An optimized stepwise algorithm combining rapid antigen and RT-qPCR for screening of COVID-19 patients

We investigated the combination of rapid antigen detection (RAD) and RT-qPCR assays in a stepwise procedure to optimize the detection of COVID-19. From August 2020 to November 2020, 43,399 patients were screened in our laboratory for COVID-19 diagnostic by RT-qPCR using nasopharyngeal swab. Overall, 4,691 of the 43,399 were found to be positive, and 200 were retrieved for RAD testing allowing comparison of diagnostic accuracy between RAD and RT-qPCR. Cycle threshold (Ct) and time from symptoms onset (TSO) were included as covariates. The overall sensitivity, specificity, PPV, NPV, LR-, and LR+ of RAD compared with RT-qPCR were 72% (95%CI 62%-81%), 99% (95% CI95%-100%), 99% (95%CI 93%-100%), and 78% (95%CI 70%-85%), 0.28 (95%CI 0.21-0.39), and 72 (95%CI 10-208) respectively. Sensitivity was higher for patients with Ct [less than or equal to] 25 regardless of TSO: TSO [less than or equal to] 4 days 92% (95%CI 75%-99%), TSO > 4 days 100% (95%CI 54%-100%), and asymptomatic 100% (95%CI 78-100%). Overall, combining RAD and RT-qPCR would allow reducing from only 4% the number of RT-qPCR needed. This study highlights the risk of misdiagnosing COVID-19 in 28% of patients if RAD is used alone. A stepwise analysis that combines RAD and RT-qPCR would be an efficient screening procedure for COVID-19 detection and may facilitate the control of the outbreak.