ABSTRACT
SARS-CoV-2 diagnosis is a cornerstone for the management of coronavirus disease 2019 (COVID-19). Numerous studies have assessed saliva performance over nasopharyngeal sampling (NPS), but data in young children are still rare. We explored saliva performance for SARS-CoV-2 detection by RT-PCR according to the time interval from initial symptoms or patient serological status. We collected 509 NPS and saliva paired samples at initial diagnosis from 166 children under 12 years of age (including 57 children under 6), 106 between 12 and 17, and 237 adults. In children under 12, overall detection rate for SARS-CoV-2 was comparable in saliva and NPS, with an overall agreement of 89.8%. Saliva sensitivity was significantly lower than that of NPS (77.1% compared to 95.8%) in pre-school and school-age children but regained 96% when considering seronegative children only. This pattern was also observed to a lesser degree in adolescents but not in adults. Sensitivity of saliva was independent of symptoms, in contrary to NPS, whose sensitivity decreased significantly in asymptomatic subjects. Performance of saliva is excellent in children under 12 at early stages of infection. This reinforces saliva as a collection method for early and unbiased SARS-CoV-2 detection and a less invasive alternative for young children.