Estimating HIV Incidence, Prevalence, and Proportion Undiagnosed Based on CD4 Data

• Activity Number: 135
• Type: Contributed
• Date/Time: Monday, August 4, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Ruiguang Song*+ and Celia Landmann Szwarcwald and Timothy A. Green and H. Irene Hall
• Companies: CDC and Brazil Ministry of Health and CDC and CDC
• Keywords: HIV ; Incidence ; Prevalence ; Surveillance

Abstract

The incidence of HIV infection is difficult to estimate because there is a long latency period between infection and symptom development in HIV disease. The method using a serologic testing algorithm for recent HIV seroconversion can estimate incidence but only for years and in areas where diagnostic specimens from persons newly diagnosed with HIV are tested with an incidence assay, and these data cannot be used to estimate HIV prevalence and the proportion undiagnosed. A method that can be used to estimate both HIV incidence and prevalence is extended back-calculation. However, this method requires HIV case surveillance data since the beginning of the epidemic and these data are not available in many areas. In this study, we use only data on persons diagnosed with HIV in recent years and results of their first CD4 tests after diagnosis. We use CD4 count to measure the progression of HIV disease and estimate the distribution of delay from HIV infection to HIV diagnosis. The annual numbers of HIV infections and the proportion of infected but undiagnosed persons are then estimated using the distribution of delay based on data reported to the HIV case surveillance system.