How Dangerous is COVID-19?
How Dangerous is COVID-19? We've been conditioned that it is quite dangerous, but over time better estimates of COVID-19's IFR (Infection Fatality Rate) have been arriving. The CDC has weighed in with an estimate of 0.26%. Telling others about this proved to be highly interesting. Most of us have become so conditioned to the expectation of a much higher rate that the typical initial reaction was to dismiss this as obvious bullshit. In a prior post, an estimate of 0.53% was provided along with the note that due to the nature of the data analyzed at that time this result was likely biased high. Hence, the CDC estimate did not seem surprising to me.
Updated data allows selection of a somewhat less biased data subset selection to estimate the IFR from than that I had previously available. Analysis of this data provided a COVID-19 IFR of 0.27% which is likely still biased on the high side, but much less so. Taking a back of the envelope computational attempt at bias removal involves adjusting for the following two bias inducing scenarios.
1) A person had COVID-19, but one or even multiple tests of this person failed to detect it (potentially due to sampling issues, other test shortcomings or untimely test execution).
2) The person had COVID-19, but was never tested (data subset selection attempted to minimize this effect, but some correction is needed due to data limitations).
My bias corrected estimate for COVID-19's IFR is 0.15%. Perhaps this bias correction is an underestimate or an overestimate, but it is my back of the envelope assessment that COVID-19's IFR is likely somewhat less than the CDC estimate. Don't waste time strongly doubting the CDC estimate which was built on a much more sophisticated analysis methodology than mine; it is somewhere in the right statistical ballpark.
While I estimate that reality may be lower than the current CDC estimate, but only time and better data will answer this more clearly. A parting thought is we should put COVID-19's IFR into context as the better and much narrowed range of estimates arrive. That exercise is left to you the reader. The same statistic(s) can lead us to different conclusions depending on how we personally view risk.
Readers are strongly cautioned that an IFR is only an overall average expectation. It is not independent of age, gender, genetics, disease history and possible preconditions one may have. It is only a very generalized result for the entire population.