Background In Japan, COVID-19 vaccination of the general population began on 12 April 2021. While published studies have relied primarily on confirmed case data to estimate the direct population-level impact of the primary vaccination series in Japan, serial cross-sectional seroepidemiological studies were also conducted, allowing asymptomatic and undiagnosed infections to be addressed. The present study aimed to estimate the direct population impact of vaccination while explicitly accounting for ascertainment bias and estimating the true total number of SARS-CoV-2 infections. Methods From 13 January 2020 to 24 May 2022, surveillance data and seroepidemiological data were systematically collected and analyzed. A statistical model was developed to describe transmission dynamics over time while estimating the total number of infections based on seroprevalence data. The number of infections directly averted by vaccination was estimated by comparing a counterfactual scenario without vaccination to the estimated total number of infections under the factual scenario with vaccination. Results During the study period, the estimated total number of infections was 47,313,032 (95% confidence intervals (CI): 37,703,409–57,933,448). Of these, —38,692,320 infections (95% CI: 29,082,697–49,312,726) —were estimated to have remained undiagnosed. In the absence of vaccination, an additional 16,941,593 infections (95% CI: 12,674,530–21,566,082) were estimated to have occurred. The estimated direct impact of vaccination was 2.3 times larger than a previously published estimate based solely on confirmed case data. Conclusions Through direct effects alone, vaccination reduced the number of SARS-CoV-2 infections by more than 25% compared with the counterfactual scenario without vaccination. Combining surveillance and seroepidemiological data enables more precise evaluation of vaccination programs by explicitly addressing ascertainment bias.
Nishiyama et al. (Mon,) studied this question.