Coronavirus - Fatality Numbers & Confounders!
1) Selection bias.
We have limited testing capacity. Hence we have to decide who and where to use it best.
Unfortunately, we decided to use it in the hospitals and geographic locations where problem is more severe.
Sometimes we are testing just the very sick that have already made their way to busy hospital where no else wants to be in.
Hence we are sample is self-selected by their own illness or we have selected to test just the ones who have more severe disease.
in order to correct this bias we have to now revert to testing people who have milder or no symptoms. Ideally we should be testing the entire US population!
As of the day of this blog writing 4/26/2020 approx 1.6% of US population has been tested.
2) Observer bias.
How do we count covid-19 deaths?
Are we "accidentally" attributing people who were otherwise going to be classified as "flu and pnuemonia" deaths to the novel coronavirus?
Some hospitals have been found to favor covid-19 as cause of death for financial and reimbursement reasons.
Say a person suffers a severe heart attack and end up in hospital for it for 2 weeks. During those 2 weeks he catches coronavirus because hospitals have high rate of this virus. He subsequently dies of arrhythmia (complication of heart attack) would he be recorded as a covid19 death or cardiovascular cause of death?
Death certificates have certain rules. If you examine a blank certificate on the internet you will find that covid-19 cannot even be classified as "immediate cause of death" and will only be mentioned as one of the several conditions "contributing" to the death. Yet death certificates containing the word covid-19 be labeled anything else but coronavirus deaths?