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

Abstract

Purpose: We primarily study a possible link between 2021 COVID-19 vaccination uptake in Europe and monthly 2022 excess all-cause mortality, that is, mortality higher than before the pandemic. Methods and Results: Analyses of 31 countries weighted by population size show that all-cause mortality during the first 9 months of 2022 increased more the higher the 2021 vaccination uptake; a one percentage point increase in 2021 vaccination uptake was associated with a monthly mortality increase in 2022 by 0.105% (95% CI, 0.075-0.134). When controlling for alternative explanations, the association remained robust, and we discuss the result emphasizing causality as well as potential ecological fallacy. Furthermore, the study shows that 2021 all-cause mortality was lower the higher the vaccination uptake, but this association became non-significant when controlling for alternative explanations. Conclusion: Despite a possible preventive effect in 2021, we cannot rule out that COVID-19 vaccination uptake in Europe has led to increasing 2022 all-cause mortality between January and September.

Full text

Is There a Link between the 2021 COVID-19 Vaccination
Uptake in Europe and 2022 Excess All-Cause Mortality?
Jarle Aarstad*, Olav Andreas Kvitastein
Ab s t r A c t
Purpose: We primarily study a possible link between 2021 COVID-19 vaccination uptake in Europe and monthly 2022 excess all-cause
mortality, that is, mortality higher than before the pandemic. Methods and Results: Analyses of 31 countries weighted by population size
show that all-cause mortality during the rst 9months of 2022 increased more the higher the 2021 vaccination uptake; a one percentage
point increase in 2021 vaccination uptake was associated with a monthly mortality increase in 2022 by 0.105% (95% CI, 0.075–0.134). When
controlling for alternative explanations, the association remained robust, and we discuss the result emphasizing causality as well as potential
ecological fallacy. Furthermore, the study shows that 2021 all-cause mortality was lower the higher the vaccination uptake, but this association
became non-signicant when controlling for alternative explanations. Conclusion: Despite a possible preventive eect in 2021, we cannot rule
out that COVID-19 vaccination uptake in Europe has led to increasing 2022 all-cause mortality between January and September.
Keywords: All-cause mortality, Causal inferences, COVID-19, Ecological fallacy, Excess mortality, Individualistic fallacy, Vaccination
Asian Pac. J. Health Sci., (2023); DOI: 10.21276/apjhs.2023.10.1.6
In t r o d u c t I o n
According to Eurostat, the EU experienced excess all-cause
mortality in the rst 9 months of 2022, that is, the mortality was
higher than the average of the same months between 2016
and 2019, before COVID-19.[1] In this study, our primary aim is to
investigate if the pattern – 2022 nation-level monthly all-cause
mortality relative to pre-pandemic numbers – can be linked
to nation-level COVID-19 vaccination uptake by the end of the
previous year, 2021.
Although research has indicated that COVID-19 vaccination has
prevented SARS-CoV-2-related hospital admission and deaths,[2-4] its
preventive eect has waned.[5] Furthermore, COVID-19 vaccination
has side eects,[6-9] such as myocarditis and pericarditis,[10,11] and
along with this knowledge, it is worth noting that a recent study
falsied a suspected association between the two diagnoses
and COVID-19 virus infection.[12] Coining this information with
research showing that vaccine side eects generally tend to be
underreported,[13] we ask if COVID-19 vaccination may be associated
with long-term mortality. To study this potential link, we regress
all-cause mortality in the rst 9 months of 2022 (relative to pre-
pandemic levels) on vaccination uptake in 31 European countries
by the end of 2021. The 31 countries are the EU member states, plus
Norway, Iceland, Liechtenstein, and Switzerland. The estimates are
weighted by each country’s population size by January 01, 2020.
Furthermore, we include an interaction term between vaccination
uptake and the time that has passed since the beginning of 2022.
Our motive is to assess if a potential link between vaccination
uptake and all-cause mortality has changed over time (To further
illuminate the study’s topic, the Appendix illustrates the association
between 2021 vaccination uptake and all-cause mortality the same
year relative to pre-pandemic levels).
In the following, we present the materials and methods
in detail and report the empirical results. Then, we discuss the
ndings, address the study’s limitations, and suggest avenues
for future research. When discussing the results and addressing
study limitations, we particularly emphasize the issues of causal
inference and ecological fallacy.
Faculty of Engineering and Science, Western Norway University of
Applied Sciences, Bergen, Norway.
