Applications Of Spatial Analysis Methods In Figuring Out The Situation And Trends Of China’s HIV Epidemic

BackgroundIn China, the HIV epidemic is obviously uneven in different regions due to their different development speed of HIV epidemic and different regional social conditions. It’ll be very meaningful for HIV prevention and control if we could category out sub-epidemic areas by their practical characteristics of HIV epidemic.ObjectiveTo analyze the spatial clustering distribution of China’s reported HIV/AIDS cases, trying to find out its changing trend. To explore the related association between HIV/AIDS prevalence and macroscopic social factors. To figure out the diversity of HIV/AIDS epidemic among different areas according to their varied epidemic characteristics, trying to form sub-epidemic areas.MethodsCollect data of reported HIV/AIDS cases from1985until the end of2011, and also collect data material of provincial social economy, demography and health of2011. First, spatial scan statistic method and spatial autocorrelation analysis were used to explore the clustered areas of reported HIV/AIDS cases at province level and county level respectively. Second, extract comprehensive variables of social factors using principal component analysis, and then explore its association with HIV/AIDS prevalence using two-level negative binomial model. Third, thirteen provincial variables generated from original HIV/AIDS epidemic data and socioeconomic indicators to indicate HIV/AIDS epidemic characteristics, were introduced to hierarchical clustering analysis to form sub-epidemic areas at province level.Results1. The results of spatial analysis showed the existence of significant clustering distribution till each year from1985to2011both at the province level and the county level. And the clustering distribution of reported HIV/AIDS cases extended from Yunnan, Xinjiang and some other provinces to the central regions such as Henan province from1995to2003, causing the decline of the spatial clustering in2003(General Moran’s I coefficient was0.1596, Z=15.9926, P<0.05). And since2005, the reported cases of southwestern provinces predominated again, causing another increase of the clustering distribution of cases. By applying spatial methods to reported cases from different transmissions, it was found that cases from injecting drug use (IDU) and heterosexual routes were mainly clustered in Yunnan, Guangxi, Xinjiang and so on, cases from blood/plasma route were significantly clustered in China’s central provinces in2004, and cases from homosexual route were mainly clustered in recent several years (2008-2011) and their clustered regions were dispersed, most of which were high economic-level regions.2. Fitted two-level negative binomial model showed that, HIV/AIDS prevalence data of the county level were clustered in their higher level units or provinces (σu02=2.612, P<0.05). In other words, after the adjustment of gender and age, there was significant difference between HIV/AIDS prevalence of different provinces. Univariate and multivariate analysis of macroscopic social factors showed that, there were three comprehensive variables were significantly positive associated with HIV/AIDS prevalence (P<0.05), and they were high level of population mobility or low level of economic living standards (N3), high proportion of minority population (M4), and large number of HIV screening population (M5) respectively.3. Three HIV/AIDS sub-epidemic areas (A, B, C) were formed by taking four aspects variables into consideration, and they were early epidemic characteristic, epidemic spread and development speed, current epidemic status and macroscopic social factors. A area was the earliest and severest HIV/AIDS epidemic area and occupied75.4%hotspot counties. The epidemic in A1and A2was driven by IDU in its early period and heterosexual transmission later. Henan, the only province in A3, characterized by its HIV/AIDS epidemic among former plasma donors (FPDs) during early1990s, presented strong spatial clustering of blood/plasma transmission occupying about80%blood/plasma hotspots. B area occupied most high economic-level regions, and its epidemic was driven by heterosexual populations, and recently by men who have sex with men (MSM) which had been largely spread since2005. B4and B5occupied50.0%MSM hotspots. For C6and C7, only sporadic HIV/AIDS infections occurred in the past years among FPDs and heterosexual populations, whereas the prevalence among MSM had been increasing.ConclusionIn China, the epidemic of HIV/AIDS displayed different models of spatial distribution during different periods and within different provinces. And the HIV epidemic is closely associated with regional macroscopic social factors. It’s more reasonable to clarify sub-epidemic areas by considering their HIV epidemic characteristics and social background, which is good for determining targeted prevention policies.