Are seroepidemiological surveys for human immunodeficiency virus infection based on tests on pools of serum specimens accurate and cost-effective?

Serum specimens (n = 17668) from UK antenatal patients in the Thames Regions were tested by Wellcozyme HIV 1/2 EIA singly and in pools of 6, 12 and 24: 35 (0.2%, 1 in 505) were confirmed as anti-HIV positive. The pools of 12 were also tested for anti-HIV 1/2 by IAF Biochem, Behring and Diagnostics Pasteur EIAs. All 35 positive specimens were easily detectable after pooling in groups of 12. The false positive rate for Wellcozyme was nearly halved compared with individual testing (1 in 309 false positive compared with 1 in 174). For the other assays false positive rates on pools of 12 were: IAF Biochem 1 in 193, Behring 1 in 140, Diagnostics Pasteur 1 in 1547. Twenty-two known anti-HIV 2-positive sera were detected by all four EIAs when diluted as in pools of 6 and 12, but by only three EIAs in pools of 24 and 48. Pooling in groups of 6 did not seem to delay detection of HIV 1 seroconversion, but pooling in groups of 12, 24 and 48 might delay it by 1, 2 and 3 weeks respectively. For this study the effect of pooling in groups of 12 would have been a reagent saving of 87-91% and a labour saving of about 50%. Because of the low HIV incidence and rarity of specimens collected around seroconversion in UK, little, if any, loss of sensitivity would result from it. Pooling in groups of 12 has therefore been chosen for the screening of anonymous antenatal specimens in the UK.