The US greatly undercounted coronavirus cases at the beginning of the pandemic, missing 90% of them – mostly because of a lack of testing, a new study finds.
The research, published in the journal Nature Communications, supports the Centers for Disease Control and Prevention’s own estimates that 90% of cases have been missed.
The United States may have experienced over 6.4 million cases of COVID-19 by 18 April 2020, according to a probability analysis published in Nature Communications. In the same period, there were 721,245 confirmed cases.
By mid-April, the US probably already had more than 6 million cases of coronavirus, the team at the University of California Berkeley estimated. That’s just about what the current official count is now, four months later.
Researchers Jade Benjamin-Chung and colleagues used a statistical method known as Bayesian probabilistic bias analysis to account for incomplete testing and less than perfect test accuracy. They went through actual case counts in each state and accounted for likely undercounts to calculate what the true number of cases should have been.
“We estimate 6,454,951 cumulative infections compared to 721,245 confirmed cases in the United States as of April 18, 2020,” they wrote.
“Accounting for uncertainty, the number of infections during this period was three to 20 times higher than the number of confirmed cases.”
In June, CDC Director Dr. Robert Redfield said his agency had likely undercounted cases by a factor of 10.
They said 86% of the difference was due to incomplete testing, and 14% due to imperfect test accuracy. Incomplete testing was a result of policy.
“For the first few months of the pandemic, the U.S. Centers for Disease Control and Prevention (CDC) recommended that physicians prioritize testing hospitalized patients, who tend to have moderate to severe symptoms,” the team wrote.
“Yet, evidence from studies that conducted broader testing suggest that 30–70% of individuals who test positive have mild or no symptoms and that asymptomatic and pre-symptomatic individuals can transmit SARS-CoV-2,” they added.
“Thus, a substantial number of mild or asymptomatic infections in the U.S. may be undetected.”
Even so, most people in the US have yet to have been infected. “Even in a best-case scenario in which SARS-CoV-2 infection produces immunity for 1–2 years, as is common for other betacoronaviruses, our results contribute to growing consensus that a very small proportion of the population has developed immunity and that the U.S. is not close to achieving herd immunity,” they concluded.