Corresponding Author: Jarle Aarstad, Western Norway University
of Applied Sciences, PO Box 7030, NO-5020 Bergen, Norway.
E-mail:jarle.aarstad@hvl.no
How to cite this article: Aarstad J, Kvitastein OA. Is There a Link
between the 2021 COVID-19 Vaccination Uptake in Europe and 2022
Excess All-Cause Mortality? Asian Pac. J. Health Sci., 2023;10(1):25-31.
Source of support: Nil.
Conicts of interest: None.
Received: 02/01/2023 Revised: 10/02/2023 Accepted: 17/02/2023
MAt e r I A l s A n d M e t h o d s
As a dependent variable, we applied data from Eurostat, which
models the countries’ all-cause mortality each month divided by
the average of the same month between 2016 and 2019 multiplied
by 100.[1] Finally, the expression is subtracted by 100, which implies
that positive values indicate excess mortality and negative values
indicate submortality. Vaccination uptake is the percentage of the
total population in each country that has received a “primary course”
by week 52, 2021. We gathered the data from the COVID-19 vaccine
tracker by the European Center for Disease Prevention and Control,[14]
except for Switzerland, where we used Our World in Data as a source[15]
[for details on countries’ vaccination uptake and size in population,
Table1]. Each month in 2022 is modeled straightforwardly, where
January takes the value of one, February two, etc.
We control for pre-COVID life expectancy in 2019 using
Eurostat data[16] as it is a proxy for health status in each country.[17]
Finally, we control for average all-cause mortality in 2020 and 2021
divided by the average between 2016 and 2019.[18] Our motive for
the control variable is that relatively high mortality in 2020 and
2021 may induce relatively low mortality in 2022 and vice versa.
On the other hand, relatively high mortality in 2020 and 2021 may
indicate a deteriorating health status not captured by pre-COVID
life expectancy. For consistency, we multiply the measure by 100.
©2023 The Author(s). This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/
licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
ORIGINAL ARTICLE e-ISSN: 2349-0659 p-ISSN; 2350-0964

www.apjhs.com Jarle Aarstad and Olav Andreas Kvitastein: COVID-19 Vaccination Uptake and Excess All-Cause Mortality
Asian Pacic Journal of Health Sciences | Vol. 10 | Issue 1 | January-March | 2023 26
Below, we rst present descriptive statistics and correlations.
Next, in our statistical models, we apply linear multilevel mixed-
eects random intercept regression analyses, where monthly
observations are nested at the country level.[19,20] As noted,
observations are weighted according to the population size by
January 01, 2020, and we carry out all analyses in Stata 17.0.[21]
re s u lt s
Table 2 reports minimum values, maximum values, descriptive
statistics, and correlations. We observe strong correlations between
vaccination uptake and the control variables average 2020–2021
mortality and life expectancy, which may cause problems with
multicollinearity.[22] Later, we explain how we deal with the issue.
Model 1, Table 3, shows that vaccination uptake has a non-
signicant direct association with the dependent variable, but the
signicant monthly association indicates that the overall mortality
tends to increase over time. The interaction between vaccination
uptake and time passed in months since the beginning of 2022
(V*M) is strongly signicant and implies that the mortality
increases the higher the vaccination uptake. Specically, it shows
that a one percentage point increase in 2021 vaccination uptake is
associated with an increase in 2022monthly mortality by 0.105%
(95% CI, 0.075–0.134).
We mean-center the variables to minimize multicollinearity
problems,[23] and the variance ination factor (VIF) taking a value
of one concerning the month variable and the interaction term
shows that multicollinearity is not a problem. However, the VIF
concerning the vaccination uptake taking a value of 5.17 may be
problematic. To amend the issue, we omit the control variables in
Model 2 and observe that all the VIFs are now low. Furthermore, the
remaining estimates and their error terms are not much altered.
As robustness checks, in Models 3 and 4, we add an interaction
term between each control variable and the month variable at
a time. We observe that statistical conclusions concerning the
independent variables and their interaction term are unaltered.
The only exception is vaccination uptake showing a borderline
signicant association with the dependent variable in Model 3.
Model 5 replicates the rst model except that it excluded July
when a heat wave occurred. The overall increasing monthly
association is now borderline signicant, but the interaction term
association remains robust. Finally, we carried out unreported
analyses controlling for 2018 nation-level median age and 2019
per-capita GDP adjusted for purchasing power, respectively,
as done in Models 3 and 4, but without altering any statistical
conclusion (analyses are available on request).
