1/ Finally some demographic information on the South Korean #COVID19 cases, as 3/4/2020. Thread analyzing this data vs. China's.
First, here is a plot of cases vs. mortality rate by age.
First, here is a plot of cases vs. mortality rate by age.
2/ Earlier, I had done a similar plot based on data released by the China CDC as of 2/11 and also base-lined this data against China's general population pyramid.
3/ Some caveats and observations. Relative to the onset of its crisis, South Korea's data is earlier than China. By 2/11, gross new cases were already on the decline in China:
4/ Whereas, South Korea's cases are still on the rise. Here are its cases over the last six days where the rate of growth of confirmed infections appears to still be constrained by the pace at which they are increase lab-test output.
5/ While CFRs are lower for South Korea than China (e.g. 15% for >80 in China, only 6% in South Korea), this is most likely because many more Koreans (especially elderly) will unfortunately pass away in the coming days. Getting more data on severe/critical would help normalize.
6/ Another difference is the spike among 20-29 year olds in South Korea's data. I posit that this is not because 20-somethings in Korea are any more likely to catch the disease than China's, but highlights differences in South Korea's testing approach and unique circumstances.
7/ Specifically, Korea has concentrated its initial effort on the Shincheonji mega-cluster in Daegu, which still accounts for the majority (56%) of the confirmed cases. I believe they have largely completed testing of this cluster, and are now focusing on smaller ones.
8/ By contrast, China's was more spread out and did not feature any mega-clusters a large as Shincheonji. As such, its testing strategy was based more on pure priority - to get people into the right clinical treatment pathway based on symptoms or epidemiological history.
9/ Based on the data, I posit that a significant proportion of Shincheonji's tested population were younger adults in the 20-29 range, accounting for that bump. As South Korea tests more people outside this mega-cluster, the curve should look more like China's.
10/ I would also posit that a significant number of Shincheonji's population were females. China's data was evenly split between genders (51.4% male). Korea's initial dataset is 62.5% female.
11/ Adjusting for these differences, Korea's newly released demographic data largely corroborates the data that came out of China a month ago, and should some additional confidence to its validity.
12/ As more demographic data is released in other countries, + updated data from China and Korea, we should start talking about mortality rates on an age-adjusted basis. Some countries are older and all things equal, their potential rate should correspondingly be higher.
13/ For example, Italy and Japan's age-adjusted mortality rate would be lower than India's. However, age is clearly just one factor at play - others are quality of healthcare, speed of response, intensity of crisis + this idea of binary scenarios (i.e. Wuhan vs. Non-Wuhan).
14/ For more on the binary scenario concept:
15/ And more recently, here:
