Re:Age, Sex, Existing Conditions of Covid-19 Cases and Deaths
by mrb on Thursday February 27, 2020 @11:45AM
(#22434907)
You are exactly correct. With the formulas you described:
naïve CFR = deaths / cases
resolved CFR = deaths / (deaths + recoveries)
It can be shown pretty dramatically in outbreak simulations how more accurate the resolved CFR is than the naïve CFR. See for example the chart https://blog.zorinaq.com/assets/outbreak2.svg In that specific example, the naïve CFR underestimates by 5x, while the resolved CFR is only off by ~1.3x. I ran this simulation as part of my blog post on how to estimate the Case Fatality Ratio of the coronavirus: https://blog.zorinaq.com/case-fatality-ratio-ncov/
naïve CFR = deaths / cases
resolved CFR = deaths / (deaths + recoveries)
It can be shown pretty dramatically in outbreak simulations how more accurate the resolved CFR is than the naïve CFR. See for example the chart https://blog.zorinaq.com/assets/outbreak2.svg In that specific example, the naïve CFR underestimates by 5x, while the resolved CFR is only off by ~1.3x. I ran this simulation as part of my blog post on how to estimate the Case Fatality Ratio of the coronavirus: https://blog.zorinaq.com/case-fatality-ratio-ncov/