The WHO’s estimate of the COVID 19 case fatality rate is 3.4 percent. According to John Ioannidis, the estimate is inaccurate. His reasoning is simple but comes from basic epidemiology principles; it should come from a random sample of a population and an accurate estimate of the number of infections in the host population.
He uses Diamond Princess’s 700 passengers as the closest closed sample to calculate the estimate although this small population cannot be considered as a random sample. According to him, the case fatality rate of this population is 1 percent (7 deaths/700 passengers); this population is largely elderly as against the general population in any country. If we extrapolate this onto the general population, the case fatality rate should be much less than 1 percent.
He then projects this rate onto the US age structure and estimates the case fatality rate as 0.125 percent. Since the previous 1 percent is not at all a random sample, he stretches the confidence interval from 0.025 percent (five times lower) to 0.625 (five times higher) percent. Adding extra uncertainties such as underlying chronic medical conditions, his final estimate varies from 0.5 percent to 1 percent.