When computing a fatality rate for COVID-19, it's important to have a reasonable handle on the number of deaths caused by COVID-19 and the number of people infected with the COVID-19 virus.
Apparently Mark Levine, Chair of New York City Council health committee, is claiming that a number of COVID-19 deaths in NYC are going unreported because there is insufficient testing resources to test everyone who dies in the city.
Ok, that could cause the number of deaths to be underreported, at least for places that have insufficient test resources
However, it is also apparent that NYC is counting anyone who dies and had COVID-19 as a COVID-19 death. Having the virus and having the virus cause death are two different things.
This would cause the number of deaths to be overreported. (I suppose this could be adjusted by considering the number of elderly people that normally pass away from other causes.)
Disease modelers at the University of Texas at Austin believe that the number of actual COVID-19 cases could be higher than reported (by as much as a factor of 10). Insufficient testing means a number of people could have the virus but don't know it.
If true, this would mean that the fatality rate is lower than it appears to be.
Colorado's health website (
https://covid19.colorado.gov/case-data) states that they also count people who are "highly-likely" to be infected but haven't been tested but that number is a very small portion of the total case count.
We could probably find many other examples of poor data-collection that would skew the fatality rate calculation.
It doesn't take much messing with the numerator and the denominator to create wild variations in the computed/estimated fatality rate.
But I don't want to cherry-pick in either direction.