The demographic and geographic impact of the COVID pandemic in Bulgaria and Eastern Europe in 2020

Abstract

The COVID-19 pandemic followed a unique trajectory in Eastern Europe compared to other heavily affected regions, with most countries there only experiencing a major surge of cases and deaths towards the end of 2020 after a relatively uneventful first half of the year. However, the consequences of that surge have not received as much attention as the situation in Western countries. Bulgaria, even though it has been one of the most heavily affected countries, has been one of those neglected cases. We use mortality and mobility data from Eurostat, official governmental and other sources to examine the development and impact of the COVID-19 pandemic in Bulgaria and other European countries. We find a very high level of excess mortality in Eastern European countries measured by several metrics including excess mortality rate (EMR), P-scores, potential years of life lost (PYLL) and its age standardised version (ASYR), and working years of life lost (WYLL). By the last three metrics Eastern Europe emerges as the hardest hit region by the pandemic in Europe in 2020. With a record EMR at ~0.27% and a strikingly large and mostly unique to it mortality rate in the working age (15–64 years) demographics, Bulgaria emerges as one of the most affected countries in Eastern Europe. The high excess mortality in Bulgaria correlates with insufficient intensity of testing, with delayed imposition of “lockdown” measures, and with high prevalence of cardiovascular diseases. We also find major geographic and demographic disparities within the country, with considerably lower mortality observed in major cities relative to more remote areas (likely due to disparities in the availability of medical resources). Analysis of the course of the epidemic revealed that individual mobility measures were predictive of the eventual decline in cases and deaths. However, while mobility declined as a result of the imposition of a lockdown, it already trended downwards before such measures were introduced, which resulted in a reduction of deaths independent of the effect of restrictions. Large excess mortality and high numbers of potential years of life lost are observed as a result of the COVID pandemic in Bulgaria, as well as in several other countries in Eastern Europe. Significant delays in the imposition of stringent mobility-reducing measures combined with a lack of medical resources likely caused a substantial loss of life, including in the working age population.

Full text

1
Vol.:(0123456789)
Scientic Reports | (2022) 12:6333 | https://doi.org/10.1038/s41598-022-09790-w
www.nature.com/scientificreports
The demographic and geographic
impact of the COVID pandemic
in Bulgaria and Eastern Europe
in 2020
Antoni Rangachev1,2, Georgi K. Marinov3* & Mladen Mladenov4
The COVID-19 pandemic followed a unique trajectory in Eastern Europe compared to other heavily
aected regions, with most countries there only experiencing a major surge of cases and deaths
towards the end of 2020 after a relatively uneventful rst half of the year. However, the consequences
of that surge have not received as much attention as the situation in Western countries. Bulgaria, even
though it has been one of the most heavily aected countries, has been one of those neglected cases.
We use mortality and mobility data from Eurostat, ocial governmental and other sources to examine
the development and impact of the COVID-19 pandemic in Bulgaria and other European countries.
We nd a very high level of excess mortality in Eastern European countries measured by several
metrics including excess mortality rate (EMR), P-scores, potential years of life lost (PYLL) and its age
standardised version (ASYR), and working years of life lost (WYLL). By the last three metrics Eastern
Europe emerges as the hardest hit region by the pandemic in Europe in 2020. With a record EMR at
~0.27% and a strikingly large and mostly unique to it mortality rate in the working age (15–64 years)
demographics, Bulgaria emerges as one of the most aected countries in Eastern Europe. The high
excess mortality in Bulgaria correlates with insucient intensity of testing, with delayed imposition
of “lockdown” measures, and with high prevalence of cardiovascular diseases. We also nd major
geographic and demographic disparities within the country, with considerably lower mortality
observed in major cities relative to more remote areas (likely due to disparities in the availability of
medical resources). Analysis of the course of the epidemic revealed that individual mobility measures
were predictive of the eventual decline in cases and deaths. However, while mobility declined as a
result of the imposition of a lockdown, it already trended downwards before such measures were
introduced, which resulted in a reduction of deaths independent of the eect of restrictions. Large
excess mortality and high numbers of potential years of life lost are observed as a result of the COVID
pandemic in Bulgaria, as well as in several other countries in Eastern Europe. Signicant delays in the
imposition of stringent mobility-reducing measures combined with a lack of medical resources likely
caused a substantial loss of life, including in the working age population.
e SARS-CoV-2 virus and COVID-19, the disease it causes1–3, have emerged as the most acute public health
emergency in a century. e novel coronavirus spread rapidly before signicant eorts at containment were
implemented in much of the world, resulting in devastating early outbreaks in the United States and Western
Europe, starting in late February and early March of 2020.
Some combination of lockdown measures, imposed in response to surging infections, voluntary changes in
behavior, and the onset of the summer season is thought to have caused the major decline in COVID-19 cases
in Europe in the summer of 2020. However, winter in the Southern hemisphere, during which large epidemics
developed in South Africa and South America, together with the well-documented seasonality of common-cold
coronaviruses4, strongly suggested that a major second wave was to be expected in Europe with the arrival of
winter5, and when it eventually arrived expectation turned into reality.
During the early months of the pandemic, a dichotomy emerged between countries in Western and Eastern
Europe (with the possible exception of Russia). Western Europe was heavily aected—by June 2020 ocial
OPEN
1Department of Mathematics, University of Chicago, Chicago, IL 60637, USA. 2Institute of Mathematics and
Informatics, Bulgarian Academy of Sciences, Soa 1113, Bulgaria. 3Department of Genetics, Stanford University,
Stanford, CA 94305, USA. 4Premier Research, Morrisville, NC 27560, USA. *email: GKM359@gmail.com
Content courtesy of Springer Nature, terms of use apply. Rights reserved

