Safely return to schools and offices: early and frequent screening with high sensitivity antigen tests effectively identifies COVID-19 patients

Abstract

Background: In-person interaction at school and offices offers invaluable experience to students and benefits to companies. In the midst of the pandemic, ways to safely go back to schools and offices have been argued. Centers for Disease Control and Prevention (CDC) recommends taking all precautions such as vaccination and universal indoor masking. However, even if all the precautions are implemented and transmission is perfectly prevented in the facilities, they may be infected outside of the facilities, which would be a source of transmission in the facilities. Therefore, identifying those infected outside of the facility through screening is essential to safely go back to schools or offices. However, studies investigating the effectiveness of screening are limited. Further, it is not well clarified now which screening strategy (e.g., low or high sensitivity antigen tests, intervals and the number of tests) effectively identify infected and infectious individuals to avoid transmission in facilities Methods: We assessed the effectiveness of various screening strategies in schools and offices through quantitative simulation. The effectiveness was assessed by the proportion of identified infected and infectious participants. Infection dynamics in the facility is governed by transmission dynamics of the population they belong to, and the screening is initiated at different epidemic phases: growth, peak, and declining phases. The viral load trajectory over time for each infected individual was modelled by the viral dynamics model, and the transmission process at the population level was modelled by a deterministic compartment model. The model parameters were estimated from clinical and epidemiological data. Screening strategies were varied by antigen tests with different sensitivity and schedules of screening over 10 days. Results: Under the daily screening, we found high sensitivity antigen tests (the detection limit: 63000 copies/mL) yielded 88% (95%CI 86-89) of effectiveness by the end of 10 days screening period, which is about 20% higher than that with low sensitivity antigen tests (2000000 copies/mL). Comparing screening scenarios with different schedules, we found early and frequent screening is the key to maximize the effectiveness. Sensitivity analysis revealed that less frequent tests might be an option when the number of antigen tests is limited especially when the screening is performed at the growth phase. Discussion: High sensitivity antigen tests, high frequency screening, and immediate initiation of screening are the key to safely restart educational and economic activities allowing in-person interactions. Our computational framework is useful in assessment of screening strategies by incorporating additional factors for specific situations.

Analysis of the risk and pre-emptive control of viral outbreaks accounting for within-host dynamics: SARS-CoV-2 antigen testing as a case study

“…To demonstrate our multi-scale approach in a concrete setting, we focussed on the risk of local SARS-CoV-2 outbreaks. We used nonlinear mixed effects modelling to fit a within-host model that has been used extensively to model SARS-CoV-2 viral dynamics (35)(36)(37)(38)(39)(40)(41)(42) to data from 521 individuals with omicron variant infections (43). The nonlinear mixed effects approach enabled us to quantify the variability in within-host dynamics between individuals which, in turn, could be used to characterise heterogeneity in individual infectiousness profiles (describing how the transmission risk varies during each infection).…”

“…2. By developing symptoms (we assumed previously undetected hosts to be Supposing that an infected individual conducts an antigen test when their instantaneous viral load is ܸ, we assumed (similarly to previous work (40,42)) that a positive test result occurs with probability ‫‬ ା ሺܸሻ ൌ Prob൫ܸ ܸ ‫כ‬ ൯. Here, ܸ represents a measured viral load, assumed to be normally distributed on the log scale (independently of previous viral load measurements) such that log ଵ ሺܸ ሻ ‫‬ ܰሺlog ଵ ሺܸሻ, ߪ ଶ ሻ; ܸ ‫כ‬ is the detection limit (the choice of ܸ ‫כ‬ is described in Supplementary Table 1); and the value of the measurement error level, ߪ, was assumed to be equal to the corresponding quantity that we estimated for PCR testing (Supplementary Table 1).…”

“…Using nonlinear mixed effects modelling, we fitted a within-host viral dynamics model (35)(36)(37)(38)(39)(40)(41)(42) to data from 521 individuals with infections due to the omicron SARS-CoV-2 variant (43). The temporal viral load profile of infected individuals, using population estimates of within-host model parameters (Supplementary Table 1), is shown in Figure 2A.…”

“…We used a simple within-host model of SARS-CoV-2 viral dynamics (35)(36)(37)(38)(39)(40)(41)(42), given by…”

“…We derive an equation satisfied by the outbreak risk under a multi-scale model, and verify our analytically derived outbreak risk estimates using simulations of an individual-based stochastic outbreak simulation model. Focussing on the case study of SARS-CoV-2, we characterise the viral dynamics by fitting a within-host model (35)(36)(37)(38)(39)(40)(41)(42) to data from 521 individuals with infections due to the omicron variant (43). We first consider the outbreak risk in the absence of interventions, before exploring the extent to which the outbreak risk can be mitigated through regular rapid antigen testing of the entire local population.…”