To our knowledge, this is the largest application of the BED-CEIA to estimate HIV-1 incidence and the first application among discordant and pre-marital couples in China. Simulation of the HIV epidemic among IDUs in Dehong from 2004 to 2008 showed that the epidemic continues to expand but that the rate has been slowing in the most recent years. Our incidence estimates suggest that lower incidence may explain the slowing of the growth of the epidemic among IDUs. The reasons why HIV prevalence among IDUs remained stable while the incidence decreased over the five-year period (but still remained relatively high) may include several factors: 1) A high death rate among HIV-infected IDUs; 2) High mobility among IDUs in Dehong and neighboring areas with lower HIV prevalence; 3) The impact of local government policies on strengthening the use of force to combat drug abuse; 4) Temporal changes in the number of newly developed IDUs and drug addiction levels in different age groups. Finally, misclassification of transmission route may also be a possible explanation.
The epidemic of HIV/AIDS in China consists of at least eight sub-epidemics [17
] along with the spread of other STDs, such as syphilis [18
]. China's HIV epidemic remains low prevalence overall [19
] with high infection among specific sub-populations and in certain geographic areas such as Dehong Prefecture near the border with Myanmar [20
]. Until recently, IDU was the leading route of HIV transmission in China and HIV continues to spread rapidly among IDUs in Dehong Prefecture and other major drug trafficking corridors in China [1
]. However, the HIV transmission mode has been changing gradually from IDU to sexual transmission. Our study demonstrates that HIV incidence among IDUs in Dehong is decreasing. However, incidence among the other four focal populations, all representing sexual transmission of HIV, has remained stable. In our study, sexual transmission of HIV appears concentrated in several high-risk populations (especially discordant couples and FSWs) and has not diffused substantially into the general population.
In a previous study, we showed that IDUs had the highest HIV incidence rate in Yunnan Province, varying between 2.2% and 8.0% from 2004 to 2008, and HIV-1 genetic diversity had become increasingly complex [2
]. HIV incidence among IDUs in Dehong was also higher than in other regions of China (except Xinjiang Province in the Northwest) [8
]. We speculate that the decrease in incidence among IDUs may be partially explained by the changes in risk behaviors, notably uptake in needle exchange and a reduction in needle sharing [21
]. While there has been a notable increase in access to antiretroviral therapy among IDUs and some likely reduction in AIDS- and drug-related deaths, more extensive modeling would be needed to estimate whether or not these have also contributed to lower prevalence and incidence estimates. After IDUs, negative persons in HIV-discordant relationships had the next highest risk of transmission, with incidence estimated by BED varying from 3.4% to 7.2% and is very similar to that in a cohort study of stable sex partners in Africa [22
]. Because of the high mobility of FSWs and the fact that many are only in the business for a relatively short period of time, the incidence and prevalence estimates in this group are relatively close.
Since 2004, China Comprehensive Response Project Areas were founded in Dehong, and a series of HIV control and prevention efforts have been initiated, including MMT, needle and syringe exchange, prevention of mother-to-child transmission, and distribution of free condoms as part of a 100% condom policy for sex workers and their clients. Although coverage of intervention programs has rapidly expanded in Dehong, the HIV epidemic was likely influenced differently by these various intervention efforts and we also found that the reported needle sharing by IDUs did not markedly decrease over the period of this study. Major challenges to harm reduction include the high relapse rate for MMT and the limited ability to reach the majority of heroin users due to various barriers [21
]. Therefore, further multivariate analysis is needed to explore which measures may be most effective for different populations. Our data emphasize the importance of accelerating HIV prevention programs like needle exchange and MMT to reduce needle sharing among IDUs and condom distribution and promotion to reduce unprotected sex among discordant couples and among FSWs and their clients.
There are a number of different laboratory methods to estimate HIV incidence [23
] and the use of these methods is still evolving, especially in populations with HIV-1 non-B subtypes. It is likely that BED-CEIA will have its utility as a sensitive screening test, with avidity used to exclude false positives, and modeling to further estimate the remaining pool of false positives. The utility of the BED-CEIA assay has been demonstrated in a number of studies, in particular for elucidating trends in incidence [11
]. Regional data from sub-Saharan Africa on the use of BED have indicated that it overestimates the HIV incidence [27
] and further adaptations have been made to enhance accuracy, especially in populations infected with non-B subtypes [13
]. In the current study, the trend in incidence among IDUs was very similar to that of a prospective cohort study conducted in the same area [29
]. The prevalence and incidence among pregnant women and pre-marital couples were very similar, with slightly higher rates among pre-marital couples. These findings indicate that the HIV incidence calculated from the BED-CEIA assay is credible. This has previously been shown also in another study from China [8
]. We excluded patients with CD4+ cell count ≤200 and those on ART, as recommended [13
]. This could lead to a potential bias of the estimates because relatively less vigorous exclusion was applied in the earlier years than in late years during the period from 2004 to 2008. IDUs recruited from detoxification centers could be an easier target for harm reduction programs than IDUs in the community, which are generally harder to reach; therefore, the trends of HIV incidence among detoxification users may not reflect accurate trends among non-detoxification users. Despite current limitations, methods that allow cross-sectional HIV incidence estimation are an important advance, and these methods will be further refined as more experience is gained in different settings. While absolute HIV incidence estimates may be imperfect, they may still be highly useful to determine trends over time [26