How to Estimate the Mortality Risk of COVID-19: A New Approach with a Three-Factor Decomposition.

Statistical analysis of COVID-19 mortality is challenging due to its non-stationarity and cross-sectional instability. In this paper, the authors introduce a unified method to evaluate the fatality rate of COVID-19 across countries, whose method provides more reliable information for cross-country comparison than the traditional case-fatality rate（CFR）. It emerges that the new method, the blockwise case-fatality rate（BCFR）, varies for different countries and in different periods. The authors also decompose the COVID-19 fatality data by three factors: 1) The virus infection dynamics over population in different countries, 2) pure distribution and evolution of instantaneous death rate attributed to different individual’s physical characteristics such as age and health, and 3) individual countries’ variations affecting interactions between the virus infection and the instantaneous mortality due to individual’s physical characteristics. Based on the new three-factor model, the authors obtain six key findings of the COVID-19 fatality rate. Our study suggests that, on average, the estimated instantaneous fatality rate contributes about 57.0% to the global BCFR while the time-varying weight contributes about 41.5% in December 2020. The country-specific contribution of instantaneous fatality rate is significantly higher than that of the time-varying weight. Besides, the country-specific characteristics in demographical, social, and economic aspects would affect the relative severity of the disease.