Abstract: | The characteristics of primary (first) tests with three enzyme-linked immunosorbent assay (ELISA) kits for human immunodeficiency virus (HIV) antibody were determined. The three ELISAs were performed on 3,229, 3,130, and 685 specimens from high-risk individuals using the Litton (LT; Litton Bionetics Laboratory Products, Charleston, S.C.), Dupont (DP; E. I. du Pont de Nemours & Co., Inc., Wilmington, Del.), and Genetic Systems (GS; Genetic Systems, Seattle, Wash.) kits, respectively. Evaluation was based on the distribution of quantitative test results (such as optical densities), a comparison with Western blot (WB) results, reproducibility of the tests, and identification of seroconverters. The performances of the GS and the DP kits were good by all four criteria and exceeded that of the LT kit. Primary ELISA-negative results were not always confirmed with repeat ELISA and by WB testing. The largest percentage of these unconfirmed negative test results came from samples with quantitative results in the fifth percentile nearest the cutoff. Thus, supplementary testing was indicated for samples with test results in this borderline negative range. Similarly, borderline positive primary ELISA results that were quantitatively nearest (fifth percentile) the cutoff value were more likely to be antibody negative on supplementary testing than samples with high antibody values. In this study, results of repeated tests by GS ELISA showed the least change from first test results. DP ELISA showed more unconfirmed primary positive test results, and LT ELISA showed more unconfirmed primary negative test results. Designation of a specimen with a single ELISA quantitative level near the cutoff value as positive or negative should be viewed with skepticism. A higher than normal proportion of specimens with high negative optical densities by GS ELISA (fifth percentile nearest the cutoff) and also negative by WB were found to be from individuals in the process of seroconversion. |