Figure1a-c, based on Model 1, Table3, concludes the ndings.
It indicates increasing all-cause excess mortality [Figure 1a], and
more strikingly, the increase is higher the higher the national
vaccination uptake [Figure 1b]. As noted, a one percentage point
increase in 2021 vaccination uptake is associated with an increase in
2022monthly mortality by 0.105% (95% CI, 0.075–0.134). In countries
with relatively high vaccination uptakes, which include numerous of
Europe’s states and the most populous [Figure1d, based on Table1],
the result estimates a mortality increase [Figure 1c]. In countries
with relatively low vaccination uptakes, the result estimates a low or
even negative mortality increase [Figure1c].
dI s c u s s I o n
This study shows that the all-cause mortality during the rst
9 months of 2022 in 31 European countries increased more
the higher the 2021 vaccination uptake. The association is
strongly signicant [Table 3], but to make causal inferences,
it is further necessary to (1) rule out reverse causality and
(2)account for alternative explanations.[23] Below we discuss both
issues. Furthermore, we discuss our study in light of potential
(3) ecological fallacy, which is a “failure in reasoning that arises
when an inference is made about an individual based on aggregate
data for a group.”[24]
Concerning potential reverse causality (1), the timing of the
independent and dependent variables is crucial. Relating the
timing condition to our study, we nd it unreasonable, actually
Table 1: Vaccination uptake and countries’ population size
Country Vaccination uptake Population
Bulgaria 27.7 6951482
Romania 40.8 19328838
Slovakia 48.5 5457873
Croatia 53.3 4058165
Slovenia 55.6 2095861
Poland 55.8 37958138
Hungary 61.0 9769526
Estonia 61.3 1328976
Czechia 62.2 10693939
Latvia 64.9 1907675
Liechtenstein 65.4 38747
Lithuania 65.7 2794090
Greece 66.3 10718565
Switzerland 67.1 8606033
Netherlands 67.6 17407585
Luxembourg 68.2 626108
Sweden 69.3 10327589
Cyprus 69.4 888005
Austria 70.4 8901064
Germany 72.8 83166711
Norway 73.0 5367580
France 73.3 67320216
Italy 74.9 59641488
Spain 74.9 47332614
Finland 75.0 5525292
Belgium 76.2 11522440
Iceland 76.4 364134
Ireland 76.6 4964440
Denmark 78.7 5822763
Malta 82.1 514564
Portugal 83.1 10295909
Table 2: Descriptive statistics and correlations weighted by countries’ population size
Min. Max. Mean SD Variable 1 2 3 4
−25.4 53.9 9.98 7.24 Month. 2022 mort. rel. to 2016–19 avg. (1)
101.4 126.1 112.5 5.58 Avg. 2020–2021 mort. rel. to 2016–2019 (2) 0.016
75.1 84.3 81.5 2.47 Life expectancy in 2019 (3) 0.113† −0.620***
1 9 5 2.59 Month number in 2022 (4) 0.256*** 0 0
27.7 83.1 68.7 10.3 Vaccination uptake by the end of 2021 0.071 −0.742*** 0.857*** 0
Two-tailed tests of signicance. † P<0.10; * P<0.05; ** P<0.01; *** P<0.001 n=248

Jarle Aarstad and Olav Andreas Kvitastein: COVID-19 Vaccination Uptake and Excess All-Cause Mortality www.apjhs.com
Asian Pacic Journal of Health Sciences | Vol. 10 | Issue 1 | January-March | 2023
27
logically impossible, that increasing monthly all-cause mortality in
2022 could have occurred before and hence caused vaccination
uptake in the previous year, 2021. Accordingly, ceteris paribus it
is more likely to assume that 2021 vaccination uptake will have
caused monthly increases in 2022 mortality than the other way
around.
Concerning alternative explanations (2), we controlled for
average all-cause mortality in 2020 and 2021 divided by the
average between 2016 and 2019. The all-cause mortality during
2020 and 2021 was relatively low in countries with relatively
high vaccination uptakes [Table 2], and had we not controlled
for this issue, one could alternatively have argued for a so-called
“bounce-back eect,” that is, relatively low mortality at one period
is followed by relatively high mortality later, and vice versa. For
example, the relatively high mortality in the 1st months of 2022
could have been due to relatively low mortality in the previous
2years, but, as mentioned, we controlled for this issue. Moreover,
we not only controlled for average 2020 and 2021 mortality relative
to the average between 2016 and 2019, but we additionally
included an interaction term between this variable and the
month variable simultaneously with the interaction term between
vaccination uptake and the month variable [Table 3, Model 3].
