CFR is case fatality rate
Eric Feigl-Ding @ericding
WHY 🦠 CFR MATH IS HARD: Uncertainties of naive case count estimates. Graph shows how the ratio of the number of confirmed deaths and case counts changed over time. Case counts are corrected for 3-fold or 30-fold under-reporting and diagnostic delay 🧵:
smw.ch/article/doi/smw.2020.20203
2019-Novel Coronavirus (2019-nCoV): estimating the case fatality rate – a word of caution
2) ...In addition, example of diagnosis taken 2, 4, and 7 days prior to the date of the count of confirmed death. The latter is meant to illustrate the effect of the delay between diagnosis and death or recovery. This actual delay is likely longer than one week.
3) Implication is that while a) underdiagnosis will mean the real CFR is lower, b) a longer delay from diagnosis to death/recovery will mean the real CFR is higher. Hence there is a tug of war between A vs B above for estimating true CFR. That’s why we can’t know true CFR yet
4) Authors argue: “The higher case fatality rate reported from Wuhan may be overestimated. The true number of exposed cases may be vastly underestimated. With a focus on serious cases, mild/asymptomatic courses might remain largely unrecognized, in particular during flu season.”
4) Authors continue: “Under-detection of mild or asymptomatic cases may be further fueled after further growth of the outbreak, as healthcare-facilities and testing capacities in Wuhan have reached their limits.”
5) On flip side, authors argue: “The lower case fatality rates outside Wuhan may be underestimated... As the epidemic arrived later in other regions and countries, there may be a delay of fatal cases arising and their reporting...
6) Moreover, the authors argue that “The low number of documented recovered cases might indicate that days and weeks can pass until death occurs. Hence, the numbers, e.g. Guangdong with 970 cases and no death occurring, might be false low because severe cases might still [die].”
7) Finally, “Case fatality rates may truly differ among different regions of the world. Differences in CFR may be caused by differences in medical care during a large epidemic versus care for single cases.” Special thx to first author @MBattegay for publishing this great piece!
8) For extended discussion, @DellAnnaLuca has a detailed thread 🧵 on this CFR issue. The LAG again is much longer than just 1 week to resolution of the virus 🦠 unfortunately. Bottom line, don’t go by the naive 2% mortality so far.
t.co/9xFMSjuOZs
The report states the following which needs to be kept in context here:
One intriguing aspect of the outbreak so far is the discrepancy between the estimates of the case fatality rate reported from Hubei province, from different regions of China and from other countries. As of February 7, 2020, 30’536 have been confirmed. Thereof, 22’112 occurred in the Hubei province of China with a death toll of 619 (= 2.8%). This contrasts with 16 deaths among 8’702 recorded cases in other regions of China and further countries, suggesting at first glance a case fatality rate of 0.18%. The uncertainties and spatio-temporal variation discussed above could explain this divergence
In terms of the discussion of 'vulnerable people being dispensible' this is a fundamental misunderstanding of the point being made.
I'll try and explain :
In a normal flu outbreak x number of patients were going to die and were expected to die. Let's call this the natural rate of expected flu related deaths.
What we need to work out is how much higher NCP is compared to this. You count all the people who would have died in a normal season and discount them from the calculation. What we want to know is the number of people extra who you would not expect to die in a normal outbreak.
As the thread above illustrates, at present we really don't know the CFR and therefore the extra number of deaths in an NCP outbreak - if indeed any.
Especially since there is this massive difference between the report rates in Wuhan compared to outside.
When talking about vulnerable patients with pre existing conditions, I primarily am referring to those who fail into the 'natural rate' of flu deaths. People who would have died anyway from normal flu.
Beyond that, at present we don't know how many additional deaths there have been compared to a normal flu event.
That's the question that still needs to be explored -- it could be very close to the natural death rate expected, in which case anxiety of a high cfr will be very misplaced. Or it could be massively out and the death rate substantially higher. And there might be significant differences between different areas for a variety of reasons too.
The point being, it's not about the number dead but the percentage of deaths 'that are unexpected '
Until we have a clearer idea, we need to stop getting quite so worked up about what Wuhan means for us in the UK. Are we both apples or are we apples and pears?