I see a lot of people looking at the Tokyo Govt website and concluding "Tokyo is only testing 300 people a day or "1/3 to 2/3 of people in Tokyo are testing positive for coronavirus". While the testing numbers are still very low compared to other capitals, this is not the case 1/

The figures given under 検査実施人数/Number of people tested don't include tests conducted at medical facilities covered by insurance. They don't currently give the breakdown but when they have, these make up the majority of tests 2/ https://twitter.com/ryutakahashi217/status/1251054549898391552?s=20

However, the positive cases are INCLUDED in the 陽性患者数/number of confirmed cases figures on a daily basis regardless of how the test was conducted. This means it's not possible to get a test-positivity rate from this data as they're not measuring the same thing. 3/

The other tests are added to the 検査実施件数/number of tests conducted data every Friday through to the previous Wednesday (e.g. now updated to April 22 and will stay that way until Friday). The discrepancy is large -- e.g. April 276 people tested but 1,607 tests conducted. 4/

One caveat: the tests conducted figure would also include some people who were tested twice, which is one of the requirements to be released. It's not clear how many people that would be. 5/

Yes, this a very unhelpful way for Tokyo to present the data and presumably is a quirk of the way its collated. The testing done is still very low compared to other capitals, so there's no need to exaggerate it. On a national basis Japan is doing about 5000-7000 tests/day. 6/end