COVID-19 – Tracking European Mortality
April 20, 2020
Tom Jefferson, Carl Heneghan
When Dr John Snow investigated the 1854 cholera outbreak around Golden Square in Soho London, he made a first basic decision on the type of evidence he was going to look at – deaths. Deaths have a great advantage. Deaths are dichotomous outcomes. You are either dead or alive, there is no in-between.
Of course, you can die for a variety of causes, often more than one cause at a time. The death registration certificate, refined since William Farr’s days, could be wrong or based on uncertain information. The certificates could also be late, creating artefacts such as “parcelling” when a stack of delayed death notifications lands on the desk of an analyst all in one go.
However, mortality does not lie, it is not open to alternative explanations, something which bedevils observational data. All this is summed up in one famous William Farr quote: “the death rate is a fact; anything beyond this is an inference”.
If we can, we should take advantage of our forefathers’ prescience and consult seasonal mortality trends which are available in the Mortality monitoring in Europe Bulletin, familiarly called EUROMoMo.
EUROMoMO weekly reports mortality data from 24 countries listed on its front page. At the time of writing the latest data is from week 15 of 2020 (the week starting 6th of April). The text warns about the potential shortcoming of the data including “catch up” delay.
Even so, the excess deaths during the preceding weeks are stark and as the authors report, they are driven by over 65 mortality in certain European countries. The peak of the deaths has passed, illustrated by the downward curve of the delay-adjusted death numbers (green line).
Comparing the excess death curve with that of previous seasons we see the trend is consistent with previous outbreaks. The peaks and troughs are reflected in the stratified baseline (the red line), which is following a similar seasonal pattern to previous years.
The simple rule IS: 1) examine, analyse and interpret the data to fully understand what is going on; 2) Don’t model “what’s next” until you fully understand point 1.
We think both Snow and Farr would agree.
Click to enlarge
Tom Jefferson is a senior associate tutor and honorary research fellow, Centre for Evidence-Based Medicine, University of Oxford. Disclosure statement is here
Carl Heneghan is Professor of Evidence-Based Medicine, Director of the Centre for Evidence-Based Medicine and Director of Studies for the Evidence-Based Health Care Programme. (Full bio and disclosure statement here)