Photo: Ariana Cubillos, AP
For more coverage, visit our full coronavirus section here.
Medical experts have long assumed that the United States is significantly underreporting the actual number of COVID-19 infections across the country due to limited testing and a high number of asymptomatic cases.
Large-scale antibody testing should give researchers a sense of the scale of the epidemic, and preliminary results from the first such test in Santa Clara County suggest that we are underreporting cases of at least minus 50 times.
In early April, researchers from Stanford University conducted an antibody test on 3,300 county residents, recruited via targeted Facebook ads. The researchers hoped to create a representative sample of the county’s population by selecting individuals based on their age, race, gender and postal code to extrapolate the results of the study to the wider community.
The results of the study are preliminary and are not peer reviewed, but the overall conclusions seem to strongly contribute to the notion that a large number of cases of COVID-19 have gone undetected.
Due to questions about the effectiveness of antibody testing, the researchers adjusted the performance characteristics of the test using data from the test manufacturer and a sample of controls tested at Stanford University. Again, the results are preliminary and the study was not peer reviewed, but the researchers found a crude and unadjusted antibody prevalence of 1.5%, which was increased to 2.5 -4.2% when adjusting for population and test characteristics.
Researchers estimate that while 2.5 to 4.2 percent of the county has already been infected, the actual number of total cases in early April – active and cured – ranges from 48,000 to 81,000. The county reported just under 1,000 cases at the time of the study, which would mean that the cases are underreported by a factor of 50 to 85.
“Our results suggest that there are somewhere between 50 and 80 times more infections in our county than is known by the number of cases reported by our public health department,” said Dr. Eran Bendavid , the Stanford professor who led the study, told ABC News.
If the study figures are correct, the actual death and hospitalization rates for COVID-19 are both significantly lower than current estimates, and due to the lag between infection and death, researchers project a rate real mortality between 0.12 and 0.20. The results also suggest that the county is far from “collective immunity”, as scientists estimate that 50-60% of the population should be infected so that the virus cannot spread anywhere.
However, the study’s authors caution against extrapolating the results from Santa Clara County to the rest of the country. The county reported its first case on January 31 and was one of the first national “hot spots” where the virus had caught on. The researchers also recognize other limitations of the study, including an overrepresentation of white women aged 19 to 64, and “other biases, such as bias favoring healthy individuals able to visit our sites test, or bias favoring those with a history of COVID-type diseases seeking confirmation of antibodies. “
In San Mateo County, health worker Scott Morrow made a similar estimate Monday evening.
“I am reluctant to give you the following figures, because first they are a guess, and second because some will think they are too low to act,” he wrote in a message to the community. “My best guess is that about 2-3% of the CMS population is currently infected or has recovered from the infection. This represents approximately 15 to 25,000 people and they are throughout the county and in each community. I don’t believe that number is shifted by a factor of 10, but it could be shifted by a factor of 2 to 3. “
At the time of its announcement, the county had 699 confirmed cases of the virus, which means that if Morrow’s figures are correct, the county identified only 2.7-4.5% of its total cases.
MORE CORONAVIRUS COVERAGE:
Sign up for The Daily newsletter for the latest coronavirus news here.
Eric Ting is an SFGATE digital reporter. Email: [email protected] | Twitter:@_ericting