Date of reporting, not date of death
Update 3 April, about 2:30pm — looks like we may have some more sensible reporting in place, on the day of actual death. Have yet to explore it in detail, but see https://www.england.nhs.uk/statistics/statistical-work-areas/covid-19-daily-deaths/
Update 4 April, 11am — Guardian article following up on all this — https://www.theguardian.com/world/2020/apr/04/why-what-we-think-we-know-about-the-uks-coronavirus-death-toll-is-wrong
So, this to me is astonishing, I am baffled.
We have various graphs plotting deaths from COVID-19, and we (everyone, from the experts to the media to the public) is looking at the curves and saying things like “ooh, that many deaths today, makes us two weeks behind Italy”.
But, that’s not the case. Well, maybe, but maybe not. We’re noting the deaths when they are centrally reported, not when those patients actually died.
Today, Thurs 2 April 2020, NHS England in their press release announced 561 newly-recorded deaths. See screenshot
These are daily announcements. The data includes cases reported to NHS England over the previous 24 hours. (I think just England at that stage of the day, not including the rest of the UK)
I had assumed that the vast majority of these (for example, from today) 561 deaths happened in the previous 24 hours. Fair to assume some lag time for reporting and clarifying the occasional cause of death where it’s not clear, so thus you’d expect figures would be skewed slightly.
NHS England send out this press release each day. I am unsure if it gets made public (which I also find strange, but maybe it is out there somewhere), so alas I am not going to reproduce it here. Within the press release, there’s unhelpfully-structured lines of text with each trust, their reported numbers of deaths, and the date of the death (it’s not handily in a table, even the date formatting differs slightly).
Today, I took the dates of death, and spent an hour manually counting them.
84 deaths were from 1 April, the day before the press release.
42 deaths from 31 March
158 deaths from 30 March
129 deaths from 29 March
And the rest from earlier dates, going back to 16 March.
So, there is a significant and inconsistent lag time in reporting. Trusts are under huge pressures so I can understand delays in getting high-profile data accurate and sent off.
But, what baffles me is
a) I believe the style of reporting is going to change. But why are NHS England reporting the deaths in this manner at all? Why not simply collate the days of death and update lag data as it comes through? At the moment, we’re simply seeing the administrative data of when it gets reported. That’s not terribly interesting, and I say that as a full-on nerd who looks at all kinds of dull numbers. There’s a pandemic going on out there, there’s really interesting stuff everywhere you look.
b) Why are the media reporting the daily data as ‘new deaths’, when much of it appears not to be new deaths? They have been repeatedly challenging the government over, for example, not scaling up testing to higher levels. Why is there this acceptance of the daily numbers, when they are (presumably?) also seeing these press release numbers.
c) And what are the actual numbers of deaths per day. I know there’s a thirst for real-time reporting, but why this?! And what are the actual numbers, even if its from last week? Are we mirroring Italy?
d) And also, how are other countries reporting their numbers? Like this? Or the day of death?
So, I just don’t get it. Maybe I’ve really missed something. Maybe my brain has shut down for easter. If I have got it wrong, please do say, I am very happy to be corrected about this.
But for now, I remain puzzled.
