Is there a Link between the 2021 COVID-19 Vaccination Uptake in Europe and 2022 Excess All-Cause Mortality?

Comment: An interesting and very controversial article. As you mention, the correlation should be analysed by age cohorts. A possible explaining hypothesis could be that countries with high vaccination rates (Denmark) have vaccinated a younger strata of the population, including more young adults and children. Countries with lower rates (Bulgaria) vaccinated only the high risk population, the elderly. If negative side effects of the vaccines are more extended among the young, the vaccines might indeed have caused higher mortality.

Comment: Please have a look at the following manuscript from 2021, which analyses all cause mortalities for age cohorts in the USA and in Europe, in relation to COVID19 injections. The analyses of the european data also indicate the dynamics of deaths in relation to duration since the 1st injection.
https://www.researchgate.net/publication/355581860_COVID_vaccination_and_age-stratified_all-cause_mortality_risk
More updates of these data are also available at my Researchgate profile.
Best,
Herve Seligmann

Comment: How is it possible to relate mortality to vaccines, as history of vaccinations is readily available but past history of covid is not reliable as asymptomatic undetected undiagnosed covid will be missed out. After 2021, almost more than 85 percent population had covid, either asymptomatic undetected undiagnosed covid or symptomatic covid.
Antibodies to N antibodies have short half life and they disappear faster but T cell immunity persists.
Unless TCell immunity studies are done in all population, it is not possible to rule out past covid only from N antibodies studies..
How do we know about side effects of
vaccines or something to do with side effects of covid itself .

Comment: The linearity assumptions in month is quite strong assuming that the excess death rate is constant from month to month. Paper mentioned a secondary analysis that sounded to me like you treated month as categorical — do you have that analysis or can you post supplementary materials with secondary analyses?

Also the plots in the paper just show the linear fits which of course we know are lines by construction but we can’t see how well it fits the data. Do you have a scatterplot with excess deaths by country/month with fitted line overlaid so we can see how well it fits, and also plotting the scatterplot of excess deaths vs vaccination with line overlaid? Many readers might appreciate those plots.

Comment: Dear authors,
I performed a very simple reanalysis:
Plotted using using Excel scatter graph your vax uptake values versus excess deaths p scores (via https://ourworldindata.org/)
The result was clear correlation: higher the vaccination uptake the lower the excess deaths at end of 2022. Apparently the opposite to your finding.

https://twitter.com/Paul_A_Coombes/status/1638562850976288770?s=20
Regards,
Paul Coombes

Comment: An interesting article that shouldn't be controversial.

It will be interesting to remove certain causes of mortality that are likely to be not directly due to vaccination, e.g. road accidents and crime. Although these data may not be readily available for all the countries included in the analysis.

Comment: I am not a statistician, but surely, if, for whatever reason, there is a year (2021) with lower than average excess deaths, we have a following year/years where there has to be a catchup.
If 2022 is that year, is it surprising that the excess deaths take a jump, or am I mis-reading something?

Comment: Hi

When you aim at discovering that of Covid-19 role it can be done directly by the original mathematical method. It shows the role can be only a minimal one, and a very vast majority of excess deaths must be due to not Covid-19 causes, including vaccines, like in the U.S. (or in the U.K) in 2020 >90% of "Covid-19 deaths" were wrongly attributed to Covid-19. ...It is new place from October is the major based in Switzerland preprint (as blocked, without any excuse, by the U.S ones, including journals):
https://zenodo.org/record/7372672 To some people it seems so difficult that they give up.

Comment: Is there a version which has been peer-reviewed yet?

Comment: I may be wrong, but I don't think you are using age-standardised death rates? Death counts can be distorted by population growth and ageing.

All the countries quoted as having >75% vaccination rates have populations that are growing, whereas all but one of the countries with <60% vaccination rates have shrinking populations.
If you're comparing 2022 death rates to the death rate in ~2017, and there are 5% fewer Bulgarians in 2022, you would expect to see 5% fewer deaths in 2022 from that reason alone. Conversely, Ireland has grown by >5% in that time, so you would expect 5% more deaths in Ireland (assuming the population growth had the same demographics as the existing population).

