Abstract
BACKGROUND: Countries measure trends in HIV incidence to assess the impact of HIV prevention and treatment programmes. Most countries have approximated trends in HIV incidence through modelled estimates or through trends in HIV prevalence among young people (aged 15-24 years) assuming they have recently become sexually active and have thus only been recently exposed to HIV. METHODS: Trends in HIV incidence are described and results are compared using three proxy measures of incidence: HIV prevalence among young women attending antenatal clinics (ANCs) in 22 countries; HIV prevalence among young male and female nationally representative household survey respondents in 14 countries; and modelled estimates of adult (ages 15-49 years) HIV incidence in 26 countries. The significance of changes in prevalence among ANC attendees and young survey respondents is tested. RESULTS: Among 26 countries, 25 had evidence of some decline in HIV incidence and 15 showed statistically significant declines in either ANC data or survey data. Only in Mozambique did the direction of the trend in young ANC attendees differ from modelled adult incidence, and in Mali and Zambia trends among young men differed from trends in adult incidence. The magnitude of change differed by method. CONCLUSIONS: Trends in HIV prevalence among young people show encouraging declines. Changes in fertility patterns, HIV-infected children surviving to adulthood, and participation bias could affect future proxy measures of incidence trends.