BackgroundThe continued need for molecular testing for SARS-CoV-2 and potential for self-collected saliva as an alternative to nasopharyngeal (NP) swabs for sample acquisition led us to compare saliva to NP swabs in an outpatient setting, without restrictions to avoid food, drink, smoking, or tooth-brushing.MethodsA total of 385 pairs of NP and saliva specimens were obtained, the majority from individuals presenting for initial evaluation, and were tested on two high-sensitivity RT-PCR platforms: the Abbott m2000 and Abbott Alinity m (both with limits of detection [LoD] of 100 copies of viral RNA/mL).ResultsConcordance between saliva and NP was excellent overall (Cohen’s κ=0.93), for both initial and followup testing, for both platforms, and for specimens treated with guanidinium transport medium as preservative as well as for untreated saliva (κ=0.88-0.95). Viral loads were on average 16x higher in NP specimens than saliva specimens, suggesting that only the relatively small fraction of outpatients (∼8% in this study) who present with very low viral loads (<1,600 copies/mL from NP swabs) would be missed by testing saliva instead of NP swabs, when using sensitive testing platforms. Special attention was necessary to ensure leak-resistant specimen collection and transport.ConclusionsThe advantages of self-collection of saliva, without behavioral restrictions, will likely outweigh a minor potential decrease in clinical sensitivity in individuals less likely to pose an infectious risk to others for many real-world scenarios, especially for initial testing.Key pointsSaliva has comparable sensitivity and specificity to nasopharyngeal swabs for RT-PCR-based COVID-19 testing (concordance, κ=0.93; n=385 participants), albeit with slightly lower recovery of viral RNA. Treatment with a readily available guanidinium preservative within 24 hours of sample collection improves recovery.