Despite this – in our opinion, conservative test – we still observed
a signicant association between 2021 vaccination uptake and the
2022monthly increase in all-cause mortality. Nonetheless, future
research should extend the time frame beyond 2022’s rst 9months
to assess whether the relatively high all-cause mortality persists in
countries with high vaccination uptakes. If so, there is even more
reason to assume that there is a genuine association between 2021
COVID-19 vaccination uptake and 2022 excess mortality. Second,
we controlled for 2019 pre-COVID life expectancy. The variable
correlated positively with vaccination uptake and negatively with
the other control variable – average 2020 and 2021 mortality
relative to the 2016–2019 average mortality [Table2]. Accordingly,
the inclusion of pre-COVID life expectancy as a control variable
partakes to rule out other potential alternative explanations of our
ndings. Furthermore, we included an interaction term between
this variable and the month variable simultaneously with the
interaction term between vaccination uptake and the month
variable, but without altering any statistical conclusion [Table3,
Model 4]. Finally, we carried out unreported analyses controlling
for 2018 nation-level median age and 2019 per-capita GDP
adjusted for purchasing power, respectively, as done in Models
3 and 4 [Table3], but without altering any statistical conclusion
(analyses are available upon request).
Concerning ecological fallacy (3), we are cautious about
Table 3: Multi-level mixed-eects random intercept linear regressions with robust standard errors and 2022 monthly all-cause mortality
compared to 2016–19 monthly averages as the dependent variable. Weighted by countries’ population size
Fixed eects Model 1 Model 2 Model 3 Model 4 Model 5a
Intercept 9.98*** 9.98*** 9.98*** 9.98*** 9.21***
(0.439) (0.481) (0.423) (0.463) (0.470)
Vaccination uptake by
the end of 2021 (V)
0.016 0.050 0.130† −0.066 0.004
(0.105) (0.048) (0.067) (0.077) (0.105)
[5.17] [1.00] [2.23] [2.76] [5.18]
Month number in
2022 (M)
0.716** 0.716** 0.716** 0.716** 0.484†
(0.262) (0.262) (0.249) (0.258) (0.280)
[1.00] [1.00] [1.00] [1.00] [1.00]
Avg. 2020–2021 mort.
rel. to 2016–2019 (A)
0.190 0.198† 0.105
(0.137) (0.117) (0.146)
[2.23] [2.23] [2.23]
Life expectancy 2019
(L)
0.538† 0.565 0.210
(0.318) (0.346) (0.333)
[3.77] [3.76] [3.77]
V*M 0.105*** 0.105*** 0.090*** 0.111** 0.091***
(0.015) (0.015) (0.022) (0.032) (0.015)
[1.00] [1.00] [2.23] [3.76] [1.01]
A*M −0.037
(0.048)
[2.23]
L*M −0.031
(0.151)
[3.76]
Random eects
Residual 38.3 38.3 38.1 38.3 28.5
(7.88) (7.88) (8.08) (7.96) (5.62)
Country eect 1.52 2.53 2.00 2.02 2.39
(2.43) (2.65) (2.59) (2.64) (2.31)
Wald χ2128.2*** 134.0*** 191.1*** 179.5*** 109.0***
Log pseudo-likelihood −1.35e10 −1.36e10 −1.36e10 −1.36e10 −1.16e10
Estimates are weighted by country size in population size by January 1, 2020, and we report robust standard errors in parentheses. We report VIF: Variance
ination factors in brackets. For xed eects, we report conservative two-tailed tests of signicance. † P<0.10; *P<0.05; **P<0.01; and ***P<0.001. Models 1–4
have 279 monthly observations (nine monthly observations per each of the 31 countries). aModel 5 excludes July 2022 and has 248 monthly observations (eight
observations per each of the 31 countries)

www.apjhs.com Jarle Aarstad and Olav Andreas Kvitastein: COVID-19 Vaccination Uptake and Excess All-Cause Mortality
Asian Pacic Journal of Health Sciences | Vol. 10 | Issue 1 | January-March | 2023 28
making individual-level inferences from our nation-level ndings.
In other words, we emphasize that our study shows a positive
association between 2021 nation-level vaccination uptake and
2022 nation-level monthly increase in excess mortality while ruling
out reverse causality and accounting for alternative explanations.
