Background: Two rapid antigen tests (RATs) for COVID-19 targeting the nucleocapsid protein of SARS-CoV-2 were compared with real-time RT-PCR as the reference method.
Methods: Ninety-six nasopharyngeal swab samples, comprising 56 positive and 40 negative samples confirmed through rRT-PCR were collected and retested to determine the reliability of the two nasopharyngeal RATs.
Results: The overall sensitivity and specificity of both RATs were 64.3% (95% confidence interval 50.4 - 76.6%) and 100% (95% confidence interval 91.2 - 100%), respectively. Cohen's kappa coefficient of agreement of both RATs to rRT-PCR was 0.600 (95% confidence interval 0.457 - 0.743) (p < 0.001), showing almost perfect agreement when the Ct values were less than 25 in rRT-PCR. A significant difference in Ct values between true positives and false negatives was observed (Mann-Whitney-Wilcoxon test; p < 0.001).
Conclusions: Compared to rRT-PCR, RATs have fewer false negatives. In suspected COVID-19 cases, negative RAT results should be retested using either RAT or rRT-PCR.