Looking at daily fatalities for Coronavirus, there’s a pretty stable pattern of there being reduced deaths on Sunday and Monday:
It’s sufficiently stable that, by this point, I think it’s fair to say that it’s not just noise that coincidentally looks meaningful.
I was trying to think of why this pattern might pop out.
Option 1 would be that there’s some deficit in hospital care those days. If so, that would suggest that it should be corrected.
But that is “if so”. Another option that I can think of is that it’s simply an artifact of when people were most likely to contract the disease. If people mostly catch the disease during a weekday and mostly suffer for 8, 15, 22, or 28 days before their demise, then we would see this effect.
Any other thoughts? Any reason to prefer some option over another?
I doubt that it has much to do with when people are actually dying, it’s when those deaths get officially certified and recorded. The people responsible for doing this don’t always work on weekends.
You’ll also find ‘catch up’ and reclassified data too - sometimes a couple of months lagging, and you will only notice it if you are maintaining your own records - and often this modified data will be corrections over a period of time and not ascribed to dates - making it a little less easy to make predictions, however there is likely to be enough noise not to be too significant
I will add that this now has a huge caveat, the Pubs and Trump have decided to take control of the data management using their own underlings instead of the CDC to compile or oversee the process.
There is a huge danger in this, not only for the democratic process but also if the data is significantly compromised and Republican governors and mayors decide to use it as basis for infection control there is going to be unwarranted reductions in lockdowns and control measures which will result in a massive surge.
If that surge does happen then it will give these false accounters even more reason to falsify the figures resulting in even less effective measures being taken.
Tuesday and Wednesday traditionally have the highest amount of deaths from the flu and pneumonia. It may be similar with COVID. Heart attacks are Monday.
Perhaps deathly ill people give up after the weekend? Get visits from loved ones on Saturday, make peace with God on Sunday, then drift off…
While the exact process varies from state to state, there are always several layers of bureaucracy handling the death reporting. Somebody at the hospital has to process the paperwork first before it goes to some government office, either the county or the state. Information processed at the hospital on Friday will (usually) be available for the health department to release on Saturday; then many (but not all) hospital clerical staff are out for the weekend, and what they process on Monday is then available to the health department on Tuesday. Meanwhile, information for people who die at home or at a non-hospital facility (nursing home, hospice, jail, etc.) will get processed by another route, which may or may not take longer.
Note, for example, that Worldometer already has a figure for today’s deaths in Arizona as I write this (134), although it is still well before noon in that state. Look at the Arizona Dept of Health Services dashboard: 134 deaths reported today, but the chart of “COVID-19 deaths by date of death” shows only 2 on the 20th and none since then.