Having said that, it may be worth discussing ecological fallacy in
general and the plausibility of making individual-level inferences
from our nation-level ndings, which we do in the following
paragraphs.
First, we address the classical Robinson’s paradox.[25] At a US
state level, Robinson found that the illiteracy rate was lower the
higher the foreign-born population rate, but at an individual level,
foreign-borns had a higher probability of being illiterate than
national-borns. An explanation of the paradox is that foreign-
borns had a high proclivity to settle in states with high literacy
rates, which illustrates a reverse causality. Returning to our study,
we have argued that reverse causality is not a potential challenge
to the results’ interpretation as the timing of the independent
variable precedes the dependent variable.
Second, we address the Simpson’s paradox,[26,27] which could
imply that individual-level vaccination was associated with a
monthly decrease in the individual-level mortality probability
despite a positive association between the nation-level vaccination
uptake and increased monthly mortality. However, Simpson’s
paradox hinges on the omitted variable bias,[28] or cross-level bias,
which is the “dierence between the expectation of an estimator
from an ecological [group-level] study and the individual-level
parameter of interest.”[29] For example, the mortality may have
increased in a high-vaccinated country, not because vaccination is
detrimental to health, but because relatively few people died there
in the previous 2years. Similarly, the mortality may have decreased
in a low-vaccinated country, not due to a lower health risk than
in high-vaccinated countries, but because relatively many people
died there in the two previous years. Related to this argument, it is
relevant to address that there, for some countries, in concert with
relatively low mortality in the pandemic’s 2years, was a mortality
increase early in 2022, arguably due to the COVID-19 omicron
variant.[30] However, again, we emphasize having accounted for this
issue by controlling for the 2020–2021 mortality (plus including
other control variables, as discussed above). Furthermore, we
have shown that the mortality particularly increased during the
rst nine months of 2022 in high-vaccination countries, while the
mortality due to omicron abated early the same year.[31]
Finally, we emphasize that individual-level studies, too, can
be subject to fallacy.[29,32] For example, collective vaccination status
can be genuinely associated with reduced disease transmission.
Accordingly, individual-level studies can be limited in capturing
the association between collective vaccination status and health
outcomes. In light of this argument, Loney and Nagelkerke[29]
assert that “even when the emphasis is clearly on the individual,
with the ultimate objective of understanding the etiology of
disease, ecological [group-level] analyses can be more “truthful”
than individualistic analyses and may avoid a large portion of
confounding on the individual level.” Subramanian, Jones[32]
similarly state that “meaningful analysis of individual-level
relationships requires attention to substantial heterogeneity in
state [or group-level] characteristics. The implication is that perils
are posed by not only ecological fallacy but also individualistic
fallacy.” To illustrate their point, they showed that the individual-
level association between race and illiteracy in the US was biased
in the absence of controlling for state-level heterogeneity. Hence,
not only group-level studies but also individual-level ones can
be subject to fallacy. Returning to our data, we accordingly
acknowledge as a limitation that the association between nation-
level vaccination uptake and excess mortality may not have fully
captured an individual-level association, but, at the same time,
Figure1: Monthly estimated excess mortality (a) and estimated change in excess mortality as a function of national vaccination uptake with
95% Cis (b). Estimated excess mortality for vaccination uptake at 60% (blue), 75% (green), and 68.7% (red), which is the weighted European
average, and actual excess mortality (yellow) (c). Vaccination uptake and population size (d). a-c are based on Model 1, Table3, and d is
based on Table1.
d
c
b
a

Jarle Aarstad and Olav Andreas Kvitastein: COVID-19 Vaccination Uptake and Excess All-Cause Mortality www.apjhs.com
Asian Pacic Journal of Health Sciences | Vol. 10 | Issue 1 | January-March | 2023
29
an individual-level association could also have been biased in the
absence of controlling for higher-level heterogeneity.[32] Therefore,
we encourage future studies to address both levels simultaneously.
Other explanations of excess mortality than those studied in
this paper can be delayed diagnosis or medical treatment during
COVID-19, although we cannot see that the issues have been more
prevalent in high-vaccination versus. low-vaccination countries,
that is, we do not expect delayed diagnosis or medical treatment
during COVID-19 to substantially have induced omitted variable
bias. Nonetheless, we encourage future studies to address the
constructs as potential carrier of excess mortality, or eventually
study them in concert with vaccination uptake.
