Estimating the number of people who have been infected is harder still. Most infected people are never formally diagnosed and never become one of the "cases" in the news. The limitations of the tests and the difficulty of attracting a representative population to be tested make it hard to estimate the true number of infections. The preferred test (reverse transcription polymerase chain reaction-based tests) uses RNA technology to see if the virus is present in nasal or oral swabs. It is a good test, but still may miss infections in up to 30% of cases.
A second type of test uses blood samples to look for an antibody called immunoglobulin (Ig)G that implies the person was previously infected. Based on IgG test results, the CDC assumes that 5% to 8% of the population has been infected. That would mean 24 million Americans have already had COVID-19 or a very similar illness. That is more than 10 times the number of confirmed cases.
The number is consequential: a higher infection rate for the same number of deaths implies that the virus is less deadly. A review by a prominent epidemiologist considered 23 population studies with sample sizes of at least 500 people and found the percentage who have positive antibodies ranged from 0.1% to 48% -- a 480-fold difference. Although the study was robustly criticized and at odds with highly cited, peer-reviewed research, it has appeared in over 30 news outlets, and the range of estimates allows people to pick a number that justifies their political position.
