The geographic distribution patterns of HIV-, HCV- and co-infections among drug users in a national methadone maintenance treatment program in Southwest China

Human immunodeficiency virus (HIV) infection and hepatitis C virus (HCV) infection are major public health problems worldwide. More than 34 million persons currently live with HIV/AIDS and 170 million people may be infected with HCV [1‐3]. By the end of 2011, China had an estimated 780,000 (620,000-940,000) people living with HIV/AIDS including 154,000 (146,000-162,000) AIDS cases. 28.4% of the 780,000 persons were infected from intravenous drug use [4]. In 2011, an estimated 48,000 individuals were newly infected with HIV, and in the same year 28,000 people died from AIDS [4]. As HIV and HCV share similar routes of transmission, including blood-blood contact, injected drug use, and sexual contact, co-infection with HIV and HCV is very common. Intravenous drug users (IDUs) often share contaminated needles or syringes for intravenous drug injection [1‐3, 5‐7]. It has been estimated that 25% of people infected with HIV in the United States are also infected with HCV. The reported prevalence of co-infection with HIV and HCV is above 90% among IDUs [5, 8, 9]. The accelerated liver disease found in HCV patients leads to increased morbidity and mortality in the HIV/AIDS patients. Co-infected patients bear a significant proportion of the mortality [1].

In China, there are approximately 2.4 million IDUs, the world’s largest population of IDUs. The proportion of IDUs living with HIV/AIDS is high [10]. In 1989, the first HIV outbreak in China appeared among IDUs from Southwest China. This has become one of the most severe HIV/AIDS epidemic areas in China [4, 10]. Since then, China has undergone an ever-growing increase in HIV/AIDS prevalence, initially fuelled by IDUs. In 2001, up to 66.5% of newly diagnosed HIV infections were related to drug use [11]. In order to constrain the dual epidemics of HIV/AIDS and drug use, China has adopted a methadone maintenance treatment program (MMTP) [11]. The MMTP was initiated in early 2004 as a small pilot project in just eight government-supported clinics of five provinces, and subsequently expanded into a nationwide program encompassing 738 clinics covering 27 provinces and serving some 344,254 drug users by the end of 2011 [11, 12]. Numerous studies reported the HIV and HCV infection prevalence among MMTP clients in China, showing the large variations in prevalence of HIV and HCV infection in different geographic locations. However, few studies have analyzed the geographic distribution patterns of HIV and HCV infections on a smaller scale (e.g., below county level) [12]. This type of analysis is needed to better understand behavioral and risk perception factors that may contribute to infections. Advances in geographic statistical techniques and geographical information systems (GIS) provide powerful tools that help characterize and improve our understanding of the geographic distribution of diseases [13].

This study aimed to characterize the geographic distribution patterns of HIV and HCV infections among drug users at the township level using GIS-based geographic analyses involving geographic autocorrelation analysis and geographic scan statistics. The high- or low prevalence areas of HIV-, HCV- and co-infections among drug users were identified in a Yi Autonomous Prefecture, Southwest of China, where a national MMTP has been implemented since 2004. Knowledge of geographic distribution patterns of HIV and HCV infections among drug users has critical implications for future MMTP implementation and health policies regarding both HIV and HCV prevention among drug users in China. These findings will help aid the public health focus and better guide the use of limited resources.

The Yi Autonomous Prefecture is one of the largest illicit drug production and distribution centers of China. It is located along one of the main drug trafficking routes from the ‘Golden Triangle’ to northwest and central China [20]. The Prefecture has experienced an alarming increase in HIV infections since the first HIV case was identified among drug users in M county in 1995. This HIV epidemic has spread to all 16 counties and 1 city [20]. By the end of 2011, the cumulative number of individuals infected with HIV was over 25,000 [21]. We found the prevalence of HIV-, HCV- and co-infection were correlated across townships, similar to the previous reports [22‐24]. The degree of relationship between HIV and HCV was not high. Our results from both the geographic autocorrelation analysis and the geographic scan statistical analysis showed that HIV-, HCV-, and co-infections in the Prefecture all exhibited significant geographic clustering at the township level. The geographic distribution patterns of these infections were clearly different. The areas of high HIV infection prevalence were mainly in the I, M, R and P counties where the proportion of Yi ethnic group was more than 50%. High HCV infection prevalence mainly occurred in the A, C, E, P, Q and R counties. Of these 6 counties, 3 (A, C, E) had a proportion of Yi population less than 50%. The high prevalence of co-infection with HIV and HCV mainly occurred in the P and R counties. These findings indicate that there might be some isolation of these communities, with more spread of infection within the communities than between them. The geographic heterogeneity of the two virus infections might be related to the geographic distribution of the ethnic groups involved. Our results revealed that HIV infection prevalence of the MMTP clients increased gradually, while HCV infection prevalence decreased gradually, as the proportion of Yi increased at the county level. This might be related to the special cultural traditions of the Yi ethnic group and the low proportion of the Yi injecting drug use or sharing syringes compared to the Han. This Yi ethnic minority still maintains traditional social norms and values including arranged marriage between members of the same social status and condoned causal sex. Hence, the transmission of these infections might be easier within some Yi communities than between them. Casual sex is also more frequent among Yi males than Han males, and unmarried Yi females have more frequent casual sexual behavior than married Yi females. Furthermore, condom use occurs infrequently among the Yi ethnic group [20, 25]. More than 86% of the MMTP clients were less than 40 year old, so heterosexual transmission was more frequent among the Yi ethnic group than among the Han ethnic group. We also found that the proportion of Han both injecting drug and sharing syringes was higher than that of the Yi. Although HIV and HCV share routes of transmission, HCV is most efficiently transmitted via exposure to contaminated blood. For example, the transmission efficiency of HCV has been estimated to be some 10 times greater than that of HIV for needle-stick injuries, while the sexual and perinatal transmission of HCV is inefficient [1, 9, 26, 27]. In A City with an 81.2% Han ethnic group, the HCV antibody prevalence among IDUs infected with HIV was 81.3% and the HIV/HCV prevalence ratio within the IDU population was 0.3. According to Vickerman et al. (2013) [28], the median estimated percentage of existing HIV infections due to sexual HIV transmission was only 38% in A city. In M County with a 97.1% Yi ethnic group, the HCV prevalence among IDUs infected with HIV was 57.5% and the HIV/HCV prevalence ratio within the IDU population was 1.3. Thus, the median estimated percentage of existing HIV infections due to sexual HIV transmission was 88% [28]. High HIV infection prevalence was more inclined to appear in some Yi cluster communities, while high HCV infection appeared in both Han cluster communities and Yi cluster communities.

