Performance of an immunoassay at detecting recent infection among reported HIV diagnoses

An enzyme immunoassay to detect recent HIV-1 infection (EIA-RI) of less than 6 months is routinely performed on diagnoses reported to the National HIV case surveillance in France. We assessed the performance of the EIA-RI infection on this country population scale by measuring its agreement with other indicators of time since infection that were obtained through clinical, biological or testing history recorded on the surveillance reporting form. We used data from the National HIV case surveillance from its debut in March 2003 to June 2007. Infection within 6 months was defined as a negative test reported within 6 months prior to diagnosis. We further ascertained this definition by adding information about of a symptomatic primary infection or biological evidence of recent seroconversion. Infection established for more than 6 months was defined when a positive test had occurred more than 6 months prior to the reported diagnosis. Time since infection could be ascertained in 6782 of 15, 331 (44.2%) HIV diagnoses. Assay sensitivity and specificity were 73.8 and 83.7%, respectively. Among the 1940 cases originating from Sub-Saharan Africa, sensitivity and specificity were 54.1 and 90.8%, respectively. Assessment of the performance of the EIA-RI on a large and heterogeneous population revealed two major findings--significant discrepancies in timing from infection near the 180-day cutoff, and a performance that depends on the geographic origin of patients. This has implications for estimating the assay window period and in the perspective of incidence estimation from HIV case surveillance.