The number of COVID-19 infections nationwide is six to 24 times higher than the 3.9 million confirmed cases, according to new data from the US Centers for Disease Control and Prevention.

The results of the new study, published this week (July 21) in JAMA Internal Medicine by CDC researchers and state health officials, represent the largest antibody survey of its kind. Even accounting for these hidden cases detected in 10 cities across the US, the findings suggest that many of the cities are nowhere near the antibody prevalence required for herd immunity. Additionally, scientists are unsure just how long people retain their antibodies after being infected, and what that means for immunity to the disease.

“These data continue to show that the number of people who have been infected with the virus that causes Covid-19 far exceeds the number of reported cases,” Fiona Havers, a CDC researcher who led the study, tells The New York Times. “Many of these people likely had no symptoms or mild illness and may have had no idea that they were infected.” 

Researchers are still working to understand SARS-CoV-2, the virus that has killed more than 142,000 people in the US. As many as 40 percent of cases are thought to be asymptomatic, the authors report in an accompanying editorial, while the symptoms brought on by the disease can be highly variable. 

To better understand how COVID-19 moves unseen throughout the country, researchers collected blood samples from more than 16,000 people who had come into a hospital for routine procedures such as cholesterol testing between March and May. They targeted 10 large cities in states including Connecticut, Louisiana, Minnesota, Missouri, New York, Philadelphia, California, Florida, Utah, and Washington. At eight of the 10 sites, they were able to retrieve a second round of samples from later in the spring to track changes in antibody prevalence over time.

The percentage of people with COVID-19 antibodies differed by region, although all 10 sites registered some level of disparity between the seroprevalence data and the reported number of cases. 

Antibody prevalence in New York City, for example, jumped from 6.9 percent in late March to more than 23 percent in May, tracking the rapid increase in cases during this time. The latter tally supports an earlier study carried out by the state that tested more than 15,000 people in April and identified antibodies in 22.7 percent of participants. According to the Times, the discrepancies between seroprevalence data and reported cases in New York City amounted to a 12-fold difference in the first round of sampling, down to a 10-fold difference in early May as testing capabilities improved.

New York was the only site to register such high percentages. In other areas, the antibody prevalence was in the single digits. In the San Francisco Bay Area, for instance, it was 1 percent of samples, while other locations ranged between 2 percent and 6 percent. 

Regions with a lower antibody prevalence sometimes yielded the greatest differences from case reports. In Missouri, where the antibody prevalence stayed around 3 percent, the first round of sampling identified 24 times more cases than had been reported, with that number dropping to 13 times more cases by late May when the state was sampled a second time, NPR reports. 

Taken together, the results further flesh out an earlier report from CDC Director Robert Redfield that cases were being underreported by up to 10 percent. They also suggest that the 700,000 tests carried out each day in the US, as reported by the Times, are not nearly enough to capture the full reach of the pandemic. 

The biggest limitation to the current study is that it relies on old data. Because the samples were collected in early spring, true counts of antibody prevalence and case numbers are likely higher now. Florida, for example, has since seen an explosion of cases that would not be reflected in the data included in this study. In addition, because samples were collected during shelter-in-place orders, the authors admit in their editorial that “the outpatient and inpatient populations included in the study are likely not representative” of the general population.

The authors go on to reject the idea that herd immunity could provide large-scale protection any time soon. While researchers have speculated about the potential for natural herd immunity as more people recover from the virus, the results show that most Americans remain susceptible to COVID-19. Even areas like New York City fall far short of the 60 percent to 70 percent threshold needed for herd immunity to become effective.

“Most of us are likely still very vulnerable to this virus and we have a long way to go to control it,” Jennifer Nuzzo, an epidemiologist at the John Hopkins Center for Health Security who was not involved in the study, tells The Washington Post. “This study should put to bed any further argument that we should allow this virus to rip through our communities in order to achieve herd immunity.”