Recent mild flu seasons result in more COVID-19 deaths
Today, I stumbled across self-published paper by Dr Chris Hope — a Reader at Cambridge University’s Judge Business School. The paper claims to have established a link between the the number of per capita deaths a country has, and the severity of the preceding two flu seasons. The paper was picked up by The Sun, who ran a story on it, but it doesn’t seem to have attracted the attention of the rest of the mainstream media.
Dr Hope is not an epidemiologist. That should not (and does not) disqualify him from participating the public discourse around COVID-19. But the paper would not have survived “first contact” with an appropriately qualified peer reviewer. The main issue is that the index that is ordinal numbers and the statistical techniques he chose are unsuitable for ordinal data. [Strangely enough, on of the few other times I have been inspired to write here was also on the subject of interval versus ordinal data.] To be fair to Dr Hope, he does appear to recognise that this might be problematic. Perhaps he just didn’t have the right statistical tools in his toolbox.
I spent a couple of house this afternoon repeating Dr Hope’s analysis, this time using Spearman Rho rather than the Pearson correlation coefficient. The other changes I made were as follows:
- I excluded countries that have a population of less that 10 million from the analysis, as smaller countries tend to generate more erratic data points.
- I used up to date death data (to 20th August 2020).
My stats may be cleaner, but the result is the same. The milder the flu seasons, the higher the per capita deaths. The rho value is -0.76 which is statistically significant at 0.01. Or in other words, we can be 99.9% confident that the per capita death rates for the countries included in the study are directly related to the severity of the two preceding flu seasons.
This is important for two reasons.
The first is that there has been much public debate about the UK government’s handling of the pandemic response, including frequent accusations that the government’s actions are directly responsible for x number deaths. This analysis does not acquit the government entirely of that charge. But what it does show is that — to an extent — some nations were predestined to suffer worse effects from the pandemic than others. That is something we all ought to bear in mind when seeking to judge how effective a government’s pandemic response has been.
The second is that it gives a clue as what we might expect in any subsequent waves of COVID-19. It’s clear that per capita death rate is a function not just of the number of infections, but [probably more importantly] the number of people vulnerable to serious complications.
In England & Wales, deaths where COVID-19 was not mentioned on the death certificate have been below the five year average for 10 out of the last 11 weeks for which we have data. And for five of the last six weeks, all-cause mortality including COVID-19 deaths has been below the five year average.
It’s fairly clear that prior to COVID-19 hitting the UK there were lots of vulnerable people who, if the last two flu seasons hadn’t been so mild, would not have been alive to see the start of the pandemic. It’s also clear that, there are a significant number people who, if they had not sadly been taken by COVID-19, would not have lived much longer anyway.
The good news, if you can call it that, is that the remaining population is — on average— less vulnerable. And that means that, even if we do see a second wave, we should not expect to see a similar death toll similar to a grizzly months of April and May.