False-negative testing errors in routine viral marker screening of blood donors. For the Retrovirus Epidemiology Donor Study.

Abstract

The contribution of testing errors to the risk of virus transmission by transfusion depends on the rate of false-negative testing errors and the prevalence of infected seropositive donations. Although the false-negative testing error rate has been estimated at 0.1 to 1 percent on the basis of proficiency studies, it has not previously been measured in routine donor screening. A 1991 to 1995 database containing 5,153,153 donations from 1.5 x 10(6) donors (including autologous donors) was searched to identify donors who tested seropositive for HIV, HCV, HTLV-I or II and who attempted subsequent donations. The false-negative rate in routinely screened follow-up donations was determined, and false-negative cases were investigated to identify the cause. Subsequent donations (n = 2015) by 1224 donors with confirmed-positive results were identified. Eleven (0.5%) of these donations did not react in EIA. Ten of the 11 false-negative cases were attributable to borderline-reactive donations. On subsequent donations, there were borderline-nonreactive results on HTLV-I (2 cases), first-generation HCV (5 cases), and second-generation HCV (3 cases) EIAs. The final case was strongly reactive for HCV in a second-generation EIA on two donations (signal-to-cutoff [S:C] ratio >3.5), followed by a baseline nonreactive result on a third donation (S:C = 0.05). False-negative testing results occur infrequently during routine infectious-disease donor screening. Although most false-negative results occurred with borderline-reactive HTLV-II samples and/or early-generation HCV EIAs, frank technical errors (e.g., sample mixup or failure to add sample to EIA well) also occur at a low rate (0.05%; 95% CI, 0-1.5%). Process enhancements designed to reduce errors (e.g., enhanced automation of data management and testing systems and process controls for EIAs) are warranted to detect and prevent false-negative results.