Routine saliva testing for the identification of silent coronavirus disease 2019 (COVID-19) in healthcare workers

Abstract

Objective: Current COVID-19 guidelines recommend symptom-based screening and regular nasopharyngeal (NP) testing for healthcare personnel in high-risk settings. We sought to estimate case detection percentages with various routine NP and saliva testing frequencies.

Design: Simulation modeling study.

Methods: We constructed a sensitivity function based on the average infectiousness profile of symptomatic coronavirus disease 2019 (COVID-19) cases to determine the probability of being identified at the time of testing. This function was fitted to reported data on the percent positivity of symptomatic COVID-19 patients using NP testing. We then simulated a routine testing program with different NP and saliva testing frequencies to determine case detection percentages during the infectious period, as well as the presymptomatic stage.

Results: Routine biweekly NP testing, once every 2 weeks, identified an average of 90.7% (SD, 0.18) of cases during the infectious period and 19.7% (SD, 0.98) during the presymptomatic stage. With a weekly NP testing frequency, the corresponding case detection percentages were 95.9% (SD, 0.18) and 32.9% (SD, 1.23), respectively. A 5-day saliva testing schedule had a similar case detection percentage as weekly NP testing during the infectious period, but identified ~10% more cases (mean, 42.5%; SD, 1.10) during the presymptomatic stage.

Conclusion: Our findings highlight the utility of routine noninvasive saliva testing for frontline healthcare workers to protect vulnerable patient populations. A 5-day saliva testing schedule should be considered to help identify silent infections and prevent outbreaks in nursing homes and healthcare facilities.

Keywords: COVID-19; case detection; nasopharyngeal; outbreak; saliva; testing.

Fig. 1.

Distribution of mean case-detection percentages during the infectious period using biweekly nasopharyngeal (A) and saliva (C) testing. Distribution of mean case-detection percentages during the infectious period using weekly nasopharyngeal (B) and saliva (D) testing. The red line indicates the mean of the distribution, and the box plot represents the interquartile range (IQR) with whiskers extending the range from minimum (25th percentile minus 1.5 IQR) to maximum (75th percentile plus 1.5 IQR). The density on the y-axis is the number of experiments from 500 iterations (Monte-Carlo simulations) that resulted in a mean case detection shown on the x-axis.

Fig. 2.

Distribution of mean case-detection percentages during the presymptomatic stage using biweekly nasopharyngeal (A) and saliva (C) testing. Distribution of mean case-detection percentages during the presymptomatic stage using weekly nasopharyngeal (B) and saliva (D) testing. The red line indicates the mean of the distribution, and the box plot represents the interquartile range (IQR) with whiskers extending the range from minimum (25th percentile minus 1.5 IQR) to maximum (75th percentile plus 1.5 IQR). The density on the y-axis is the number of experiments from 500 iterations (Monte-Carlo simulations) that resulted in a mean case detection shown on the x-axis.