2
Vol:.(1234567890)
Scientic Reports | (2022) 12:6333 | https://doi.org/10.1038/s41598-022-09790-w
www.nature.com/scientificreports/
COVID mortality reached 600 to 800 deaths per million (DPM) in countries such as Spain, Italy, the UK,
Belgium, France, and Sweden, with excess mortality rates even higher6–10. In contrast, most Eastern European
countries registered relatively few deaths, possibly because of much earlier implementation of social distancing
measures relative to the development of the outbreak.
is dichotomy disappeared during the second wave at the end of 2020, with both countries in Western and
Eastern Europe ocially registering a large number of COVID-related fatalities, as well as in some cases consid-
erably larger excess mortality. However, the development of the pandemic in Eastern Europe has so far generally
received much less attention than that in the West even though multiple countries in the region were heavily
aected by it. We show this using multiple excess mortality measures, which quantify the pandemic-related loss
of life and allow for standardized comparisons between countries.
Among Eastern European countries, Bulgaria has emerged as perhaps the most heavily aected by the pan-
demic as suggested by excess mortality analysis6. Here we analyze the development and impact of the pandemic
on Bulgaria, in the broader European context, across demographic groups within the country, and for its regional
subdivisions, as well as the inuence of human mobility changes and government-imposed quarantine measure
on the course of the pandemic. We use these analyses to identify correlate factors likely responsible for particu-
larly high unexplained excess mortality in certain settings.
Results
Mortality during the COVID pandemic in Bulgaria. We analyzed overall excess mortality patterns in
Bulgaria for the year 2020 and compared it to data for other European countries, which submit mortality data
to Eurostat,for the same period. We focus on excess mortality rather than ocially registered COVID deaths
because limited testing and varying standards for ocial reporting of COVID deaths can result in large dis-
parities between public gures for COVID-related mortality and the actual burden the disease has imposed on
the population6. While some of the excess deaths are caused by the collapse of healthcare services during peak
moments of COVID waves, when a particularly large discrepancy between ocial COVID deaths and excess
deaths is observed, and when the shapes of the excess mortality and ocial COVID case and death counts match
closely, excess mortality is likely mostly due to underreporting of COVID deaths due to insucient testing and
other irregularities like reporting almost exclusively COVID-19 deaths that occurred in hospitals as it is the case
with Bulgaria.
In total, we estimate that 19,004 lives have been lost in Bulgaria in 2020 in excess of the baseline from previ-
ous years (Fig.1A). is amounts to an EMR of 2,734 DPM, or ~0.27%, for the year and ranks the country as
the most highly aected within the EU (Fig.1A; according to P-scores Lithuania, Italy and Spainrank higher).
COVID mortality is in most countries higher in males than in females32, and this is also what is observed in
Bulgaria and most other EU countries (Fig.1B,C). For females, an EMR of 2,248 DPM is observed (P-score of
19.3), compared to an EMR of 3,250 DPM for males (P-score of 24.1) across all ages.
ese observed EMR values are much higher than the ocially reported COVID-attributed population fatality
rate (PFR), by a factor of ~2.5
×
. Examination of the EMR/PFR ratios in Europe showed that excess deaths are
higher than ocial COVID death tolls in most countries (Fig.2). However, a clear dichotomy emerges between
Eastern and Western Europe, with the EMR/PFR ratio being considerably higher in countries in Eastern Europe
such as Bulgaria, Romania, Poland, Slovakia, Lithuania, and others.
ese estimates and geographic patterns are in agreement with other recent analyses of excess mortality33,34.
Loss of life as a result of the COVID pandemic. We next examined the impact of the pandemic in
terms of years of life lost using the PYLL and ASYR metrics based on excess mortality (Fig.3), where the latter
metric is the more suitable one for cross-country comparison. Both metrics paint a similar picture. However,
there are some notable dierences when using P-scores from (Fig.1). For instance, Italy, Spain and Belgium
are among the countries with highest P-scores - 2nd, 3rd, 7th, respectively – but in terms of ASYR valus these
countries rank 9th, 10th, and 14th.
Using standardized ASYR and PYLL values (per 100,000 population; Supplementary Figs.2A-C and 1), we
nd that the highest total loss of life among the examined countries occurred in Lithuania and Bulgaria for
both males and females, followed by Romania, Poland, Serbia, Montenegro, Czechia and Hungary. Out of the
top 13 countries in SupplementaryFig.1A 11 are in Eastern Europe. is higher loss of life burden in Eastern
European countries is explained not only by their high EMRs but also by a large numbers of deaths in younger
age groups (
<65
years). Lithuania and Bulgaria exhibit 3,351 and 3,195 years of life lost per 100K standardised
population, whereas countries with similar P-scores like Italy and Spain show 1,506 and 1,498 years of life lost
per 100K standardised population, respectively. is drastic dierence is explained by the age distribution of
excess deaths relative to the life expectancy for each country.
Calculation of WYLL values, which show the loss of working years of life, showed Lithuania and Bulgaria to
have incurred the highest such loss within the set of examined countries (Supplementary Fig.2D-E; note that
the high total WYLL value for Iceland is possibly an artifact of the small population of the country). In Lithuania
and Bulgaria,
25%
and
21%
of excess deaths, respectively, are of people in the age group 40–64. In contrast, only
7%
of excess deaths in Italy and Spain are in the age group 40–64 (see Fig.4D).
In countries such as Italy, Spain, France and Belgium, only 17–20% of excess deaths are under 75 years of age,
while 67–70% of all excess deaths in these countries are in the age group
80+
.
For Bulgaria, we nd an average PYLL value of
12.91 ±0.08
in total,
12.24 ±0.24
for males, and
12.67 ±0.08
per female (Supplementary Fig.2). Excluding outliers (note that average PYLL values based on excess mortality
are very high in countries such as Iceland, Luxembourg due to stochasticity associated with the very low number
of excess deaths), these values are generally higher than what is seen in Western Europe. e only three countries
Content courtesy of Springer Nature, terms of use apply. Rights reserved

