COVID-19 cases in the United States may be vastly underestimated, according to a study published Wednesday by the UC Berkeley School of Public Health.
From Feb. 28 to April 18, there were 721,245 confirmed COVID-19 cases, according to the study. Using probabilistic bias analysis, the researchers estimated that in reality, the United States may have experienced more than 6.4 million cases during that same time frame.
This disparity in confirmed cases and estimated total cases suggests that 89% of cases were undocumented, largely due to incomplete testing and partially because of limited test accuracy, the researchers concluded.
A lot of the testing currently available in the United States has been primarily reserved for people with moderate to severe symptoms, according to the study.
Despite prioritizing people with more prominent symptoms, studies suggest that 30-70% of individuals who test positive for the coronavirus have mild symptoms or none at all, the study states.
The study further found that when using confirmed case counts or the estimated number of infections, incidents of COVID-19 were highest in the Northeast, Midwest and Louisiana. Conversely, underestimating the number of cases occurred more frequently in California, some southern states and Puerto Rico.
As there is no transmission model incorporated into this research, the findings cannot be used to forecast future rates of infection, according to the study.
Instead, the study provides a more accurate representation of the infection burden at any specific point in time.
“The most basic knowledge used for infection control is estimating the prevalence of the disease,” said study co-author and campus biostatistics professor Alan Hubbard in an email. “The results can be used in several ways, such as to calibrate models to produce information more relevant to the actual state of the disease in the population than the raw counts.”
A previous version of this article incorrectly included information from a source who was referring to a different study.