injection.
The reductions in estimated HIV incidence in Ning Ming and Lang Son are dramatic, but similar to reductions in HIV incidence associated with large-scale risk reduction/prevention programme expansion observed in other areas. For example, estimated HIV incidence among IDU in New York City fell from 13 per 100 person-years in 1978–1983  to seven per 100 person-years in 1984–1985 : HIV incidence among IDU in Amsterdam fell from 10.2 per 100 person-years in 1986 to 5.6 per 100 person-years in 1987 ; and HIV incidence among IDU in Vancouver fell from 10.3 per 100 person-years in 1997 to 2.5 per 100 person-years in 1998 .
The level of programme implementation achieved in Ning Ming and Lang Son was approximately 10 000–15 000 syringes or pharmacy vouchers in each area provided per month for an estimated 1500 IDU in each area. Full implementation corresponded to a ‘coverage’ of seven to 10 syringes per IDU per month. This compares favorably with syringe exchange coverage in the United States , and estimates of the coverage needed to control HIV epidemics among IDU .
This intervention study was not a randomized controlled trial, so it is difficult to assess the importance of potential causal factors. It is very likely that there were interactions (synergies) among potential causal factors leading to the observed reductions in prevalence and estimated incidence among new injectors. Peer education would have increased knowledge and motivation to reduce risky behaviors.
It is also highly likely that many HIV-seropositive IDU would have developed clinical AIDS and died during the period of the study. Although we do not have mortality data among HIV-positive individuals in Ning Ming, mortality was clearly high among HIV-positive individuals in Lang Son. The first case of HIV in Lang Son province was not recorded until December 1993 and rapid transmission occurred among IDU in the mid and late 1990s. By 2006, there was a cumulative total of 2393 known cases of HIV infection, of whom 1343 had developed AIDS and 1248 (52%) had died, and many of these deaths would have occurred during the period of our surveys (2002–2005) . Note also the significant decline in HIV prevalence among all subjects over the serial cross-sectional surveys in Lang Son (Table 2). A large number of deaths caused by AIDS would have both removed potential HIV transmitters from the population and further increased motivation to reduce risky behaviors.
Increased access to sterile injection equipment would then have helped translate increased motivation into effective risk reduction. The potential for synergy among interventions is the part of the rationale for including peer outreach and increased access to sterile injection equipment within ‘comprehensive’ HIV prevention programming for IDU .
In this study, interventions known to be effective for preventing HIV infection among IDU in industrialized countries were adapted to developing country sites and implemented on a public health scale with coordination across an international border. Cooperation was obtained from law enforcement to reduce barriers to practising safer injection. Implementation was followed by substantial reductions in HIV prevalence and estimated HIV incidence among new injectors in both areas. Finally, there was no indication that the project led to any increase in individuals starting to inject drugs.
There are many high prevalence HIV epidemics among IDU in developing and transitional countries . We hope that the present study can be used as a strong argument for implementing large-scale HIV prevention programmes for IDU throughout other parts of China and Vietnam and in many other areas.
Sponsorship: This paper was partly supported by an anonymous donation and National Institutes of Health grant nos.: 2R01 DA003574-21, 1R01 DA1470301 and P30 DA 11041.
Conflicts of interest: None.
1. Committee on the Prevention of HIV Infection among Injecting Drug Users in High Risk Countries. Preventing HIV infection among injecting drug users in high risk countries: an assessment of the evidence
. Washington: Institute of Medicine; 2006.
2. Stimson GV, Des Jarlais DC, Ball A, editors. Drug injecting and HIV infection: global dimensions and local responses
. London: UCL Press; 1998.
3. Saidel TJ, Des Jarlais DC, Peerapatanapokin W, Dorabjee J, Siddharth Singh S, Brown T. Potential impact of HIV among IDUs on heterosexual transmission in Asian settings: the Asian Epidemic Model. Int J Drug Policy 2003; 14:63–74.
4. Coyle S, Needle R, Normand J. Outreach-based HIV prevention for injecting drug users: a review of published outcome data. Public Health Rep 1998; 113:19–30.