3
Vol.:(0123456789)
Scientic Reports | (2022) 12:6333 | https://doi.org/10.1038/s41598-022-09790-w
www.nature.com/scientificreports/
Figure1. Excess mortality in Bulgaria and other European countries in 2020. (A) Overall P-scores and
excess mortality (in deaths per million; DPM) for all ages in Bulgaria (highlighted in red) and other European
countries; (B) P-scores and excess mortality for females of all ages; (C) P-scores and excess mortality for males
of all ages. All error bars in this and subsequent gures represent 95% condence intervals.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

4
Vol:.(1234567890)
Scientic Reports | (2022) 12:6333 | https://doi.org/10.1038/s41598-022-09790-w
www.nature.com/scientificreports/
with an average PYLL
≥
13 are Estonia, Lithuania, Serbia, and Bulgaria, compared to values in the 7 to 9.5 years
range for countries such as Switzerland, Sweden and Belgium. Despite males exhibiting higher mortality due
to COVID-19, the average PYLL based on excess deaths in Bulgaria is higher for females (it is also higher for
females in several other European countries; Supplementary Fig.3).
Using ocial COVID-attributed deaths, for Bulgaria we obtain an average of 12.37 years lost for males and
14.01 years lost for females. Based on the ocial COVID-19 mortality data for Czechia (the other country for
which exact data about the age of the diseased was availableto us) we obtain 9.78 and 9.35 for males and females,
respectively. In both cases, the estimates we obtain for the average PYLLs from excess mortality and ocial
COVID-19 deaths data are in agreement (note, however, that there are substantial dierences between Bulgaria
and Czechia in other aspects – for example, the average age of ocially registered COVID-19 deaths for women
is 71 years in Bulgaria compared to a life expectancy of 78.4 years, while in Czechia, the average age of the female
COVID-19 deaths is 80.81, which is very close to the 82.1 life expectancy for women in that country).
ese observations suggest that the impact of the pandemic in late 2020 on countries in Eastern Europe
was not only large in absolute terms but also aected heavily younger demographics than in Western Europe.
One possible explanation for this discrepancy is the underlying comorbidity structure of population. Car-
diovascular diseases (CVD) are a known risk factor for severe COVID, so we carried out a correlation analysis
between the dierent excess mortality metrics we used and the prevalence of CVDs (based on 2018 data) in
Europe. As a direct indicator for CVD prevalence we used CVDs death rates. We found strong correlation
between PYLLs and CVD death rates, ASYRs and CVD death rates, and between WYLLs and CVD death rates
restricted to age
<65
years (Supplementary Fig.9); correlation between EMR values/P-scores and CVDs was
not statistically signicant. ese correlation warrant further investigation.
Demographic-specic mortality patterns in Bulgaria. By the ocial statistics of the Ministry of
Health18,19 the average age of a deceased male and female from COVID-19 are 69 and 71, respectively. e lead-
ing comorbidity is cardiovascular disease (
55%
), followed by diabetes (
17%
), pulmonary disease (
12%
), obesity
(
3%
), and
30%
are listed with no known comorbidity. An overwhelming majority of
94.5%
of all 7,576 ocial
COVID-19 deaths occurred in the hospitals with working age deaths comprising
28%
of all COVID-19 deaths.
For the working age group females on average died at age 55.9 and males at age 55.7 with
45%
of the deceased
having a cardiovascular disease.
Data on excess mortality for people under 65 reveals a slightly dierent picture. e working age group excess
deaths are
21%
of all excess deaths with an average age of the deceased
55.65 ±0.07
for men and
57.57 ±0.28
for
women. e reason for the higher average age for women is that our data does not reveal excess deaths in women
under 40, whereas in the ocial statistics
5%
of the casualties are of ages between 10 and 39.
Figure2. Ratio between excess mortality and ocial COVID-attributed deaths in European countries in 2020.
Note that the high EMR/PFR ratios for 2020 in countries like Finland and Estonia might be an artifact of overall
low both excess and COVID-attributed mortality.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

