AEGiS-14IAC: Further refinements of HIV incidence estimation using Detuned Assay testing.

14th International AIDS Conference

Barcelona, Spain - July 7-12, 2002

Further refinements of HIV incidence estimation using Detuned Assay testing.

Int Conf AIDS 2002 Jul 7-12; 14:(abstract no. MoOrC1039)

Cleghorn FR, Sill A, Jack NE, Smith H, Forsythe-Duke V, Kreisel K, Banfield-Roach D, Bartholomew C, Constantine N, Blattner WA
University of Maryland, Baltimore, United States

BACKGROUND: The use of cross-sectional sampling methods to estimate HIV incidence via sensitive/less sensitive (S/LS) assays has expanded in the past 5 years. Current data suggest that incidence using this approach is over-estimated by 10-20%, based largely on the imperfect sensitivity and specificity of the assay.

OBJECTIVE: To devise a sensitive/less sensitive confirmatory testing algorithm to accurately identify persons with recent HIV infection, and to compare incidence estimates with standard approaches.

METHODS: A gold standard of 178 well characterized serial samples from 28 cases with known clinical dates of infection ascertained and followed in the STD clinic in Port of Spain, Trinidad were used to calibrate the screening (UNIGOLD rapid S/LS assay) and confirmatory assay (Organon Technika Vironostika (DV) S/LS assay) to optimally define patients with recent infection. The assays' performance characteristics were modified to maximize 2-stage specificity so that the predictive value for early infection was 100%. The algorithm was then applied to all HIV seropositive samples in the clinic in the 2000-2001 period.

RESULTS: Using the standard DV S/LS assay, HIV incidence increased from 2.2% to 4.2% (95% CI 4.1-4.5) between 2000 (N=2047) and 2001 (N=2906). When the confirmatory algorithm was applied to the 2001 seropositive samples [262/2906, 9% seroprevalence], the more accurate incidence was refined to 3% (2.9-3.2), a difference of 29% in the incidence estimate, and reflecting the 29/45 recent samples confirmed as recent.

CONCLUSIONS: A 2-stage algorithm using a rapid S/LS screening test in conjunction with an S/LS ELISA provides more accurate HIV incidence data than classical serological detuning assay alone. These methods may be useful for rapidly confirming incidence rates in populations targeted for vaccine studies.

Keywords: AEGIS, Incidence, HIV Infections, HIV Seropositivity, Research Design, Sensitivity and Specificity, Algorithms, Enzyme-Linked Immunosorbent Assay, Seroepidemiologic Studies, Spain, Trinidad and Tobago, Human, epidemiology, utilization, methods

020707
MoOrC1039