Diagnostic accuracy of three prevailing rapid antigen tests for detection of SARS-CoV-2 infection in the general population: cross sectional study

ABSTRACT

ObjectiveTo assess the diagnostic accuracy of three rapid antigen tests (Ag-RDTs) for detecting SARS-CoV-2 infection in the general population. DesignCross-sectional study with follow-up using pseudonymised record linkage. SettingThree Dutch public health service COVID-19 test sites. ParticipantsConsecutively included individuals aged 16 years and older presenting for SARS-CoV-2 testing. Main outcome measuresSensitivity, specificity, positive and negative predictive values of BD-Veritortm System (Becton Dickinson), PanBio (Abbott), and SD-Biosensor (Roche Diagnostics), applying routinely used sampling methods (combined oropharyngeal and nasal [OP-N] or nasopharyngeal [NP] swab), with molecular testing as reference standard. For SD-Biosensor, the diagnostic accuracy with OP-N sampling was also assessed. A viral load cut-off ([≥]5.2 log10 SARS-CoV-2 E-gene copies/mL) served as a proxy of infectiousness. ResultsSARS-CoV-2 prevalence and overall sensitivities with 95% confidence intervals were 188/1441 (13.0%) and 129/188 (68.6% [61.5%-75.2%]) for BD-Veritor, 173/2056 (8.4%) and 119/173 (68.8% [61.3%-75.6%]) for PanBio, and 215/1769 (12.2%) and 160/215 (74.4% [68.0%-80.1%]) for SD-Biosensor with routine sampling, and 164/1689 (9.7%) and 123/164 (75.0% [67.7%-81.4%]) for SD-Biosensor with OP-N sampling. In those symptomatic or asymptomatic at sampling, sensitivities were 72.2%-83.4% and 54.0%-55.9%, respectively. With a viral load cut-off, sensitivities were 125/146 (85.6% [78.9%-90.9%]) for BD-Veritor, 108/121 (89.3% [82.3%-94.2%]) for PanBio, 160/182 (87.9% [82.3%-92.3%]) for SD-Biosensor with routine sampling, and 118/141 (83.7% [76.5%-89.4%]) with OP-N sampling. Specificities were >99%, and positive and negative predictive values >95%, for all tests in most analyses. 61.3% of false negative Ag-RDT participants returned for testing within 14 days (median of 3 days, interquartile range 3) of whom 90.3% tested positive. ConclusionsThe overall sensitivities of the three Ag-RDTs were 68.6%-75.0%, increasing to at least 85.6% after the viral load cut-off was applied. For SD-Biosensor, the diagnostic accuracy with OP-N and NP sampling was comparable. Over 55% of false negative Ag-RDT participants tested positive during follow-up.