5
Vol.:(0123456789)
Scientic Reports | (2022) 12:6333 | https://doi.org/10.1038/s41598-022-09790-w
www.nature.com/scientificreports/
Next, we examine mortality in Bulgaria within the working age population in detail. Due to the well-docu-
mented age-related skew of COVID fatalities, we focused on two subgroups of working age individuals – those
in the 30–39 and those in the 40–64 age ranges.
We nd no elevated mortality in females in the 30–39 age group, while mortality is elevated in males of the
same age bracket, with P-scores of –0.39 and 9.37, respectively (Fig.4A–C).
In contrast, we nd highly elevated excess mortality in both males and females in the 40–64 age group, in
which Bulgaria and Lithuaniarank highest in the EU (Fig.4D–F), with P-scores of 23.2 and 28.6 for Bulgar-
ianmales and females, respectively. e dierence between males and females is remarkable, as, unlike the
typical situation, elsewhere in the world in this group in Bulgaria excess mortality measured by P-scores is lower
for males than for females. We discuss the possible explanation for these observations in subsequent sections.
Figure3. Geographic distribution of excess mortality-based ASYR and PYLL values for European countries in
2020. Shown are the total (per 100K people) values. (A) ASYR values for the whole population; (B) ASYR values
for females; (C) ASYR values for males; (D) PYLL values for the whole population; (E) PYLL values for females;
(F) PYLL values for males.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

6
Vol:.(1234567890)
Scientic Reports | (2022) 12:6333 | https://doi.org/10.1038/s41598-022-09790-w
www.nature.com/scientificreports/
Regional disparities in COVID pandemic-related mortality in Bulgaria. Following from the obser-
vation of considerable disparities between dierent European regions, we then analyzed regional dierences in
pandemic impacts within Bulgaria (Fig.5). As a reminder, the overall statistics for Bulgaria are an EMR of
0.27%
,
P-score of
21.8%
, CFR of
3.7%
, an EMR/PFR ratio of 2.5, and a percentage of population tested positive of
2.9%
.
e rst major such disparity we observe is that between the four most populated provinces and the rest of
the country. e excess deaths in these four major regions – Soa (city), Plovdiv, Varna and Burgas – account for
just
32%
of all excess deaths even though
≥50%
of the Bulgarian population lives there. Moverover, Soa (city),
Varna and Burgas have the lowest EMR of all provinces (Fig.5A) and P-scores in the range 12–25% (Fig.5B).
e provinces of Soa (city) and Burgas also show the two lowest CFR values (Fig.5F).
In contrast, the more peripheral regions are among the most heavily aected. For example, the regions of
Vidin and Silistra exhibit some of the highest EMRs – 0.46 and 0.40, respectively. Vidin also has the second
highest CFR (
8%
). e close to the average for the country P-score of
24%
in Vidin is likely a result of already
very high pre-pandemic mortality in the region (the region has one of the fastest aging populations in the EU
and the death rate there is 22 per 1000 people per year, whereas the death rate for Bulgaria is 15 deaths per 1000
people per year). In Silistra, the EMR/PFR ratio is 3.00, the second highest in the country and the P-score is
28%
. In Kardzhali, the EMR/PFR ratio is 4.00, the highest in the country, and the P-score is
22%
. With a P-score
of
30%
and EMR of 0.40, Smolyan is one of the hardest hit regions in the country, and it also has a CFR of
8.9%
,
which is the highest among all provinces.
ese regions also tend to show a lower percentage of the population that has tested positive, despite exhibit-
ing the highest excess mortality (Fig.5E), oen high CFRs (Fig.5F), and high EMR/PFR ratios (Fig.5G).
We also nd curious disparities in regional patterns of male- and female-specic excess mortality (Fig.5C,D).
e highest male excess mortality was observed in Pazardzhik, Gabrovo, Soa (region) and Smolyan, while the
highest female excess mortality is seen in Blagoevgrad, Silistra, Pazardzhik and Targovishte.
Figure4. Excess mortality in working age populations in Bulgaria and other European countries in 2020. (A)
P-scores for the overall population in ages 30–39; (B) P-scores for females in ages 30–39; (C) P-scores for males
in ages 30–39; (D) P-scores for the overall population in ages 40–64; (E) P-scores for females in ages 40–64; (F)
P-scores for males in ages 40–64.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

7
Vol.:(0123456789)
Scientic Reports | (2022) 12:6333 | https://doi.org/10.1038/s41598-022-09790-w
www.nature.com/scientificreports/
Remarkably, only
32%
of all excess deaths in the working age group occurred in Soa (city), Plovdiv, Varna
and Burgas. For women in the working age group only
29%
of the deaths occurred in those regions. A regional
analysis reveals that the regions that with the highest P-scores in this demographic category are Blagoevgrad, Sil-
istra, Pazardzhik, Targovishte, Kardzhali, Kyustendil and Dobrich ranging from
52% ±6%
to
36% ±7%
(Fig.5C).
Women in the age group 65–69, which includes working women in retirement age, were also heavily aected
with an overall P-score of
23.7%
and exceptionally high regional P-scores in the provinces of Sliven -
76% ±10%
,
Kardzhali -
46% ±8%
Blagoevgrad -
45% ±7%
, Pazadzhik -
43% ±6%
, and Smolyan -
42% ±11%
and (Sup-
plementary Fig.4). We discuss the possible explanations for these observations in the Discussion section.
ese observations suggested that there might also be signicant disparities in the impact of the COVID
pandemic within regions, i.e. regional centers showing better outcomes than peripheral municipalities within
Figure5. Regional disparities in the impacts of the COVID-19 pandemic in Bulgaria. (A) Overall excess
mortality in Bulgarian regions (EMR units); (B) Overall excess mortality in Bulgarian regions (P-score); (C)
Excess mortality in working age (40–64) females in Bulgarian regions (P-score); (D) Excess mortality in working
age (40–64) males in Bulgarian regions (P-score); (E) Percentage of the population who have tested positive
for SARS-CoV-2 in Bulgarian regions; (F) CFR values for Bulgarian regions; (G) Ratio between excess deaths
(EMR) and ocial COVID-19-attributed deaths (PFR).
Content courtesy of Springer Nature, terms of use apply. Rights reserved