Moreover, future research should assess if all-cause mortality in
dierent age cohorts or between genders is a function vaccination
uptake or if the type of vaccination has played a role. Similarly,
future research should investigate if vaccination uptake or type of
vaccination is associated with specic causes of death. Data from
England and Wales show excess mortality from April 2022 across
gender and age cohorts and several diagnoses.[33] However, we do
not know if the pattern can be attributed to vaccination uptake,
type of vaccination, delayed diagnosis, delayed medical treatment,
or if long-COVID[34] has played a role, and future research should
investigate these issues.
(The Appendix, carrying out pre-2022 analyses, illuminates
a negative association between vaccination uptake and 2021
mortality relative to the 2016–2019 average mortality, which
indicates a short-time preventive eect. However, the association
becomes non-signicant when including relevant controls. 2019
life expectancy, on the other hand, has a signicant negative
association with 2021 mortality. Finally, the association between
2019 life expectancy and vaccination uptake is strongly positive).
co n c lu s I o n
Analyses of 31 countries weighted by population size show that
all-cause mortality during the rst nine months of 2022 increased
more the higher the 2021 vaccination uptake; a one percentage
point increase in 2021 vaccination uptake was associated with
a monthly mortality increase in 2022 by 0.105 percent (95% CI,
0.075-0.134). When controlling for alternative explanations, the
association remained robust.
co p yr I g h t A n d p e r M I s s I o n s t At e M e n t
We conrm that the materials included in this paper do not violate
copyright laws. Where relevant, appropriate permissions have been
obtained from the original copyright holder(s). All original sources
have been appropriately acknowledged and/or referenced.
dAtA A v A I l A b I l I t y
All data used in this research is publicly available. On request, the
corresponding author can provide raw data and Stata codes.
references
1. Eurostat. Excess Mortality by Month. Available from: https://ec.europa.
eu/eurostat/databrowser/view/demo_mexrt/default/table?lang=en
[Last accessed on 2022 Dec 20].
2. Haas EJ, Angulo FJ, McLaughlin JM, Anis E, Singer SR, Khan F, et al.
Impact and eectiveness of mRNA BNT162b2 vaccine against
SARS-CoV-2 infections and COVID-19 cases, hospitalisations, and
deaths following a nationwide vaccination campaign in Israel:
An observational study using national surveillance data. Lancet
2021;397:1819-29.
3. Bernal JL, Andrews N, Gower C, Robertson C, Stowe J, Tessier E, etal.
Eectiveness of the Pzer-BioNTech and Oxford-AstraZeneca vaccines
on covid-19 related symptoms, hospital admissions, and mortality in
older adults in England: Test negative case-control study. BMJ Br Med
J 2021;373:n1088.
4. Watson OJ, Barnsley G, Toor J, Hogan AB, Winskill P, Ghani AC. Global
impact of the rst year of COVID-19 vaccination: A mathematical
modelling study. Lancet Infect Dis 2022;22:1293-302.
5. Nordstrom P, Ballin M, Nordstrom A. Risk of infection, hospitalisation,
and death up to 9months after a second dose of COVID-19 vaccine:
A retrospective, total population cohort study in Sweden. Lancet
2022;399:814-23.
6. Kadali RA, Janagama R, Peruru S, Malayala SV. Side eects of BNT162b2
mRNA COVID-19 vaccine: Arandomized, cross-sectional study with
detailed self-reported symptoms from healthcare workers. Int J Infect
Dis 2021;106:376-81.
7. Klugar M, Riad A, Mekhemar M, Conrad J, Buchbender M, HowaldtHP,
et al. Side eects of mRNA-based and viral vector-based COVID-19
vaccines among German healthcare workers. Biology (Basel) 2021;10:752.
8. Finsterer J. Neurological side eects of SARS-CoV-2 vaccinations. Acta
Neurol Scand 2022;145:5-9.
9. Patone M, Mei XW, Handunnetthi L, Dixon S, Zaccardi F, Shankar-
Hari M, et al. Risk of myocarditis after sequential doses of COVID-19
vaccine and SARS-CoV-2 infection by age and sex. Circulation
2022;146:743-54.
10. Karlstad Ø, Hovi P, Husby A, Härkänen T, Selmer RM, Pihlström N, et al.
SARS-CoV-2 vaccination and myocarditis in a Nordic cohort study of
23 million residents. JAMA Cardiol 2022;7:600-12.
