Editor's Note: The headline for this story was corrected on Aug 5, 2021. It originally said cases were likely 60% higher, which is incorrect. We apologize for the mathematical error.
University of Washington at Seattle researchers estimate that 60% of US COVID-19 cases have gone unreported due to biases in test data and delayed reporting, according to a modeling study yesterday in the Proceedings of the National Academy of Sciences.
The researchers used multiple data sources, including likely unbiased representative random testing surveys from Indiana and Ohio, to estimate the true number of coronavirus infections in both the country and in each state for 1 year starting in March 2020. Their model included likelihood components that combine data on COVID-19 deaths, confirmed cases, and the number of tests administered each day.
Senior author Adrian Raftery, PhD, said in a University of Washington press release that each data source has its own flaws that would give a biased picture of the true prevalence of COVID-19. "What we wanted to do is to develop a framework that corrects the flaws in multiple data sources and draws on their strengths to give us an idea of COVID-19's prevalence in a region, a state or the country as a whole," he said.
Prevalence in US, Indiana, Ohio
About 65 million US residents (19.7%) likely had COVID-19 by Mar 7, the model shows. Until that point, only about 1 of every 2.3 infections had been confirmed, suggesting that about 60% of all infections had been unreported.
"Our results indicate that a large majority of COVID infections go unreported," the authors wrote. "Even so, we find that the United States was still far from reaching herd immunity to the virus in early March 2021 from infections alone. This suggests that continued mitigation and an aggressive vaccination effort are necessary to surpass the herd-immunity threshold without incurring many more deaths due to the disease."
Indiana saw a COVID-19 infection fatality rate (IFR) of 0.84% and a cumulative incidence of 22.9%, representing 1.5 million infections, or nearly one-quarter of the state, as of Mar 7. For every confirmed case, another 2.3 may have gone unreported, the researchers said, although they note that more undercounting was done early in the pandemic.
There had probably been more than 800 COVID-19 cases in Indiana by the time of the first reported case there on Mar 6, 2020. Soon after, Indiana Gov. Eric Holcomb declared a state of emergency, closed schools, ended indoor dining, and issued stay-at-home orders. The first pandemic wave there peaked roughly 2 weeks later, according to the model.
By May 1, 2020, 274,000 infections had likely occurred, in contrast to the 18,630 confirmed cases by that time, for a cumulative incidence of 4.1% and an undercount factor of 14.7.
In Ohio, the model revealed an estimated IFR of 0.83%. The cumulative incidence was 19.5% by Mar 7, and the undercount factor was 2.3. About a year prior, on Mar 9, 2020, Ohio Gov. Mike DeWine had declared a state of emergency and issued stay-at-home orders 2 weeks later. Businesses were allowed to start reopening on May 1, 2020.
Modeling shows that the first wave of infections in Ohio increased in March 2020, probably peaked by late April, stayed at a sustained level throughout the summer, and then picked up again in the fall.
A call for regular state-level random testing
The estimated undercounts, which varied widely across states, could have had several causes, lead author Nicholas Irons, a doctoral student, said in the press release.
"It can depend on the severity of the pandemic and the amount of testing in that state," he said. "If you have a state with severe pandemic but limited testing, the undercount can be very high, and you're missing the vast majority of infections that are occurring. Or, you could have a situation where testing is widespread and the pandemic is not as severe. There, the undercount rate would be lower."
Under the ideal scenario, Raftery said, states would conduct regular random testing of residents to estimate the true COVID-19 prevalence, like Indiana and Ohio did. "We think this tool can make a difference by giving the people in charge a more accurate picture of how many people are infected, and what fraction of them are being missed by current testing and treatment efforts," he said.