5

I was looking at the Daily Deaths from COVID-19 from Worldometers and I noticed that the number is always lower on Sundays and Mondays.

enter image description here Why is that? Am I interpreting the graphs correctly?

5
  • This isn't exactly a duplicate because this question asks why there are peaks mid-week rather than dips on weekends, but it's essentially the same observation. I'll leave it to the community to decide if it's a duplicate. – Carey Gregory May 23 '20 at 17:18
  • 4
    Twitter joke rather than a serious reference: First rule of epidemiology: weekend data doesn't count. Clinics and offices are closed on the weekends. Folks put off medical care and record keeping may be delayed. – Charles E. Grant May 23 '20 at 17:49
  • 1
    How can that be? So you come to work on Tuesday (you skip Monday?) and then your staff says: "one hundred people died on Sunday, what do we do?" and you say "Ah, never mind, just distribute some numbers on Monday, some of them on Sunday and we'll catch up". I mean, if they are closed, the numbers should be zero, yet, they add some numbers in there. Doesn't make any sense :-) Good twitter joke, though – CuriousPaul May 23 '20 at 18:30
  • 1
    @CuriousPaul They are the days reported, not the days of the deaths, which means they can just report the number and don't have to go back and revise all the previous numbers. Sources like worldometers are using all the data available, so they don't wait for state or federal agencies to report totals, they use local level data when it's reported. If some locales report on the weekend then they have partial counts. – Bryan Krause May 23 '20 at 19:04
  • the linked question is about cases and test results more than deaths. People can put off getting tests or going to the doctor on the weekend, but I don't think they put off dying. – Kate Gregory May 24 '20 at 12:53
3

While hospitals, clinics, long term care homes and such are of course open and working seven days a week, administrative staff typically only work 5 days a week. So someone comes to work Monday and reports 3 deaths (that may actually have occurred on Saturday or Sunday.) They may not do that first thing Monday, they may have other things going on.

Meanwhile whoever they report to has some cut off for when "things you tell me today" go into "today's number." Maybe it's noon, I have no idea, it could be 5pm, whatever. So some of the things that are reported to that person on Monday get reported to the big country-wide reporting mechanisms Tuesday.

As a result you will see two days with less deaths reported, and those two days are more likely to be Sunday and Monday than Saturday and Sunday. Sure, if the graphs were "deaths that happened on day X" and not "deaths we were told about on day X" this wouldn't happen. However in normal times, not all deaths are observed (never mind reported) on the day they happen: people who die alone at home and are discovered later, for example, or people who are suspected of having died in a house fire but it takes several days to find some remains, that sort of thing. The norm has been to count them on the day they are reported.

3

While people don't usually have a preference for dying on particular days determining how they died is subjected to delays. A medical examiner or other expert had to determine whether the death is from covid-19 or one of the many comorbid conditions that the person suffered from. Furthermore, there are deaths in the community that need to be examined. You can see that actual numbers change over time with hindsight, and the weekend factor introduces a delay in reporting.

One of the most pernicious ways has to do with the time it takes to tabulate deaths due to Covid-19, especially in jurisdictions where hospitals are overworked. In New York City, for example, some deaths are reported to the city’s health department within hours, but others take days — sometimes a week or more — to get tabulated. This delay means that on any given date, when officials announce the daily death toll, they do so having received only a fraction of the death reports that will eventually come in.

On the afternoon of April 1, for example, New York City officials declared that 1,374 people had died of the virus to date. But the accounts of deaths kept coming in. Two days later, the tally of people who had died on or before April 1 had climbed to more than 1,700. By April 5 it stood at 1,878. As of April 9, the official count was 2,253, a roughly 60 percent increase over the initial tally. This number will likely continue to rise, slowly, for several more days.

https://undark.org/2020/04/11/covid-19-data-lag-deaths-new-york

https://www.livescience.com/how-covid-19-deaths-are-counted.html

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.