11. Oster ME, Shay DK, Su JR, Gee J, Creech CB, Broder KR, et al. Myocarditis
cases reported after mRNA-based COVID-19 vaccination in the US
from December 2020 to August 2021. JAMA 2022;327:331-40.
12. Tuvali O, Tshori S, Derazne E, Hannuna RR, Afek A, Haberman D,
etal. The incidence of myocarditis and pericarditis in post COVID-19
unvaccinated patients-a large population-based study. JClin Med
2022;11:2219.
13. Lazarus R, Klompas M, Bernstein S. Electronic Support for Public
Health-Vaccine Adverse Event Reporting System (ESP: VAERS). Grant
Final Report, Grant ID: R18 HS. National Library of Medicine: Bethesda;
2010. p.17045.
14. European Centre for Disease Prevention and Control. COVID-19
Vaccine Tracker. Available from: https://vaccinetracker.ecdc.europa.
eu/public/extensions/COVID-19/vaccine-tracker.html#uptake-tab
[Last accessed on 2022 Dec 20].
15. Our World in Data. Coronavirus (COVID-19) Vaccinations. Available
from: https://ourworldindata.org/covid-vaccinations [Last accessed
on 2022 Dec 20].
16. Eurostat. Life Expectancy by Age and Sex. Available from: https://
ec.europa.eu/eurostat/databrowser/view/demo_mlexpec/default/
table?lang=en [Last accessed on 2022 Dec 20].
17. Roser M, Ortiz-Ospina E, Ritchie H. Our World in Data: Life Expectancy.
Oxford, UnitedKingdom: University of Oxford; 2013. Available from:
https://ourworldindata.org/life-expectancy [Last accessed on 2022
Dec 20].
18. Eurostat. Deaths and Crude Death Rate. Available from: https://
ec.europa.eu/eurostat/databrowser/view/tps00029/default/
table?lang=en [Last accessed on 2022 Dec 20].
19. Raudenbush SW, Bryk AS. Hierarchical Linear Models: Applications
and Data Analysis Methods. 2nded. Thousand Oaks, CA: Sage; 2002.
20. Snijders TA. Multilevel analysis. In: Lovric M, editor. International
Encyclopedia of Statistical Science. Berlin, Heidelberg: Springer Berlin
Heidelberg; 2011. p.879-82.
21. StataCorp. Version 17. College Station: TX StataCorp LP; 2021.
22. O’Brien RM. Acaution regarding rules of thumb for variance ination

www.apjhs.com Jarle Aarstad and Olav Andreas Kvitastein: COVID-19 Vaccination Uptake and Excess All-Cause Mortality
Asian Pacic Journal of Health Sciences | Vol. 10 | Issue 1 | January-March | 2023 30
factors. Qual Quant 2007;41:673-90.
23. Cook TD, McCleary R, McCain LJ, Reichardt CS, Fankhauser G,
Campbell DT. Quasi-Experimentation: Design and Analysis Issues for
Field Settings. Boston: Houghton Miin Co.; 1979.
24. Hsieh JJ. Encyclopedia Britannica. Ecological Fallacy; 2017. Available
from: https://www.britannica.com/science/ecological-fallacy [Last
accessed on 2022 Dec 20].
25. Robinson WS. Ecological correlations and the behavior of individuals.
Am Soc Rev 1950;15:351-7.
26. Simpson EH. The interpretation of interaction in contingency tables.
JR Stat Soc Series B (Methodological) 1951;13:238-41.
27. Blyth CR. On Simpson’s paradox and the sure-thing principle. JAm
Stat Assoc 1972;67:364-6.
28. Rothman KJ. A pictorial representation of confounding in
epidemiologic studies. JChronic Dis 1975;28:101-8.
29. Loney T, Nagelkerke NJ. The individualistic fallacy, ecological studies
and instrumental variables: A causal interpretation. Emerg Themes
Epidemiol 2014;11:18.
30. Folkehelseinstituttet. Dødelighet i Norge Under Koronapandemien
2020 til Høsten 2022. Available from: https://www.fhi.no/
contentassets/a5c83528acc94ddfbe815a5934faadf5/dodelighet-
under-koronapandemien-2020-til-hosten-2022.pdf [Last accessed on
2022 Dec 23].
31. Statista. Number of New Coronavirus (COVID-19) Deaths in Europe
since February 2020.[Available from: https://www.statista.com/
statistics/1102288/coronavirus-deaths-development-europe [Last
accessed on 2022 Dec 20].
