ABSTRACT
Purpose Saliva has been considered a suitable sample material for SARS-CoV-2 testing but uncertainty remained regarding the sensitivity and reliability of different saliva collection methods for community mass testing. This study aimed to investigate the potential utility of expectorated saliva (ES) and drooled saliva (DS) through a large cohort study. Methods ES and DS samples were collected in a two-stage non-randomized prospective cohort study. Their utility for SARS-CoV-2 RNA qRT-PCR testing was assessed by comparison with results for combined throat and nose (CTN) swabs. A total of 2,878 subjects were recruited, from which 2,747 were evaluable for statistical analyses. Results Using CTN swab-based results as reference,DS- and ES-based tests showed the same high level of concordance (98% vs 98%) or specificity (99% vs 99%). Sensitivity seemed to be higher for DS than for ES (93% vs 80%) but not significantly once viral concentration was taken into account. Multivariable analysis indicated however an inferior sensitivity of saliva-based testing for female compared to male subjects with low viral burden. Assuming no false positive qRT-PCR results, an unbiased comparison showed no significant difference in sensitivity between saliva- and swab-based testing. Conclusion SARS-CoV-2 RNA testing based on saliva showed high diagnostic accuracy and can be considered an alternative where swabbing may not be tolerated or operationally feasible. Drooled saliva yielded the same diagnostic performance compared to expectorated saliva and may present a preferred option with reduced aerosol risk and increased compliance. Observed sex-specific difference in detection performance however warrant further investigations.
ABSTRACT
Background Antigen lateral flow devices (LFDs) have been widely used to control SARS-CoV-2. Changes in LFD sensitivity and detection of infectious individuals during the pandemic with successive variants, vaccination, and changes in LFD use are incompletely understood. Methods Paired LFD and PCR tests were collected from asymptomatic and symptomatic participants, across multiple settings in the UK between 04-November-2020 and 21-March-2022. Multivariable logistic regression was used to analyse LFD sensitivity and specificity, adjusting for viral load, LFD manufacturer, setting, age, sex, assistance, symptoms, vaccination, and variant. National contact tracing data were used to estimate the proportion of transmitting index cases (with [≥]1 PCR/LFD-positive contact) potentially detectable by LFDs over time, accounting for viral load, variant, and symptom status. Findings 4131/75,382 (5.5%) participants were PCR-positive. Sensitivity vs. PCR was 63.2% (95%CI 61.7-64.6%) and specificity 99.71% (99.66-99.74%). Increased viral load was independently associated with being LFD-positive. There was no evidence LFD sensitivity differed between Delta vs. Alpha/pre-Alpha infections, but Omicron infections were more likely to be LFD positive. Sensitivity was higher in symptomatic participants, 68.7% (66.9-70.4%) than in asymptomatic participants, 52.8% (50.1-55.4%). 79.4% (68.6-81.3%) of index cases resulting in probable onward transmission with were estimated to have been detectable using LFDs, this proportion was relatively stable over time/variants, but lower in asymptomatic vs. symptomatic cases. Interpretation LFDs remained able to detect most SARS-CoV-2 infections throughout vaccine roll-out and different variants. LFDs can potentially detect most infections that transmit to others and reduce risks. However, performance is lower in asymptomatic compared to symptomatic individuals. Funding UK Government.