A team of student and faculty investigators at Yale University has devised a new method for calculating COVID-19 fatality rates in the United States that it says will better guide the country through its reopening, and more accurately drive health care policy decisions regarding the pandemic. Fatality rates have been difficult to calculate in regions without extensive testing due to many infected individuals not being detected.
The team first examined fatality rates in Germany, where testing is prevalent, after noticing that the low fatalities rates Germany reported early in the epidemic were rapidly increasing. The researchers attributed this discrepancy to the delay between patients becoming infected, and later either dying or recovering from the disease.
After correcting the German fatality rate the investigators examined whether it was similar to six other countries, including South Korea and Iceland, with similar extensive testing. Although they found considerable variation in the overall fatality rate (0.6 – 5.0%) they also found that 89% of it could be explained just by the percentage of cases 70 years and older due to their much higher fatality rate.
Based on the consistency of their findings they used the age specific fatality rates to calculate the fatality rate in regions with less extensive testing. Among their calculations, based on the age distribution of COVID-19 cases, was a fatality rate of 1.8% for New York City. This rate was higher than previous estimates with the lowest value being recently reported by the United States CDC (0.1%).
“We were surprised by how low these values were. We then tested our finding by dividing the number of COVID-19 deaths by the fatality rate to calculate the actual number of infected individuals in the city, and compared it to recent random measurements of infections in the NYC adult population,” said Douglas Rothman, PhD, a professor in the Radiology & Biomedical Imaging Department at Yale School of Medicine (YSM).
“Our infection percentage calculation (between 14.7% and 22% of the adult population) showed an excellent match with the data (15.3% to 19.9%). In contrast previous fatality rates overestimated the infection percentage by between 2- and 18-fold.
“We believe these findings may partially explain the failure of present models to accurately predict both fatalities and duration of the COVID-19 outbreak in the U.S.,” said Rothman.
The study’s first author is Jessica Rothman, a PhD student in the Department of Biostatistics at Yale School of Public Health and Douglas Rothman’s niece. The other authors are Theodore Holford, PhD, Susan Dwight Bliss Professor Biostatistics at YSM; David Eidelberg, MD, a professor at the Institute of Molecular Medicine, Feinstein Institutes for Medical Research and the Donald and Barbara Zucker School of Medicine at Hofstra/Northwell; and Samantha Rothman, a math/computer science major at Tulane University.
“Julia Rothman, the youngest sister of Jessica and Samantha, assisted in creating the database used for the analysis. Additionally, we had substantial help from friends, family and colleagues from throughout the country – all who are passionate about helping fight the COVID-19 pandemic,” Douglas Rothman said.