32. Subramanian SV, Jones K, Kaddour A, Krieger N. Revisiting Robinson:
The perils of individualistic and ecologic fallacy. Int J Epidemiol
2009;38:342-60.
33. Oce for Health Improvements and Disparities. Excess Mortality
in England. Available from: https://app.powerbi.com/view?r=ey
JrIjoiYmUwNmFhMjYtNGZhYS00NDk2LWFlMTAtOTg0OG
NhNmFiNGM0IiwidCI6ImVlNGUxNDk5LTRhMzUtNGIyZS1hZ DQ3LT
VmM2NmOWRlODY2NiIsImMiOjh9 [Last accessed on 2022 Dec 20].
34. Centers for Disease Control and Prevention. Long COVID or Post-COVID
Conditions. Available from: https://www.cdc.gov/coronavirus/2019-
ncov/long-term-eects/index.html [Last accessed on 2022Sep01].
Table A1 reports pre-2022 descriptive statistics and correlations
weighted by countries’ population size. In addition to the pre-
2022 country-level variables we already reported on, it includes
a variable measuring 2021 all-cause mortality relative to the
2016–2019 average mortality and one measuring 2020 all-cause
mortality relative to the 2016–2019 average mortality.
We observe a strong negative correlation between vaccination
uptake and 2021 mortality, which may indicate a preventive eect
that year. Furthermore, we observe a strong positive correlation
between pre-COVID life expectancy and 2021 mortality, which
may indicate an alternative or complementary explanation to
the potential vaccine eect. Finally, we observe a strong positive
correlation between pre-COVID life expectancy and vaccination
uptake.
Table A2 regresses 2021 mortality on vaccination uptake
and pre-COVID life expectancy. Furthermore, it includes 2020
Table A1: Pre-2022 country-level descriptive statistics and correlations weighted by countries’ population size
Min. Max. Mean SD Variable 1 2 3 4
101.1 137.4 113.3 8.49 2021 mort. rel. to 2016–2019 avg. (1)
99.6 120.7 111.8 5.20 2020 mort. rel. to 2016–2019 avg. (2) 0.331†
101.4 126.1 112.5 5.59 Avg. 2020–2021 mort. rel. to 2016–2019 (3) 0.901*** 0.707***
75.1 84.3 81.5 2.47 Life expectancy in 2019 (4) −0.845*** 0.029 −0.620***
27.7 83.1 68.7 10.3 Vaccination uptake by the end of 2021 −0.862*** −0.211 −0.742*** 0.857***
Two-tailed tests of signicance. † P<0.10; * P<0.05; ** P<0.01; *** P<0.001 n=31
mortality (compared to the 2016–2019 average) as a control
variable. The regression aims to assess whether vaccination
uptake and pre-COVID life expectancy is genuinely associated
with 2021 mortality. The model shows that both independent
variables have negative associations with the dependent
variable, but only the pre-COVID life expectancy association
is significant. The non-significant association between
vaccination uptake and 2021 mortality is in line with the
largely non-significant direct association between vaccination
and 2022 mortality [Table 3]. However, high VIFs concerning
the independent variables may indicate multicollinearity
problems, particularly since the model has only 31 country
observations.[22] Altogether, we conclude that vaccination
uptake may have temporarily reduced mortality in 2021 (but
not in 2022, where the monthly mortality increases the higher
the vaccination uptake, as shown earlier in the study).
Ap p e n d I x

Jarle Aarstad and Olav Andreas Kvitastein: COVID-19 Vaccination Uptake and Excess All-Cause Mortality www.apjhs.com
Asian Pacic Journal of Health Sciences | Vol. 10 | Issue 1 | January-March | 2023
31
Table A2: Pre-2022 country-level linear weighted regressions with
robust standard errors and 2021 all-cause mortality compared to the
2016–2019 average as a dependent variable
Coe. Beta
Intercept 248.5***
(29.6)
2020 mort. comp. to the 2016–2019 avg. 0.481** 0.295
(0.152)
[1.27]
Life expectancy in 2019 −2.14** −0.633
(0.552)
[4.55]
Vaccination uptake by the end of
2021
−0.209 −0.257
(0.149)
[4.76]
R-square/R-square adj. 0.854/0.837
F-value 39.5***
Two-tailed tests of signicance. n=31. Estimates are weighted by country size
in population size by January 01, 2020, and we report robust standard errors
in parentheses. We report VIFs in brackets and beta values in the left column