HIV screening and confirmation: a simplified and less expensive testing algorithm.

Abstract

In this study we investigated the performance of fourteen different assays capable of simultaneously detecting antibodies to HIV-1 and HIV-2, referred to as combined screening assays (CSAs), on a panel of 371 sera, with a prevalence of 51.5% and 1.3% for HIV-1 and HIV-2 antibodies respectively. The geographic distribution of the sera was as follows; Europe (121), Africa (203) and Latin America (47). These sera were collected from different clinical groups of patients; Asymptomatic (36), AIDS-Related Complex/AIDS patients (18), infected individuals with generalised lymphadenopathy (12), blood donors (149), and subjects with unknown clinical status (156). The Dupont Western blot (WB) kit for detection of HTLV-III antibodies and the Pasteur new Lav-Blot II kit were used for the confirmation of HIV-1 and HIV-2 infection respectively. Of the 14 tests studied, 9 were enzyme linked immunosorbent assays (ELISAs), and 5 were non-Elisa tests requiring visual reading. An alternative approach for HIV antibody testing was studied restrospectively, whereby sera positive in an initial CSA (A) were retested on a second CSA (B), that was different from the first. The use of WB was limited to sera that gave discrepant (A+B-) results in the two CSAs. A positive result in both CSAs was reported as anti-HIV positive. A negative result in the first CSA was reported anti-HIV negative. Sensitivity, specificity, cost, and the delta (delta) values (delta values of the ELISA assays) were taken into consideration when selecting suitable pairs of assays. All the ELISAs scored 100% sensitivity, but for the non-ELISAs, the sensitivity ranged from 96.0% to 100%. The specificity for the ELISAs and non-ELISAs varied from 87.4% to 100% and from 51.4% to 100% respectively. Delta (delta) values for the ELISAs ranged from 3.82 to 136.68 and from -1.15 to -3.08 for the anti-HIV positive and anti-HIV negative populations respectively. Of the 121 test combinations studied, 9 (7.4%) pairs yielded 100% sensitivity and specificity and 61 (50.4%) pairs of CSAs required further testing on WB. This implies 100% positive predictive value, at a cost that was on average 6 times less, and a testing time that was 5 times faster than the conventional algorithm. We conclude that there are several combinations of pairs of CSAs that can be used in the alternative algorithm that can provide accurate results at a much lower cost than the conventional algorithm requiring confirmation by WB of all initially reactive CSA results.(ABSTRACT TRUNCATED AT 400 WORDS)