Pooled Saliva Specimens for SARS-CoV-2 Testing

We evaluated saliva (SAL) specimens for SARS-CoV-2 reverse transcriptase PCR (RT-PCR) testing by comparison of 459 prospectively paired nasopharyngeal (NP) or midturbinate (MT) swabs from 449 individuals with the aim of using saliva for asymptomatic screening. Samples were collected in a drive-through car line for symptomatic individuals ( = 380) and in the emergency department (ED) ( = 69). The percentages of positive and negative agreement of saliva compared to nasopharyngeal swab were 81.1% (95% confidence interval [CI], 65.8% to 90.5%) and 99.8% (95% CI, 98.7% to 100%), respectively. The percent positive agreement increased to 90.0% (95% CI, 74.4% to 96.5%) when considering only samples with moderate to high viral load (cycle threshold [ ] for the NP, ≤34). Pools of five saliva specimens were also evaluated on three platforms, bioMérieux NucliSENS easyMAG with ABI 7500Fast (CDC assay), Hologic Panther Fusion, and Roche Cobas 6800. The average loss of signal upon pooling was 2 to 3 values across the platforms. The sensitivities of detecting a positive specimen in a pool compared with testing individually were 94%, 90%, and 94% for the CDC 2019-nCoV real-time RT-PCR, Panther Fusion SARS-CoV-2 assay, and Cobas SARS-CoV-2 test, respectively, with decreased sample detection trending with lower viral load. We conclude that although pooled saliva testing, as collected in this study, is not quite as sensitive as NP/MT testing, saliva testing is adequate to detect individuals with higher viral loads in an asymptomatic screening program, does not require swabs or viral transport medium for collection, and may help to improve voluntary screening compliance for those individuals averse to various forms of nasal collections.