Quickie
Colonel
There is a time-related distortion if one calculates daily death/daily resolved cases because the average time to discharge for mild cases, the average time to discharge for serious cases and the average time to death are different.
For example, suppose that it takes exactly two weeks for a mild case patient to discharge, 4 weeks for a serious case to discharge and 5 weeks for a patient to die. If you calculate daily death/daily resolved cases, what you're calculating is roughly dying patient with onset 5 weeks ago/(patient with mild onset 2 weeks ago+surviving patient with serious onset 4 weeks ago + dying patient with onset 5 weeks ago). Since the epi curve is going to be a bell curve, this is either an underestimation of the final CFR (before the peak of the bell curve) or an overestimation of the final CFR (after the peak). As mild cases make up the bulk of cases, this distortion is rather severe, but can be minimized if we look instead at death rate for serious cases only. This is of course an extremely simplified model, but it illustrates the idea.
As the makeshift hospitals are closed and mild cases being discharged at a rapid rate, if you calculate "sum of daily death" / ("sum of daily death + daily cured)", expect the rate to increase drastically in coming days. But don't be alarmed, it's just a distortion.
You are assuming a situation where for example 1000 patients suddenly got infected and admitted to the hospital at the same time. In such a case you will see a jump in death rate at around a certain time point (let's say 2 weeks) depending on the type of disease.
In the real world, such a thing doesn't happen. the 1000 patients that are infected that came in the last few days may have caught the virus at very different times during the past 10 days or so. See the difference?
Since you're so detemined, I calculated for Hubei Province using Excel the "sum of daily death" / ("sum of daily death + daily cured)" for 10 days using the table provided by Supercat.
Here is the result. As you can see there is no "distortion". The mortality rate has actually decreased during the last 4 days.
Daily Death __________________________________________________ Daily Cured _________________________________________ Mortality rate
45 | 2492 | 0.017737 |
34 | 2292 | 0.014617 |
42 | 2570 | 0.01608 |
31 | 2410 | 0.0127 |
37 | 2389 | 0.015251 |
31 | 1923 | 0.015865 |
29 | 1487 | 0.019129 |
28 | 1502 | 0.018301 |
27 | 1543 | 0.017197 |
21 | 1422 | 0.014553 |
Total: | 0.161431 | |
Average : | 0.016143 |
The average Mortality rate for Hubei for the 10 days is 1.61% (Compared this to the early days!)
The Mortality rate for Hubei has been decreasing from 1.91% to 1.83% to 1.71% to 1.46% for the last four days.
Last edited: