SMU Pom Mom wrote:JasonB wrote:When calculating the death rate, please be aware of what you are using as the numerator and denominator.
The numerator is number of deaths, easy enough.
The denominator, however, is NOT the number of cases. The hundreds of thousands of folks who have tested positive in the last two weeks haven't had the 6+ weeks it takes to die (for the small percentage who will die).
The denominator, in fact, should be the number of recoveries. Deaths divided by recoveries tells you the percentage of people who die from the disease.
One tiny correction: The denominator should be the sum of deaths and recoveries, i.e., total completed cases, not just recoveries.
Than you, yes, great correction.
Ponyboy, I haven't changed any of the formulas.
T0- when you are infected.
T1 - when you start to show symptoms - 1 to 2 weeks later on average
T2 - If you have to go to the hospital - 2 weeks after you show symptoms.
T3 - if you die - 2 to 3 weeks after you go to the hospital. Plus, in Texas, they are doing additional confirmation of Covid as the cause of death, which delays death reporting an additional 5 days.
When you look at the graph of the "number of cases" and how they rise - the majority of testing is still being done at point T2. More T1 than before, but the very high positivity rate indicates that T0 and T1 aren't happening as much as we would like.
When you are looking at the graphs, look at the number of active hospitalizations. You will see a gradual increase of hospitalizations from last week of May until mid June, and then acceleration of hospitalizations at that point forward.
For deaths, because we don't do analysis on the weekends, they tend to burst up and down, peaking on Wednesday each week. So, have your graphs illustrate the 7 day rolling average - it will give you a better trend line to follow than the up and down of individual days. If you look at that graph, you will show a gradual increase in deaths starting in mid June (3 weeks after the increase in hospitalizations) until last week, when (3 weeks after the accelleration of hospital cases) we saw a significant increase in deaths.
Hopefully, that helps. But taking the current number of deaths divided by current number of cases isn't accurate, because you haven't given about 100,000 cases a chance to die yet. Right now in Texas we have had since the beginning 274,712 cases.134,953 are considered to be active cases. So your rate right now is 3,340/(274,712-134953). Which has dropped since the other day, thankfully, to 2.3%.
The same thing happens with the flu, BTW. If you try and calculate a death rate from the flu in January, when cases are peaking, you are going to get a really invalid number if you divide deaths by the total number of cases. Flu doesn't have as long of a drawn out hospitalization before deaths, so the difference between the numbers isn't as bad, but it still would be inaccurate at that time.
Now, the Covid death rate thing comes with two HUGE caveats: 1) the denominator is based on the number of people tested. As we all know, there are a lot of non-symptomatic or very minor cases with this thing, where people never get tested. Same thing happens with the flu, but I would say the percentage of people who get Covid and don't have symptoms and won't get tested is much higher than the flu because we know there is a good number of minor cases. So there are problems with the denominator. 2) If you look at the CDC Excess Deaths counts, in Texas, we have been running above the normal expected deaths since April, and by those counts, we are closer to 4K excess deaths. so, our total deaths number from Covid might be a little bit off.
Hopefully that helps!