Recently, some areas have begun to test the new coronavirus antibodies in the serum of residents. Because people infected with the virus do not necessarily have symptoms, and most people seek virus testing only when they have symptoms, estimating the prevalence of the virus by the number of confirmed cases is not only inaccurate, but also overestimates the infection mortality rate. Many public health experts therefore advocate that antibody testing is a better way to estimate prevalence. On April 14, 17 experts and scholars published an antibody detection study in Santa Clara County, California on the medRxiv website . This site publishes preprints that have not been peer-reviewed, and this study received particular attention because one of its authors, a prominent Stanford University School of Medicine professor, has criticized tradition with frequency in recent years.
John PA Ioannidis, famous for his method of statistical testing. The results of the study found that the weighted prevalence estimate was 2.8%, meaning that 54,000 residents in Santa Clara County had been infected with the virus, which was 54 times the official number of confirmed cases in the county at the time! But the study has been criticized for using a voluntary Wedding Photo Editing sample, not a random sample, which has tarnished Ioannidis' reputation. On May 18, one of the authors of the above report, a research team led by USC scholar Neeraj Sood , published an article in the Journal of the American Medical Association (JAMA) to estimate the prevalence of the new coronavirus using antibody test results. research letter.
This letter provides what is likely to be the first study to be published in a major medical journal using random samples for antibody testing. In mid-April, a random sample of 863 residents in Los Angeles County, California, underwent a lateral flow immune assay test. The test results showed that 35 subjects tested positive, accounting for 4.06% of all subjects tested. After weighting by demographic variables, sensitivity and specificity, the positive rate was adjusted to 4.65%, and its 95% confidence interval was 2.52% to 7.07%. Based on an estimated prevalence of 4.65%, 367,000 adults in the county should have been infected with the new coronavirus, which is 43.5 times the 8,430 confirmed cases in the county at the time. As for the infection fatality rate, the