8
Vol:.(1234567890)
Scientic Reports | (2022) 12:6333 | https://doi.org/10.1038/s41598-022-09790-w
www.nature.com/scientificreports/
each region. Examination of available data at the level of individual municipalities in Bulgaria indeed shows such
as a pattern (Fig.6, Supplementary Fig.6).
CFR values are lower in regional centers, and the most heavily aected municipalities in each region tend to
be smaller ones outside the region’s core city, and this is not an artifact of the their small population size. Some
examples include: in Vidin, the region center’s P-score is 19.5% (±4.0%) while in Kula it is 39.4% (±10.5%); in
Pernik, Pernik’s P-score is 16.3% (±1.3%) while in Tran it is 42.3% (±15.0%); in Gabrovo, Gabrovo’s P-score is
9.9% (±2.5%) while Tryavna has a P-score of 39.5% (±9.2%) and Dryanovo has a P-score of 36.3% (±9.4%).
The trajectory of the pandemic in Bulgaria and the eectiveness of implemented pandemic
control measures. Finally, we mapped the trajectory of the pandemic in Bulgaria onto the timeline of
imposition of social distancing measures and independent measures of actual changes in societal mobility to
understand the relationship between those factors and its development (Fig.7).
e rst period of the COVID-19 pandemic in Bulgaria, from March to the end of September, was marked
by a slight elevation of the new conrmed cases in the summer, peaking at 242 cases per day towards the end of
July. e excess mortality for this period is around 300 people and the ocial COVID-19 death toll amounts to
820 people, with a daily death rate of up to 10 deaths until the middle of October.
Figure6. Regional disparities in the impacts of the COVID-19 pandemic in Bulgaria at the county/
municipality level. (A) Overall excess mortality in Bulgaria at the county/municipality level; (B) CFR values
for Bulgaria at the county/municipality level (note that there are two municipalities named “Byala”, and
available data does not distinguish between the two, thus they are colored in white as missing data. Regions are
demarcated in black, regional centers are shown as black dots.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

9
Vol.:(0123456789)
Scientic Reports | (2022) 12:6333 | https://doi.org/10.1038/s41598-022-09790-w
www.nature.com/scientificreports/
Figure7. Development of the COVID pandemic in Bulgaria over 2020 and the eectives of measures implemented in order
to control it. (A) Number of tests conduced and test positivity percentage over the second half of 2020. aOcial daily testing
data was only made available from 06 Jun 2020–Open Data Portal (https:// data. egov. bg/ data/ resou rceVi ew/ e59f9 5dd- afde-
43af- 83c8- ea291 6badd 19). (B) Ocially registered COVID cases. Note that rapid antigen tests were only included in statistics
starting from December 22nd 2020. (C) Ocially registered weekly COVID deaths and overall weekly excess mortality over
the course of 2020. (D) Social mobility changes and the timing of imposition of restrictions. Arrows indicate the time of
imposition of “lockdown” measures. “Time Home” refers to the change of the number of visitors to residential areas relative to
the period before the pandemic. “Time Retail and Recreation” refers to the change of the number of visitors to places of retail
and recreation relative to the period before the pandemic. is includes restaurants, cafes, shopping centers, theme parks,
museums, movie theatres, libraries.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

10
Vol:.(1234567890)
Scientic Reports | (2022) 12:6333 | https://doi.org/10.1038/s41598-022-09790-w
www.nature.com/scientificreports/
Rapid growth in the number of new conrmed cases started around the end of September. en an explo-
sion of cases occurred in late October, November and the rst half of December (Fig.7B). e peak in the 7-day
moving average of the number of conrmed cases occurred on November 19th.
Ocial COVID-19 deaths peaked on December 6th with a 7-day moving average of 140, or ~18 DPM/day;
excess deaths started decreasing around the same time (Fig.7C). Excess deaths began diverging from ocial
statistics with the start of the Fall surge, in the middle of October, and peaked at ~54 DPM/day in the week
ending on November 27th. is corresponds to a
112.3%
increase in relative age-standardised mortality rates
(rASMRs) according to the ONS35; a higher number in Europe in 2020 was observed only in Spain for the week
ending on April 3rd at at
142.9%
.
One of the obvious candidate explanations for the discrepancy between ocial and excess deaths is insuf-
cient testing36.
Indeed, that appears to be the case for Bulgaria. Test positivity rates peaked ~40% in late November. However,
a curious pattern is observed in the number of tests recorded in ocial statistics, which actually began decreasing
while the positivity was still increasing in the month of November (Fig.7A). An explanation for this pattern is
that the results of rapid antigen tests were not included in ocial statistics until late December, and a consider-
able portion of testing shied from PCR to antigen tests as the Fall wave developed. is likely accounts for at
least some of the discrepancy between recorded and excess deaths.
We then examined the factors responsible for the Fall surge eventually receding using the stringency index
and mobility metrics (see Methods), changes in which have been shown before to be predictive of the trajectory
of COVID epidemics37–40.
e stringency index was at 35.19 from mid September until October 29th (Figs.7D,8B), when the Bulgarian
government imposed some new restrictions (high schools and universities moved to remote learning; nightclubs,
Figure8. Mobility metrics, stringency of restrictions and mortality and cases at the peak of the late-2020
wave in Bulgaria. (A) Google Mobility Data and Stringency Index at the peak of the fall wave in Bulgaria and
other EU countries. “Time Home” refers to the change of the number of visitors to residential areas relative to
the period before the pandemic. “Time Retail and Recreation” refers to the change of the number of visitors
to places of retail and recreation relative to the period before the pandemic. is includes restaurants, cafes,
shopping centers, theme parks, museums, movie theatres, libraries. (B) Timeline of imposition of social
distancing measures and of reductions in mobility in Bulgaria around the peak of the late-2020 wave. e peak
occurred around November 11th 2020, as demonstrated by mortality data, which at any given moment reects
the dynamic of new cases in Bulgaria approximately 2.5 weeks prior to that moment.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

