BACKGROUND

Asian HIV epidemics are concentrated among particular behavioural groups, but large variations exist in epidemic types, timing, and geographical spread between countries and within countries, especially in China. We aimed to understand the complexity of HIV epidemics in China by systematically analysing prevalence trends by data source, region, population group, and time period.

METHODS

We collected HIV prevalence data from official national sentinel surveillance sites at the provincial level from Jan 1, 1995, to Dec 31, 2010. We also searched PubMed, VIP Chinese Journal Database (VIP), China National Knowledge Infrastructure, and Wanfang Data from Jan 1, 1990, to Dec 31, 2012, for independent studies of HIV prevalence. We integrated both sets of data, and used an intraclass correlation coefficient test to assess the similarity of geographical pattern of HIV disease burden across 31 Chinese provinces in 2010. We investigated prevalence trends (and 95% CIs) to infer corresponding incidence by region, population group, and year.

RESULTS

Of 6850 articles identified by the search strategy, 821 studies (384,583 drug users, 52,356 injecting drug users, 186,288 female sex workers, and 87,834 men who have sex with men) met the inclusion criteria. Official surveillance data and findings from independent studies showed a very similar geographical distribution and magnitude of HIV epidemics across China. We noted that HIV epidemics among injecting drug users are decreasing in all regions outside southwest China and have stabilised at a high level in northwest China. Compared with injecting drug users, HIV prevalence in female sex workers is much lower and has stabilised at low levels in all regions except in the southwest. In 2010, national HIV prevalence was 9·08% (95% CI 8·04-10·52) in injecting drug users and 0·36% (0·12-0·71) in female sex workers, whereas incidence in both populations stabilised at rates of 0·57 (0·43-0·72) and 0·02 (0·01-0·04) per 100 person-years, respectively. By comparison, HIV prevalence in men who have sex with men increased from 1·77% (1·26-2·57) in 2000, to 5·98% (4·43-8·18) in 2010, with a national incidence of 0·98 (0·70-1·25) per 100 person-years in 2010. We recorded strong associations between HIV prevalence among at-risk populations in each province, supporting the existence of overlap in risk behaviours and mixing among these populations.

CONCLUSIONS

HIV epidemics in China remain concentrated in injecting drug users, female sex workers, and men who have sex with men. HIV prevalence is especially high in southwest China. Sex between men has clearly become the main route of HIV transmission.