5. Des Jarlais DC, Perlis T, Arasteh K, Torian LV, Beatrice S, Milliken J, et al
. HIV incidence among injection drug users in New York City, 1990 to 2002: use of serologic test algorithm to assess expansion of HIV prevention services. Am J Public Health 2005; 95:1439–1444.
6. Emmanuelli J, Desenclos J. Harm reduction interventions, behaviours and associated health outcomes in France, 1996–2003. Addiction 2005; 100:1690–1700.
7. Muga R, Sanvisens A, Bolao F, Tor J, Santesmases J, Pujol R, et al
. Significant reductions of HIV prevalence but not of hepatitis C virus infections in injection drug users from metropolitan Barcelona: 1987–2001. Drug Alcohol Depend 2006; 82:S29–S33.
8. European Monitoring Centre for Drugs and Drug Addiction. Annual report on the state of the drugs problem in the European Union and Norway
. Luxembourg: European Monitoring Centre for Drugs and Drug Addiction; 2004.
9. Burrows D. High coverage sites: HIV prevention among injecting drug users in transitional and developing countries: case studies. UNAIDS Best Practice Collection
. Geneva: UNAIDS; 2006. pp. 1–80.
10. Cohen J. HIV and heroin: a deadly international affair. Science 2003; 301:1657–1658.
11. Wiebel W. The indigenous leader outreach model: intervention manual.
In: Levin L, editor. Rockville: National Institute on Drug Abuse; 1993.
12. Hammett TM, Bartlett NA, Chen Y, Ngu D, Cuong DD, Phuong NM, et al
. Law enforcement influences on HIV prevention for injection drug users: observations from a cross-border project in China and Vietnam. Int J Drug Policy 2005; 16:235–245.
13. Des Jarlais DC. Structural interventions to reduce HIV transmission among injecting drug users. AIDS 2000; 14(Suppl 1):S41–S46.
14. Greene W. Econometric analysis
, 4th ed. Upper Saddle River: Prentice-Hall Inc.; 2000.
15. Hammett TM, Kling R, Johnston P, Liu W, Ngu D, Friedmann P, et al
. Patterns of HIV prevalence and HIV risk behaviors among injection drug users prior to and 24 months following implementation of Cross-Border HIV prevention interventions in northern Vietnam and Southern China. AIDS Educ Prev 2006; 18:97–115.
16. Des Jarlais DC, Friedman SR, Novick DM, Sotheran JL, Thomas P, Yancovitz S, et al
. HIV-1 infection among intravenous drug users in Manhattan, New York City, from 1977 through 1987. JAMA 1989; 261:1008–1012.
17. Des Jarlais DC, Friedman SR, Marmor M, Cohen H, Mildvan D, Yancovitz S, et al
. Development of AIDS, HIV seroconversion, and potential co-factors for T4 cell loss in a cohort of intravenous drug users. AIDS 1987; 1:105–111.
18. van Haastrecht HJ, van den Hoek JA, Bardoux C, Leentvaar-Kuypers A, Coutinho RA. The course of the HIV epidemic among intravenous drug users in Amsterdam, the Netherlands. Am J Public Health 1991; 81:59–62.
19. Tyndall M, Johnston C, Craib K, Li K, Spittal P, O'Shaughnessy M, et al
. HIV incidence and mortality among injection drug users in Vancouver – 1996–2000. Can J Infect Dis 2001; 12:69B.
20. Vickerman P, Hickman M, Rhodes T, Watts C. Model projections on the required coverage of syringe distribution to prevent HIV epidemics among injecting drug users. J Acquir Immune Defic Syndr 2006; 42:355–361.
21. Hanh H. Welcome speech. Joint Meeting on Project “Vietnam-China Cross Border HIV/AIDS Prevention”
. Lang Son, Vietnam; 2006.
22. UNAIDS/WHO. AIDS epidemic update: December 2006
. Geneva: Joint United Nations Programme on HIV/AIDS (UNAIDS) and World Health Organization (WHO); 2006.