Estimating HIV incidence from surveillance data indicates a second wave of infections in Brazil

Abstract

Emerging evidence suggests that HIV incidence rates in Brazil, particularly among men, may be rising. Here we use Brazil’s integrated health systems data to develop a mathematical model, reproducing the complex surveillance systems and providing estimates of HIV incidence, number of people living with HIV (PLHIV), reporting rates and ART initiation rates.An age-structured deterministic model with a flexible spline was used to describe the natural history of HIV along with reporting and treatment rates. Individual-level surveillance data for 1,077,295 cases (HIV/AIDS diagnoses, ART dispensations, CD4 counts and HIV/AIDS-related deaths) were used to calibrate the model using Bayesian inference.The results showed a second wave of infections occurring after 2001 and 56,000 (95% Credible Interval 43,000–71,000) new infections in 2015, 37,000 (95% CrI 28,000–54,000) infections in men and 16,000 (95% CrI 10,000–23,000) in women. The estimated number of PLHIV by end-2015 was 838,000 (95% CrI 675,000–1,083,000), with 80% (95% CrI 62–98%) of those individuals reported to the Ministry of Health. Women were more likely to be diagnosed and reported than men; 86.8% of infected women had been reported compared with 75.7% of men. Likewise, ART initiation rates for women were higher than those for men.The second wave contradicts previous estimates of HIV incidence trends in Brazil and there were persistent differences in the rates of accessing care between men and women. Nevertheless, the Brazilian HIV program has achieved high rates of detection and treatment, making considerable progress over the past ten years. Keywords: HIV/AIDS, Incidence estimation, Brazil, Deterministic model, Prevalence, HIV care cascade