Far more people were infected with the novel coronavirus than previously reported in several corners of the U.S., according to data released Tuesday by the Centers for Disease Control and Prevention.
The agency conducted a survey looking at antibodies to the virus in 10 U.S. regions. It found prevalence was highly variable from one region to the next, but far higher than the reported number of cases across the board. The presence of antibodies in the blood is evidence that a person’s body had mounted an immune response to the coronavirus.
Included in the data was a study conducted in late April by the Indiana University Fairbanks School of Public Health at IUPUI that estimated that nearly 2.9% of Hoosiers either had the virus or previously had the virus. The findings, released preliminarily in May, were published Tuesday in Morbidity and Mortality Weekly Reports, a scientific publication of the CDC.
The study indicated that 186,000 people in Indiana were estimated to be infected or previously infected with COVID-19 at the end of April, far more than actual testing, which showed 17,000 cases in the state at that time.
In the New York City metropolitan area, the CDC estimated, based on samples collected in March and April, that 6.9% of the population had contracted the virus, a level that would be equivalent to at least 12 times the number of reported cases. In the San Francisco Bay area, samples collected in April showed a far smaller percentage of people contracted the virus, at 1%. But that was still nine times the reported case load in the region.
Assessing how many people have antibodies—what’s known as seroprevalence—helps shed light on how many infected people either didn’t show signs of the virus, or didn’t have serious enough symptoms to seek medical care. The data can help determine not only how widespread the virus is, but how deadly it is.
The CDC’s study suggests the numbers of such people could be vast. But it also indicates that even the hardest-hit locales are nowhere near the 60% infection level needed to establish so-called herd immunity.
Many infected people don’t develop symptoms, and the CDC findings suggest a need for more testing to detect outbreaks and contain the virus’s spread. At present, the U.S. is conducting between 700,000 and 800,000 tests a day, according to the Covid Tracking Project. That number has grown significantly in recent weeks, but is still a fraction of what experts suggest is needed.
The CDC’s study is the largest of its kind to date. Researchers analyzed blood samples to look for antibodies in people who either had routine clinical tests or were hospitalized. The study published in the Journal of the American Medical Association on Tuesday expands on early data from six cities and states released in June. Researchers also posted more data for eight regions to the CDC’s website. The agency said it plans to test about 1,800 samples collected from each of the 10 regions and update the data every three to four weeks.
The data shows that some regions of the country have been much worse hit than others. It also showed how the gap between estimated infections and reported cases narrowed as testing and reporting improved. In New York City, for example, the 12-fold difference reported based on samples from March and April became a 10-fold difference a few weeks later.
Previously, seroprevalence studies have attracted criticism for flawed methodology and widely varied findings. A Stanford University study of Santa Clara County, Calif., published before peer review raised particular concern after concluding that more than twice as many people were infected with COVID-19 than counts at the time suggested.
Antibody tests have also faced questions about their accuracy. In April, the White House recommended that to more accurately assess the spread of COVID-19, the U.S. should use multiple antibody tests at once. In its survey, the CDC notes that potential false positives are a limitation of the study.