Due to the shared routes of transmission, co-infection with HIV and HCV was very common in IDUs [1]. The HIV/AIDS epidemic in the Yi Prefecture was initially driven by transmission via injection drug use, and the majority of the early reported HIV cases occurred in young adults infected by injection drug use [20, 25]. Among the 4 counties (I, M, R and P) with a high prevalence of HIV infection, the areas of high prevalence of co-infection with HIV and HCV were the P and R counties. County I had a low prevalence of co-infection with HIV and HCV. One reason for the substantial geographic heterogeneity in co-infection with HIV and HCV may be the difference in drug use behaviors between the two regions. The proportion of IDUs among the MMTP clients was 64.5% in R County and 52.1% in P County, but only 39.4% in I county. The other reason for this heterogeneity may be the difference in sexual HIV transmission between the two regions. The HCV antibody prevalence among IDUs infected with HIV was 86.7% in P County and 83.7 in R County. The HIV/HCV prevalence ratio within the IDU population was 0.64 in P County and 0.90 in R County. The median estimated percentage of existing HIV infections due to sexual HIV transmission was 55% in P County and 67% in R County [28]. However, in I County, the HCV prevalence among IDUs infected with HIV was 12.9% and the HIV/HCV prevalence ratio within the IDU population was 5.1. The median estimated percentage of existing HIV infections due to sexual HIV transmission was 88% [28]. These findings imply that intravenous drug use and heterosexual intercourse might be two important transmission routes in both P and R Counties, whereas in I county, heterosexual intercourse might be the dominate transmission route. The route of transmission in I county might be reflective of a change in the route of transmission from intravenous drug use to heterosexual sex. This needs further study. Analysis of surveillance data from the Yi Autonomous Prefecture showed that, over time, the percentage of intravenous drug use accounting for HIV transmission decreased, while transmission by heterosexual intercourse increased [29].

Our findings of substantial geographic heterogeneity of HIV and HCV infections at the township level imply that it is may be necessary to inform or invent site-specific intervention strategies to address the special cultural traditions of the Yi ethnic group in endemic areas and to better allocate currently limited resources to combat these two viruses. For example, P and R Counties have a high prevalence of co-infection with HIV and HCV. Because HIV/HCV co-infection in IDUs has become a rapidly emerging global public health problem [30, 31], multiple approaches are imperative to prevent both HIV and HCV infections, identify new cases, and provide care. Local MMTP clinics should be strengthened and expanded using local multi-sector cooperation (e.g., public security and public heath sectors) to increase the coverage and retention of MMTP and improve accessibility of services [11]. Thus, more drug users will be engaged in the MMTP clinics and receive more effective interventions, including uninterrupted drug substitution treatments, and both HIV and HCV testing. Effective prevention and treatment for both infections will be available and provide an ability to halt the tide of increasing morbidity and mortality due to HCV/HIV-related disease.

There are some limitations to our study. First, only a part of all the drug users enrolled in the MMTP clinics. It is difficult to confirm whether MMTP participants are a truly representative sample of a population of community-dwelling drug users. However, it has been estimated that the national HIV prevalence among MMTP clients is not significantly different from that among non-MMTP drug users reported by national sentinel surveillance during 2004–2009 [12]. Second, HIV and HCV are transmissible infections, and they often share similar transmission routes. Therefore, the clustering found with the spatial methods might be related to the clustering of risk behaviors, to natural geographic boundaries between human populations, or to both. However, due to the limitations of these analytic approaches (e.g., they did not consider HIV, HCV and co-infections as communicable diseases, and are limited in the ability to identify which reason may contribute to the clustering), we cannot analyze unambiguously what resulted in the clustering. In addition, the information regarding the use of condoms and sexual behaviors is not available and individual risk factors were not considered in the framework of our geographic analyses. These limit the ability to further analyze why the geographic distribution of these infections in our study field was heterogeneous. Some studies show that casual sexual behaviors are frequent, and condoms were infrequently used among the Yi ethnic group compared to the Han ethnic group [20, 25].

This was the first study to analyze the geographic distribution patterns of HIV-, HCV, and co-infections among drug users in a national MMTP in a Yi Autonomous Prefecture, Southwest of China. These infections exhibited substantial geographic heterogeneity at the township level. Their geographic distribution patterns were also obviously different. These findings imply that it is may be necessary to inform or invent specific intervention strategies, particularly regarding the Yi ethnic group. Currently limited resources will need to be reallocated according to these findings.