—First, to determine whether there is evidence for loss of human immunodeficiency virus type 1 (HIV-1 ) antibody in seroreactive individuals. Second, if true seroreversion occurs, to determine its incidence relative to errors in the testing process.
—A retrospective cohort study reviewing the results of 5 446 161 HIV-1 antibody tests performed on 2 580 974 individuals (the US Army HIV Data System) from 1985 through 1992. For all patients with one or more seroreactive sample followed by one or more nonreactive sample, we examined available records and retested the samples.
—Serum samples had been obtained from active-duty and retired military personnel, their dependents, and applicants to the military.
—Of 4911 individuals reported to be seroreactive for HIV-1 by two independent samples, only six were potential seroreverters. Review of the six cases revealed that five actually were HIV-seroreactive patients who had samples from nonreactive individuals mistakenly attributed to them, while the sixth had a testing error proven by retesting the discrepant specimen. Errors in the testing process were identified (n=23) or suspected (n=3) in another 26 individuals who had not had independent confirmation of reactivity by a second sample. The cumulative error rate was 12.4 per 1 million patients tested. An additional group of 31 uninfected infants appeared to serorevert due to loss of antibody acquired from their HIV-1— infected mothers.
—Review of this database demonstrates no evidence for true seroreversion of HIV-1 antibody status. We conclude that if seroreversion occurs at all, it is exceedingly rare. In fact, most (if not all) cases of apparent seroreversion represent errors of attribution or testing.(JAMA. 1993;269:2876-2879)