Saliva is more sensitive for SARS-CoV-2 detection in COVID-19 patients than nasopharyngeal swabs

Anne Louise Wyllie; John Fournier; Arnau Casanovas-Massana; Melissa Campbell; Maria Tokuyama; Pavithra Vijayakumar; Bertie Geng; M. Catherine Muenker; Adam J. Moore; Chantal B. F. Vogels; Mary E. Petrone; Isabel M. Ott; Peiwen Lu; Alice Lu-Culligan; Jonathan Klein; Arvind Venkataraman; Rebecca Earnest; Michael Simonov; Rupak Datta; Ryan Handoko; Nida Naushad; Lorenzo R. Sewanan; Jordan Valdez; Elizabeth B. White; Sarah Lapidus; Chaney C. Kalinich; Xiaodong Jiang; Daniel J. Kim; Eriko Kudo; Melissa Linehan; Tianyang Mao; Miyu Moriyama; Ji Eun Oh; Annsea Park; Julio Silva; Eric Song; Takehiro Takahashi; Manabu Taura; Orr-El Weizman; Patrick Wong; Yexin Yang; Santos Bermejo; Camila Odio; Saad B. Omer; Charles S. Dela Cruz; Shelli Farhadian; Richard A. Martinello; Akiko Iwasaki; Nathan D. Grubaugh; Albert I. Ko

To determine if saliva performs as well as the U.S. CDC recommendation of using nasopharyngeal swabs for SARS-CoV-2 diagnostics, we collected clinical samples from 44 COVID-19 inpatient study participants ( Table 1 ). This cohort represents a range of COVID-19 patients with severe disease, with 19 (43%) requiring intensive care, 10 (23%) requiring mechanical ventilation, and 2 (5%) deceased as of April 5th, 2020. Using the U.S. CDC SARS-CoV-2 RT-PCR assay, we tested 121 self-collected saliva or healthcare worker-administered nasopharyngeal swabs from this cohort. We found strong concordance between the U.S. CDC "N1" and "N2" primer-probe sets ( Extended Data Fig. 1 ), and thus calculated virus titers (virus copies/mL) using only the "N1" set. From all positive samples tested ( n = 46 nasopharyngeal, 37 saliva), we found that the geometric mean virus titers from saliva were about 5⨉ higher than nasopharyngeal swabs ( p < 0.05, Mann-Whitney test; Fig. 1a ). When limiting our analysis to only patient-matched nasopharyngeal and saliva samples ( n = 38 for each sample type), we found that SARS-CoV-2 titers from saliva were significantly higher than nasopharyngeal swabs ( p = 0.0001, Wilcoxon test; Fig. 1b ). Moreover, we detected SARS-CoV-2 from the saliva but not the nasopharyngeal swabs from eight matching samples (21%), while we only detected SARS-CoV-2 from nasopharyngeal swabs and not saliva from three matched samples (8%; Fig. 1c ). Overall, we found higher SARS-CoV-2 titers from saliva than nasopharyngeal swabs from hospital inpatients. . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.