Introduction In 2015, the estimated HIV prevalence in Brazil was 0.4%. This figure has been stable in recent years, but it can mask disparities among regions. In this study we present the spatial distribution of the variation in HIV incidence rates(IR) in Brazilian municipalities from 2009–14 and assess the existence of spatial clustering of increase or decrease in these IR.
Methods We used the AIDS reporting system(Sinan) and programmatic data on Viral Load(VL) exams and ARVs. The diagnosis date used was the earliest among 1st detectable VL, 1st ARV dispensation or diagnosis date in Sinan. Annual IR were generated by municipality. To smooth the IR, we used 3 year averages and applied the local empirical Bayesian method. To assess IR time trends, we calculated the percent IR variations in the period. For spatial statistical analysis, a simple adjacency matrix was generated, and Global and Local Moran’s I autocorrelation tests were applied.
Results The Global Moran’s I for the IR variation was 0.42 (p<0.001) which points to spatial clustering. We generated 2 maps, one for the percent IR variation and another to represent the statistically significant high-high and low-low clusters. In the 1st, we observed that most municipalities in the North(N) and Center-West(CW) present increases in the period; in the Southeast(SE), the state of Sao Paulo(SP) reveals the most relevant decreases in the country; the Northeast(NE), South(S) and some areas of the SE show mixed patterns. The 2nd map makes regional disparities even clearer. There are big clusters of increasing IR in most states of the N, and smaller ones in areas of the CW and NE. Several clusters of declining IR are seen in SP, Minas Gerais(SE), Santa Catarina, Rio Grande do Sul(S), and areas of the NE.
Conclusion Spatial dependency in HIV IR variations in Brazil was evidenced. The methods used in this study have proved useful in monitoring spatiotemporal trends, pointing out important regional differences. Similar analysis can be performed at state and city levels, contributing to improved diagnoses of local epidemics.