11
Vol.:(0123456789)
Scientic Reports | (2022) 12:6333 | https://doi.org/10.1038/s41598-022-09790-w
www.nature.com/scientificreports/
pubs and bars were closed), which is reected by an increase in the stringency index to 48.15. No further sub-
stantial epidemiological measures were introduced until aer the peak of the fall wave – on November 27th,
restaurants, bars, malls, schools and gyms were closed. However, the stringency index, though now increased to
53.7%, remained considerably below the levels of restrictions imposed in other European countries (Fig.8A), as
no stay-at-home orders or curfews were imposed, non-essential stores and hair-dressing salons remained open,
and gatherings of up to 15 people were permitted.
As the peak of restrictions occurred around the time of the peak of excess mortality and thus aer the peak of
infections, it is likely that restrictions were not the main cause for the eventual decline in cases. Indeed, changes
in people’s behavior as reected in social mobility measures were observed much earlier than the imposition of
restrictions, likely due to fear of becoming infected spreading among the population, a pattern previously noted
elsewhere in the world41. Our data supports this:
40%
of the total increase from October 1st to November 27th
of the time spent at home occurred in the period November 3rd to November 27th when no new substantial
measures were introduced (see8B).
e 7-day running average Google mobility data measured on November 19th shows a total decrease of at
home compared to the baseline (Figs.7D,8B). However, as with the stringency index, these values are still the
lowest among analyzed European countries (Fig.8B), which is likely a contributing factor to the very high excess
mortality resulting from the pandemic.
Finally, we examine hospitalization trends in Bulgaria and several other European countries. We nd that
at their peak on December 12th, hospitalizations in Bulgaria reached a level of
0.1%
of the population, which
is one of the highest hospitalization rates up to date (it has since been exceed by Hungary and by Bulgaria itself
during the subsequent March surge; Fig.9).
Discussion
In this survey, we analyze the impact of the COVID-19 pandemic in 2020 in Bulgaria and the broader Eastern
European context. Aer a relatively low level of COVID-19 cases and deaths prior to that, in the concentrated
span of less than three months in October, November, and December 2020, Bulgaria recorded the largest (per
capita) number of excess deaths among the examined countries. Similar, though somewhat lower, large excess
mortality increases were observed in most other countries in Eastern Europe. However, ocial COVID-19-at-
tributed deaths account for only less than half of the excess deaths.
is discrepancy is likely caused by a combination of multiple factors – COVID-19 cases leading to death
that were not reported as such in ocial statistics, COVID-19 cases resulting in death some time aer recovery
due to longer-term complications from the disease, and deaths from other causes that increased as a result to
the inability of the healthcare system to treat them due to it being overwhelmed by COVID-19 patients. As the
disparity between excess deaths and ocial COVID-19 mortality is very large in the case of Bulgaria – excess
deaths amount to ~0.27% of the population while ocial COVID-19 are at ~0.11%, i.e. a ~0.16% dierence – and
excess mortality is highly temporally concentrated in a short time span of about ten weeks (i.e. the contribution
of the latter two factors is unlikely to have been so large in such a brief period), it is most likely that the bulk of
excess deaths were caused directly by COVID-19.
Why they were not recorded as such is also probably due to a multitude of factors. Testing in Bulgaria has
been greatly insucient throughout the pandemic and even more so during the late-2020 surge and the low-
est among the examined countries (Supplementary Fig.8; in addition to that, the decision to not include rapid
antigen tests in public statistics certainly has contributed to the underreporting. Most reported deaths occurred
in hospitals, and many of those who could not be hospitalized due to healthcare systems being overwhelmed
and died at home were not recorded as having died of COVID. Whether additional social and socioeconomic
factors could have contributed, as has been suggested to be the case elsewhere in the world42, is a subject for
Figure9. Peak hospitalizations in Bulgaria and other countriesfrom the beginning of the pandemic till April
2021.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

