Meta-analysis for the pooled sensitivity and specificity of anti-human immunodeficiency virus Ab rapid tests

Abstract

Background: Many immunochromatography (ICA) kits for anti-human immunodeficiency virus type (HIV) antibody (Ab) have been introduced to improve the accessibility of HIV Ab tests. However, qualified evaluation reports for HIV rapid tests are not enough to validate their performances. Meta-analysis for the performances of the HIV Ab rapid tests was performed in this study. Methods: PubMed database was searched with combination of search terms, 'human immun-odeficiency virus', 'HIV Ab', 'rapid test', 'immunochromatography', 'performance', 'sensitivity', and 'specificity'. Criteria of inclusion were performance studies for HIV ICA kits with serum or EDTA whole blood. Methodological qualities were evaluated with standards for reporting of diagnostic accuracy studies (STARD) checklists by two investigators. Homogeneity among selected studies was evaluated and then pooled sensitivity and specificity were calculated. Positive and negative predictive values were simulated with presumed HIV prevalence in Korea. Results: Twenty-three studies were selected from 12 high-qualified papers with STARD checklists. The performance of 23 studies were found to be heterogeneous (P<0.1) and random effect model was used. Pooled sensitivity was 99.71% (95% CI: 99.45-99.97%) and pooled specificity was 99.27% (95% CI: 98.83-99.70%). With HIV prevalence of 0.03%, positive and negative predictive values were presumed to be 3.936% and 99.999%, respectively. Conclusions: This meta-analysis for HIV ICA rapid tests showed good performance. In consideration of low positive predictive values of HIV rapid tests, confirmation by enzyme immunoassay or Western blot is still needed. This study would be helpful in evaluating and establishing proper performance guideline for those kits not fully evaluated.