The Mozambican Ministry of Health began the stepwise

introduction of combined antiretroviral therapy (cART) throughout the country in 2005. In the MDH in Manhiça, cART was introduced in 2005. BMS-354825 in vivo Estimation of HIV incidence in the current analysis was based on the methodology validated by Hallett et al. [1] to estimate HIV incidence between two prevalence surveys. The method relies on the decomposition of prevalence changes by age group of width r (usually 5 years) between two cross-sectional surveys separated by T years of time. Thus, the HIV prevalence in the second of two cross-sectional surveys represents the sum of new HIV infections plus the survivors of previously recorded HIV-infected individuals. Five HIV prevalence points were available from the studies described above (1999, 2003, 2004, 2005 and 2008). Hallett et al. [1] proposed two methods for estimating HIV incidence from prevalence. The first is based on mortality rates derived from three potential HIV epidemic scenarios. These

are: (i) an expanding epidemic, (ii) a stable epidemic and (iii) a declining epidemic. These scenarios consider mortality changes related to both prevention and treatment strategies. In this analysis we used mortality rates from the publication of Hallett et al. [1] obtained from neighbouring African countries, as HIV-specific mortality data for Manhiça were not available. The second method uses a survival distribution from infection to death by age to obtain mortality rates. The Weibull survival distribution from the publication of Doramapimod ic50 Hallett et al. [1] was used.

The incidence rate can be estimated using both methods for the ith age cohort, if the time between surveys T is equal to the age-group interval width r=5 years. If the time between surveys T is different from the age-group interval width r, the incidence rate for the ith age not group can be obtained as a weighted mean of the consecutive ith age-cohort incidences: The inter-survey global incidence estimate for individuals aged 15–45 years can be calculated using a weighted mean based on age-group size Pi as To obtain the yearly incidence rate estimates, a quadratic curve is fitted to the HIV mid-point incidence estimation between surveys: After re-sampling individuals in the prevalence surveys, bootstrap confidence intervals were generated. A sensitivity analysis was conducted by repeatedly fitting the regression model after omitting each point prevalence one by one. Five point prevalences for 1999, 2003, 2004, 2005 and 2008 were calculated from the data of the three studies, as described in the Methods section. The prevalence of HIV infection among the 180 women aged 15–45 years in the study carried out in 1999 was approximately 12% [95% confidence interval (CI) 8–18%].