In reality the effects will be less than that as population growth comes from a mix of more babies and immigration, which usually involves younger-than-average people. But that's why you need to age standardise.

As an aside, there's a nice natural experiment in the UK, where vaccination rates are pretty uniform across the country, but healthcare is organised at sub-national level and there's lots of stats on healthcare performance at regional level. It's notable that Scotland has had far fewer excess deaths in 2022 than England, which may have something to do with Scotland's health system living with Covid much better than its counterpart in England.

For instance, one of the standards is that the mean response time for Category 2 calls (chest pain etc) is 18 minutes and 90% get an ambulance within 40 minutes. In 2022 England averaged around 40 minutes and 90% in about 100 minutes, whereas Scotland more or less achieved the targets of 18/40 - except for a wobble in December. So it might be interesting to look at those kinds of healthcare statistics if you're looking for correlations with excess deaths.

Comment: Yes, covid vaccines increased mortality rate for about 15%.
https://www.preprints.org/manuscript/202301.0204/v7

Comment: On authors observations that increase in covid19 vaccine intake in 2021 being correlated with increases in all-cause mortality: Which period are you comparing with? How was the all-cause mortality trend at pre- and during covid-19 pandemic? If the trend in pre-pandemic years was going up, couldn't reasons other than vaccination be at play? Perhaps burden of covid-19 infection being the driver behind increases in all-cause mortality etc. Another important observation is that countries tended to boost up vaccination efforts with increases in covid-19 infection waves. Countries that saw less of the infections tended to be slower in ramping up the covid-19 vaccinations. Couldn't this partially or wholly explain higher all-cause mortality with increases in vaccination? I.e more to do with correlation and not causality.

Comment: Just a brief note - the concept of "borderline significance" has no meaning within the logic of Fisherian or Neyman-Pearson significance testing. It makes as much sense as being "borderline pregnant". Perhaps a Bayesian analysis might be more helpful?

Comment: There is absolutely *zero* reason to weight the independent variables by population size. All-cause mortality is already a proportional measure. I look forward to seeing a peer-reviewed version of this paper.

Comment: I read the paper with interest - thank you for making it public. I didn't see any discussion of the variance year-to-year in excess mortality and whether 2022 fell within 'normal' bounds - that would be an essential, I think - and whether normalising by a wider range of years changed any outcomes (I recognise the tension between longer normalising intervals and the desire to deal with a stationary population). Related to that, it would also be interesting to know if, in your base data set, there was any increasing trend in mortality that would make excess mortality more likely and how that varied between countries (I am sorry but I have not looked up your references for the data). It would also be interesting to see if there was any correlation between month of the year and mortality - in general - and excess or deficient mortality - in particular- in other years. Mostly, I was surprised that vaccination at any time in 2021 was somehow related to a progressive increase in excess mortality over 2022. Is that a correct interpretation - that you took the all year vaccination as opposed to the monthly vaccination rates? I did not see any discussion of lag rates so imagine that must be your method.

Comment: These results are entirely compatible with being a statistical artefact.
Both vaccination rate and excess death rate have the popuation in the denominator and weighting by the population does not remove the artefact.
Just try generating random numbers of vaccinations, random numbers of excess deaths and using the population values to compute the rates of vaccination and excess deaths by dividing by the population and then regress excess deaths on vaccination rate and you will find a highly significant relation.

With y as a random number, x as a random number and with some fixed d (denominator), y/d will have a positive relation with x/d.

Comment: On further reflection I think it will be likely to be a negative relation, but I think the problem remains that it could be an artefact.

Thanks

Stephen

Comment: 1. The excess mortality is compared to **before COVID started**.
2. You don't bother to check for correlation between vaccination rates and COVID infection rates.
3. You don't bother to remove COVID deaths from the excess mortality.

Given #1, you should at least check #2 and #3.