12
Vol:.(1234567890)
Scientic Reports | (2022) 12:6333 | https://doi.org/10.1038/s41598-022-09790-w
www.nature.com/scientificreports/
future investigations, as is the question of whether the reasons for underreporting are uniform across the more
general Eastern European region. Lack of testing on its own in turn has probably contributed to the epidemic
growing out of control and leading to such a number of excess deaths.
Another contributing factor to the high mortality rate in Eastern Europe is probably the very high prevalence
of cardiovascular diseases in the region43 also suggested by our correlation analysis in Supplementary Fig.9.
In Bulgaria over half of the COVID-19 ocially reported fatalities are listed with cardiovascular disease as a
comorbidity.
Bulgaria also exhibits one of the most highly elevated working-age excess mortality, and it is also an outlier in
terms of working-age excess mortality among females. We also observe signicant regional disparities within the
dierent regions in the country in total and in working age sex-specic excess mortality. A possible explanation
for the latter is the development of outbreaks at workplaces where mostly women work – for example, garment,
textile and shoe factories, which in Bulgaria almost exclusively employee women and which are major sectors of
the economy in provinces such as Blagoevgrad, Kardzhali, Smolyan, Sliven, and Kyustendil44. Indeed, there were
numerous reports about outbreaks in such settings. Analogous causality might be behind regional disparities
in working age male-specic excess mortality (Fig.5D). A list of reports about outbreaks in these regions can
be found in our GitHub repository, which includes reports about outbreaks in battery, automotive parts, power
transmission, sanitary ceramics, and other factories.
Regional disparities in overall excess mortality, in particular the clear dichotomy emerging between the major
population centers, in which generally better outcomes are observed, and the more heavily aected peripheral
regions, also warrant further investigation. COVID-19 is still oen considered a disease that impacts highly
populated big cities the most, where disease spread is thought to be facilitated by density; this is due to many of
the most notable initial outbreaks aecting well-connected in terms of international travel large metropolitan
areas. However, as the pandemic has spread throughout the countries that have not controlled it, it may be
the case that previously established regional disparities in healthcare infrastructure are becoming a key factor
determining dierential outcomes between generally better resourced major cities on one hand, and the less
equipped to test, track and treat COVID-19 patients countryside areas. ere is evidence that such causation is
at play in Bulgaria – many of the heavily aected regions have fewer ICU beds, fewer doctors, and fewer special-
ists in the most relevant to the treatment of COVID-19 specialties than the capital and a few other major cities
(Supplementary Fig.7). For example, Vidin and Silistra have fewer than average hospital beds, Kardzhali has the
lowest number of doctors, general practitioners and pulmonologists and the second to last number of ICU beds
per capita in the country, and Smolyan has the lowest number of ICU beds (just 9 in total for the whole region)
and a generally low number of doctors.
In addition, in some of these regions (e.g. Smolyan, one of the most heavily impacted in the country) there are
purely geographic factors that may have complicated the timely treatment of patients due to the logistic challenges
of transporting patients to the regional center (which is where the only ICU units are located) from remote small
towns through mountainous terrain while the core city’s health infrastructure is itself under immense stress (as
shown in Fig.9, Bulgaria recorded record hospitalization levels during the peaks of the pandemic). For example,
four of the peripheral municipalities in the Smolyan region have twice as high CFR values as the city of Smolyan
(see Fig.6B). Whether similar regional patterns of pandemic-related excess mortality are observed in other
areas of Europe will be informative and instructive for minimizing the impact of subsequent COVID-19 waves.
It should also be noted that healthcare disparities possibly play a role on a broader-level45–47, as Eastern
Europe’s healthcare systems as a whole are well-documented to be suering from an outow of skilled medical
labor due to large numbers of doctors and nurses emigrating to Western Europe in recent years48.
However, the main factor behind the very high levels of excess mortality is still most likely the late imposition
of restrictions on social mobility and lax governmental eorts at controlling the spread of SARS-CoV-2, as our
analysis shows. In Bulgaria these were adopted long aer exponential growth in cases had commenced and was
clearly going to overwhelm hospital resources, little testing was carried out and insucient eorts were made
to ensure the isolation of infected individuals, and even when restrictions were imposed, they were generally
the most lax in Europe; furthermore, the late-2020 epidemic appears to have begun to trend downward due
to changes in individual behavior, the onset of which actually preceded the imposition of restrictions by the
government. e high levels of excess mortality are probably a natural consequence of following these policies.
Methods
Data sources. All-cause mortality data for European countries, as well as NUTS-3 (Nomenclature of Terri-
torial Units for Statistics) regions in Bulgaria, was obtained from Eurostat11,12. e data presented in the datasets
is sex- and age-stratied, with age groups split in increments of 5 years. Since not all countries submit data at the
same time and in the same manner, only countries that have consistent weekly data for the period 2015–2020
(inclusive) were analyzed.
Country-level population data at the beginning of 2020 was collected through Eurostat13, but was further
supplemented by population data from the United Nations’ UNdata Data Service14. We further elaborate on this
topic in the subsequent section on Potential Years of Life Lost (PYLL) and Working Years of Life Lost (WYLL)
estimates.
Life expectancy values at dierent ages were obtained from three separate sources. We acquire the full life
tables for Bulgaria through the country’s National Statistical Institute15, and for Czechia through the country’s
Statistical Oce16. Abridged life tables for all European countries were obtained from the World Health Organi-
zation’s open data platform17. is dataset is partitioned by age, in increments of 5 years.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

