Pandemic deaths analysis

I did some analysis of deaths over this pandemic (because I’m morbid like that).

I calculated expected “normal” deaths for every week of 2020, based on 10-year-average age-specific death rates for each specific week of the year applied to the most recent age-specific populations. I also calculated an upper and lower significance threshold as twice the standard deviation. The graphs below compare these with the actual deaths from all causes (so far), in different age groups.

During the first wave of the coronavirus pandemic, excess deaths (defined by me in relation to expected deaths as calculated above) reached 0.98 of a month’s worth of extra deaths in England and Wales. The tally has since come down to 0.86 of a month for 2020 so far, as deaths have been a little lower than expected after the peak. In Scotland, the excess reached 0.57 of a month, and has since come down to 0.47 of a month’s worth.

The recent lower-than-expected death rates seem to indicate that the deaths of some very old or unwell people were brought forward just a little by COVID-19. Certainly it’s in the older populations that the recent death rates are low enough to reach significance.

Conversely, deaths in the 0-44 age range barely reached significant elevation during the peak. In Scotland, the death rate in that age group has not really come down since April and is still skirting significance. This can’t be due to the coronavirus directly but may be due to effects of the lockdown including reduced access to health services. NRS report that a good chunk of the excess deaths over this year were due to causes other than COVID-19, including dementia, genitourinary diseases and diabetes.

England and Wales analysis:

Scotland analysis:

If anyone reading this can help me with data for other countries, I’d love to expand the above analysis.

(Apologies, the y-axis label in each graph should just read “Deaths”, not “Deaths or cases”.)

This entry was posted in Coronavirus, science and tagged , , . Bookmark the permalink.

5 Responses to Pandemic deaths analysis

  1. Roger Watt says:

    I’d be curious to see this data as cumulative graphs. By eye it looks like the two phases – over and then under the expected – are about equally long just now. But the under phase has maybe just 10-15% of the amplitude of the over phase. If so, in the absence of any new wave that means that cumulative graphs might converge after about 50 weeks.

    And, the expected has some interesting anomalies. Week 14-15 – is this Easter? That inflection looks like people hang on for an extra week (or don’t get reported?) Then at Christmas/Hogmanay (wk 51 – 02) is that an anomalous dip in the middle, or is it peaks either of the holiday itself. Does your data have cause of death? Can you separate out accident and suicide from illness?

    • Sarah says:

      Good idea, I’ll do the cumulatives in my next update.
      I’m not sure about the anomalies in past deaths but would guess it’s reporting lags from public holidays.
      I’m not sure what data is available by cause of death. Probably not the same detail re week of the year and age sex.

  2. I’m in Scotland this is an interesting read thank you I can’t help to add to anything but just wanted to thank you for sharing

  3. Pingback: Should a forty-something woman in Scotland take the AstraZeneca vaccine? | Meaning and Truth

  4. Pingback: How much did Covid-19 change our risk of death? | Meaning and Truth

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s