The difference implies that the virus is much less deadly than it looks, but it also makes contact tracing a daunting challenge.
Newly published antibody test results from half a dozen parts of the country confirm that COVID-19 infections in the United States far outnumber confirmed cases. The ratio of estimated infections to known cases in these studies, which the U. S. Centers for Disease Control and Prevention (CDC) reported on Friday, range from 6 to 1 in Connecticut as of early May to 24 to 1 in Missouri as of late April.
These results confirm something we already knew: The COVID-19 infection fatality rate—deaths as a share of all infections—is much lower than the crude case fatality rate—deaths as a share of known cases. That is bound to be true when testing is limited and a virus typically produces mild or no symptoms. At the same time, the CDC’s antibody studies imply that efforts to control the epidemic through testing, isolation, quarantine, and contact tracing will not be very effective, since they reach only a small percentage of virus carriers.
The CDC analyzed blood samples drawn for routine tests unrelated to COVID-19 from patients in New York City, Connecticut, South Florida, Missouri, Utah, and western Washington state. Although these samples may not be representative of the general population, they provide a clearer picture of virus prevalence than screening limited to people who sought virus tests because they had symptoms consistent with COVID-19 or because they were in close contact with known carriers.
In New York City, where the samples were drawn from March 23 through April 1, nearly 7 percent tested positive for COVID-19 antibodies, implying that infections outnumbered reported cases during that period by 12 to 1. The prevalence estimated by a state-sponsored antibody study conducted from April 19 to April 28 was three times as high, although the ratio of estimated infections to known cases (about 11 to 1), was similar. The difference in estimated prevalence can be at least partly explained by the spread of the virus between early and late April.
The gap between the two estimates may also be partly due to differences between the samples used in the studies. The CDC study was based on patients whose doctors ordered routine blood tests, while the New York State Department of Health study used blood drawn from randomly selected shoppers. Infections might have been unusually common among people who ventured out to stores during the study period, either because they were more likely to encounter carriers or because they had already recovered from COVID-19 and therefore felt safe leaving their homes. (Then again, the health department study would have missed people who were self-isolating because they had symptoms or because they had close contact with people who had COVID-19.)
In South Florida, where the samples analyzed by the CDC were collected from April 6 through April 19, almost 2 percent tested positive. That is just one-third the antibody prevalence that University of Miami researchers found in a random sample of Miami-Dade County residents about a week later.
