Here’s the coronavirus death toll through April 21. Naturally, you’re all eager to hear my Solomon-like wisdom on what to do about the Swedish numbers, aren’t you? Let’s review:
- If I use the Johns Hopkins daily numbers for Sweden, they are enormously noisy thanks to Sweden’s lack of reporting on weekends. On the other hand, they’re consistent with the way every other country is reported.
- If I use the numbers from the Public Health Agency of Sweden, they’re corrected to show the actual day the death occurred. However, these numbers are updated every day to reflect new data and it takes a few days before they settle down. This means that numbers from the most recent few days are always lower than reality.
What to do? Two things. First, instead of a 6-day rolling average, I’ve switched to a 7-day rolling average. This means every dot encompasses a full week, including both weekdays and weekends. This smooths things out a little bit. Second, I’ve added the official Swedish numbers as a gray line, but I don’t include the most recent five days.
How’s that for sawing the baby in half?
One other note: the United States is already up to 40,000 deaths and appears to have plateaued, rather than peaked. At this point, even if we start to decline soon, it will probably be a long, slow decline and the total number of deaths will reach very close to 100,000. That’s a big change from last week’s optimistic assessment of 50-60,000.
The raw data from Johns Hopkins is here. The COVID Tracking Project is here. The Public Health Agency of Sweden is here.
