Web-based real-time risk assessment of coronavirus disease 2019 infection in schools and social dining settings

Abstract

Background: Schools and dining situations are associated with a high risk of indoor transmission of coronavirus disease 2019 (COVID-19). Performing risk assessment in real time could enable organizers to adjust the duration and size of indoor activities depending on the epidemic situation.

Methods: The per-hour transmission rates of COVID-19 from a single infector in school and social dining settings were estimated from COVID-19 surveillance data in Japan from January to February 2022 using a mathematical model. We then developed a web application that calculates the risk of COVID-19 infection in those settings, accounting for place of residence, vaccination history, duration of indoor activity, and the number of participants.

Results: The estimated per-hour transmission rates were 0.01934 (95 %CrI: 0.01939-0.01947) in social dining settings and 0.00324 (95 %CrI: 0.00323-0.00325) in school settings. Accounting for the epidemiological risk of having infected persons in indoor settings, a web application was devised to compute the risk of a single participant contracting COVID-19 at the event. Web application users can vary input variables including the duration and the number of participants, thereby benefiting the real time risk reduction.

Conclusions: The per-hour transmission rate was higher in social dining settings compared with that in school settings, but the greater number of participants that is typical of gatherings at schools might offset this gap in per-hour per-infector transmission risk. The proposed web application can act as an important tool for promoting risk awareness regarding COVID-19 in high-risk settings.

Keywords: Aerosol; COVID-19; Indoor transmission; Japan; Web application.

Fig. 1

Model of weekly estimates of universal and sentinel COVID-19 counts beyond May 8, 2023 in Tokyo, Osaka, Nara, and Okinawa. Fig. 1: Infection counts and immune proportion by age group at the national level. Panel (A) shows the COVID-19 infection counts in 0–9, 10–19, …, 80+ age groups by date of infection during the study period, which was estimated by back-calculation of the case counts by date of reporting using the delay distributions from infection to reporting. Panel (B) shows the estimated fraction of the population that have protection against symptomatic COVID-19 infection in 0–9, 10–19, …, 80+ age groups.

Fig. 2

Visual appearance of the web application's input sections for users. Users are expected to give inputs such as prefectures, type of location, number of participants, duration of the gathering, recent exposure to COVID-19 or vaccination. The user can also adjust the multiplier for the infection hazard that is used in the counterfactual scenario.

Fig. 3

Visual appearance of the web application's output section. On the basis of user inputs such as prefecture, type of location, number of participants, duration of the gathering, recent exposure to COVID-19 or vaccination, the probability of infection per gathering is calculated. The probability of infection when the same gathering is held 1–5 times is shown in the plot on the left (red line), together with a green line that represents the probability of infection in the “what if the number of users or the duration of the gathering was halved?” scenario. On the right side of the plot, supplementary results including the expected number of infections in the gathering are shown as outputs.

Fig. 4

Visual appearance of the web application's output section with results from counterfactual incidence rate. Visual appearance of the web application's output section with the same input as that in Fig. 3, but with additional results on the plot representing the “what if the population-level incidence rate was 10 times higher?” counterfactual scenario.