Estimating the Continuously Evolving COVID-19 Case-Fatality Ratio in the United States using a Time-Delay Correcting Algorithm.

Abstract

The COVID-19 pandemic has placed an extreme healthcare burden across the global community, and new population-based analyses are needed to identify successful mitigation and treatment efforts. The objective of this study was to design a computational algorithm to estimate the time-delay between a peak infection and associated death rate, and to estimate a measurement of the daily case-fatality ratio (D-CFR). Daily infection and death rates from January 22, 2020 through April 15, 2021 for the United States (US) were downloaded from the US Center for Disease Control COVID-19 website. A Savitzky-Golay filter estimated the moving time average of each data sequence with 5 different window-sizes. A locally-designed inflection point identification algorithm with a variable length line-fitting sub-routine identified peak infection and death rates, and quantified the time-delay between a peak infection and subsequent death rate. Although filter window-size did not affect the time-delay calculation (p = 0.99), there was a significant effect of fitting-line length (p < 0.001). A significant effect of time-delay length was found among three infection outbreaks (p < 0.001), and there was a significant difference between time-delay lengths (p < 0.01). A maximum D-CFR of approximately 7% occurred during the first infection outbreak; however, starting approximately 2.5 months after the first peak, a significant negative linear trend (p < 0.001) in the D-CFR continued until the end of the analyzed data. In conclusion, this research demonstrated a new method to quantify the time-delay between peak daily COVID-19 infection and death rates, and a new metric to approximate the continuous case-fatality ratio for the ongoing pandemic.