High false discovery rate of the Architect anti-HCV screening test in blood donors in Uganda and evaluation of an algorithm for confirmatory testing

Background and Objectives: Adequate supplies of donor blood remains a major challenge in sub-Saharan Africa. This is exacerbated by lack of confirmatory testing for transfusion transmitted infections by Blood Transfusion Services (BTS) leading to significant blood disposal owing to putatively high seroprevalence rates amongst Ugandan blood donors. We aimed to ascertain the false discovery rate of the Architect anti-HCV screening assay, categorise screen-reactive samples into 3 groups: presumed false-positive, active and past infection, and develop an algorithm for confirmatory testing. Materials and Methods: 470 screen-reactive HCV blood donations were re-tested using the Architect anti-HCV assay, an alternative antibody test (SD biosensor) and a core antigen test. Sample-to-cut-off (S/CO) ratios and pre-analytical factors (centrifugation speed, haemolysis check, time between collection and testing) were recorded. Based on S/CO ratio evaluation, we propose a testing algorithm to guide supplemental tests. Results: The false discovery rate of the Architect anti-HCV assay was 0.84 as 395/470 (84%) screen-reactive samples had no evidence of HCV infection (SD biosensor and core antigen negative) (presumed false-positive). 38/470 (8.1%) were antigenaemic and 32/470 (6.8%) had evidence of past infection. The median S/CO ratios of the presumed false-positive and active infection samples were 1.8 and 17.3 respectively. The positive predictive value of HCV positivity in samples with ratios above 12 was 91.8%. On re-testing, 104/470 (22.1%) samples became negative. Conclusion: The Architect anti-HCV assay has a very high false discovery rate in Ugandan BTSs leading to excessive blood disposal. Pre-analytical factors likely contribute to this. Introduction of confirmatory testing using an algorithm based on S/CO ratio evaluation could limit unnecessary blood wastage and donor deferral