1 Most of what I'm about to write is no different from what I've said and done in the past. However, I have been asked the same question repeatedly, so I would like to reiterate it. We have received many inquiries from overseas as well, so we should have prepared the same content in English, but due to time constraints, I'm afraid I'll have to skip it. This article is designed to be read without basic knowledge of infectious diseases and jargon, but it is rather difficult to understand. Please forgive me for that.
2 The fact that the number of COVID-19 reports in Japan is very low compared to other countries is attracting attention from home and abroad. Is it true? It has been pointed out that the number of tests is so small that we may be misreading the actual number of infected people.
3 However, this point is wrong at various layers. In the first place, Japan does not aim to capture all the numbers of COVID-19. Whether it's administrative testing or insured care, the state basically has a testing strategy in mind to diagnose, hospitalize, and isolate critically ill patients who need to be hospitalized. It is natural that they "haven't figured it out" and they don't intend to. That's not a bad thing.In fact, the situation is the same in every country, large or small, and no country, whether in the United States, Europe, or Asia, is aiming to "capture the whole number.
The WHO is not asking for such a thing. But instead, Japan gives PCR to asymptomatic returnees and isolates asymptomatic test-positive people in hospital (wasn't it home for people with minor illnesses?). It has not been coherent in its principles. So, people get anxious because "we're not sure what they want to do". It's a failure in the press.
4 The difference between Korea and Japan is the "result" and not the "purpose". In South Korea, where the number of infected people had surged in one place, we had to focus on inspections in and around the area. If such a phenomenon (let's call it an overshoot) occurs in Japan, the number of inspections will increase. When the situation is different, arguing only on the basis of the number of tests without observing the situation is like trying to say, "That team made 50 sliding tackles while this team made only one," without watching a football game. In games where you don't have to slide (e.g., when you're in possession the whole time), even 0 times isn't a "mistake," and of course 50 times isn't a mistake.
There are many diseases for which the total number of patients is not known. In Japan, we do not have a "total" number of influenza cases, but only a fixed-point observation. Because that's enough information, both epidemiologically and in terms of infection control. There is no accurate data on how many cases of the common cold occur each year in Japan. It's also a mistake to say that you can tell by looking at the receipt data, because many cold patients (like me) don't see a doctor and wait until they are cured naturally. Not only in medicine, but also in economics and political science, data are mostly based on sampling to estimate population numbers, and "whole numbers" is an inefficient way of grasping the situation.
6 We have not seen the devastation in Japan as in Italy, Spain or New York City. There is no medical collapse in a critically ill patient, no use of the operating room as an ICU, no piling up of bodies on a skating rink with no place to put them. Even if the "numbers" are not known, it is a fact that the current situation in Japan (including Tokyo) is much better controlled than in other countries.
7 Even so, you may be interested in "Well, what about the actual situation? There are estimates. For example, Dr. Hiroshi Nishiura and his group estimate that the number of mild illnesses in Japan may be twice the reported number. The catch rate is 0.44, with a 95% confidence interval of 0.37-0.50.
8 Although the study was based on data from China, there is no guarantee that the Chinese COVID-19 demographic is the same as the Japanese one. Also, since the original study did not include asymptomatic patients or those with minor illnesses that did not require hospitalization, the number of infected patients estimated on that basis would inevitably be an underestimate. If you are more paranoid, it's not unreasonable to believe that "the Japanese and Chinese viruses are different because of the mutation" (although I don't think so).
9 This does not diminish the value of the paper itself. The model must always use existing parameters, and it is often impossible to prove the external validity of these parameters. If the underlying parameters are not reasonable, the predictions will not be correct. A model assumes a simplified world insofar as it is a model. A model without simplification, which is an adjectival contradiction.
To complain about these "assumptions" of the mathematical model is like complaining, for example, "You can't explain disease B," when a randomized controlled trial is conducted for disease A. This is a meaningless tirade against the honor of the industry.
A mathematical model that assumes a certain hypothesis should have internal academic validity, but it is the responsibility of the reader, as a resident of the real world, to appraise it in the real world.
Just as the RCT findings for disease A should not be used for disease B, it is natural to understand the limitations of the mathematical model and to be careful when applying it to the real world. For example, it would be wrong to read the paper and conclude that the total number of infected people in Tokyo is about 500 as of March 26.
11 People make mistakes. The models are also wrong. Being wrong is not a big deal. The problem is to notice your mistakes and make corrections. Already, a group at Imperial College London has admitted that its original estimate that the peak of the infection should be moderated was "wrong" and has revised its prediction that the ICU will soon fail if it does not fight the virus fairly aggressively.
"Ladies and gentlemen,
迷える子羊は stray の方をつかうのでそっちに合わせてもいいかも．
↑ hygge ↓