Widely available and reliable testing for SARS-CoV-2 is essential for the public health response to the COVID-19 pandemic. We estimated the diagnostic performance of reverse transcription PCR (RT-PCR) performed on saliva and the SD Biosensor STANDARD Q antigen test performed on nasopharyngeal swab compared to the reference standard, nasopharyngeal swab (NP) RT-PCR. We enrolled participants living and/or seeking care in health facilities in North Lima, Peru from November 2020 to January 2021. Consenting participants underwent same-day RT-PCR on both saliva and nasopharyngeal swab specimens, antigen testing on a nasopharyngeal swab specimen, pulse oximetry, and standardized symptom assessment. We calculated sensitivity, specificity, and predictive values for the nasopharyngeal antigen and saliva RT-PCR compared to nasopharyngeal RT-PCR. Of 896 participants analyzed, 567 (63.3%) had acute signs/symptoms of COVID-19. The overall sensitivity and specificity of saliva RT-PCR were 85.8% and 98.1%, respectively. Among participants with and without acute signs/symptoms of COVID-19, saliva sensitivity was 87.3% and 37.5%, respectively. Saliva sensitivity was 97.4% and 56.0% among participants with cycle threshold (CT) values of ≤30 and >30 on nasopharyngeal RT-PCR, respectively. The overall sensitivity and specificity of nasopharyngeal antigen were 73.2% and 99.4%, respectively. The sensitivity of the nasopharyngeal antigen test was 75.1% and 12.5% among participants with and without acute signs/symptoms of COVID-19, and 91.2% and 26.7% among participants with CT values of ≤30 and >30 on nasopharyngeal RT-PCR, respectively. Saliva RT-PCR achieved the WHO-recommended threshold of >80% for sensitivity for the detection of SARS-CoV-2, while the SD Biosensor nasopharyngeal antigen test did not. IMPORTANCE In this diagnostic validation study of 896 participants in Peru, saliva reverse transcription PCR (RT-PCR) had >80% sensitivity for the detection of SARS-CoV-2 among all-comers and symptomatic individuals, while the SD Biosensor STANDARD Q antigen test performed on nasopharyngeal swab had <80% sensitivity, except for participants whose same-day nasopharyngeal RT-PCR results showed cycle threshold values of <30, consistent with a high viral load in the nasopharynx. The specificity was high for both tests. Our results demonstrate that saliva sampling could serve as an alternative noninvasive technique for RT-PCR diagnosis of SARS-CoV-2. The role of nasopharyngeal antigen testing is more limited; when community transmission is low, it may be used for mass screenings among asymptomatic individuals with high testing frequency. Among symptomatic individuals, the nasopharyngeal antigen test may be relied upon for 4 to 8 days after symptom onset, or in those likely to have high viral load, whereupon it showed >80% sensitivity.
Keywords: COVID-19; Peru; RT-PCR; SARS-CoV-2; antigen; diagnosis; nasopharyngeal swab; saliva; validation.