13
Vol.:(0123456789)
Scientic Reports | (2022) 12:6333 | https://doi.org/10.1038/s41598-022-09790-w
www.nature.com/scientificreports/
COVID-related mortality and testing data for Bulgaria was collected through the resources available from
the Ministry of Health18,19. COVID-related mortality for Czechia was acquired from Czech Ministry of Health
ocial website tracking the pandemic20.
Excess mortality and P-scores. To calculate excess mortality across countries as well as across Bulgarian
regions, we analyze the mortality observed between week 10 and 53 of 2020 and compare it to expected (base-
line) mortality for 2020 using the historical data for the previous ve years (2015–2019). e model we used is
the Karlinsky–Kobak regression model6:
where
D
t
,
Y is the number of deaths observed in week or month t in year Y,
β
is a linear slope across years, and
αt
are separate intercepts (xed eects) for each week or month and
ǫ∼N(0, σ2)
is Gaussian noise. e model
prediction for 2020 is
Expected Mortality
t
,2020 =ˆ
αt
+ˆ
β
·2020
. We then establish a 95% condence interval for
the expected mortality. is range is used to calculate the excess mortality
t
for a week or a month t as:
is calculation is done both as a sex- and age-stratied metric, as well as an aggregated total excess mortality
for 2020, which we denote by

. To normalize excess mortality across countries, we calculate excess mortality
per total population. To do this we use population data from Eurostat for 2020.
Set
z:= |�|/√Var[�], where Var[�]
is computed in6. If
z
is signicantly below 2 for a given country, we
consider the excess mortality for this country to be not signicantly dierent from zero. In the computations
related to the years-of-life lost metrics considered in the paper, we excluded a few countries having both
z
-values
signicantly below 2 (typically lessthan 1) for each age interval and wide condence intervals that included 0
for the excess mortality associated with each of these age intervals.
Based on the excess mortality ranges we also compute a P-score value for each country/region. A P-score
value is dened as the ratio or percentage of excess deaths over certain period relative to the expected deaths for
the same period based on historical data from the years 2015–2019 (see21). We calculate the P-score as follows:
We also calculate the ratio between excess mortality and ocial COVID-19-attributed mortality. Due to the
demonstrably low testing in Bulgaria22 and other countries, this allows us to estimate under-reported COVID-19
fatalities. We also use the total positive tests per region reported at the end of 2020 to compute a Case Fatality
Ratio (CFR) which estimates the proportion of COVID-19 fatalities among conrmed cases.
Potential Years of Life Lost (PYLL), Aged-Standardized Years of life lost Rate (ASYR), and
Working Years of Life Lost (WYLL) estimates. Potential Years of Life Lost (PYLL) is a metric that
estimates the burden of disease on a given population by looking at premature mortality. It is derived as the
dierence between a person’s age at the time died and the expected years of life for people at that age in a given
country. As such, the metric attributes more weight to people that have died at a younger age.
We compute the PYLL across countries by taking the positive all-cause excess mortality for all ages groups
(in Eurostat they are aggregated at 5 year intervals). We use the abridged life expectancy tables by the WHO
(also aggregated at 5 year intervals) and calculate a total and average PYLL value for all countries. To be more
precise, for an age interval
[x,x+4]
and sex s (if no sex is specied we assume it’s for both sexes) dene by
ED([x,x+4],s)
the excess deaths and by
LE([x,x+4],s)
the life expectancy. en the potential years of life
lost are computed as
e total PYLL is computed by summing over all age intervals. In our computations we take into account the
margin of error for each
ED([x,x+4],s)
.
A limitation on this approach is the upper-boundary aggregation value for the two datasets. e all-cause
mortality dataset’s upper boundary is 90+, while the WHO’s abridged life tables only go up to the 85+ age bracket.
To account for this, we attribute the life expectancy of the 85+ age group to the 85-89 mortality group. We have
further excluded the 90+ mortality group from our analysis. is is further elaborated on in the Limitations
subsection, where we also provide a way of correcting for this exclusion.
Two countries for which we have the exact ages and sex for each reported COVID-19 fatality are Bulgaria
and Czechia. We also have full life tables (increments of one year) for both countries provided by their respec-
tive statistical institutes. is allows us to compute and compare the PYLLs for each country based on excess
mortality data and ocial data for COVID-19 fatalities.
Finally, we standardize PYLL values across countries by diving the total sum value by the population and
normalizing it per 100,000 people:
Dt,Y=αt+β·Y+ǫ

t
=Mortality
t
,2020 −Expected Mortality
t
,2020.
P
:=
Mortality
2020
−Expected Mortality
2020
Expected Mortality2020
∗
100
PYLL([x,x+4],s)=ED([x,x+4],s)∗LE([x,x+4],s).
PYLL
std :=
PYLL
total
Total Country Population
0−89
∗
100, 000
Content courtesy of Springer Nature, terms of use apply. Rights reserved

14
Vol:.(1234567890)
Scientic Reports | (2022) 12:6333 | https://doi.org/10.1038/s41598-022-09790-w
www.nature.com/scientificreports/
e data for country-level populations in Eurostat has a similar limitation in the upper boundary of the age
distribution (a cut-o at 85+). To mitigate this limitation, we supplement the population data from Eurostat for
ages 0-84 with population size data for the 85-89 age group from the UNdata Data Service.
To compare the impact of the pandemic across European populations with dierent age structures we com-
pute the Age-Standardized Years of Life Lost Rate (ASYR)23,24. Let
([x,x+4],s)
as be an age interval for a sex
s in a standard life expectancy table for a given population. Denote by
P([x,x+4],s)
the population size of
P([x,x+4],s)
. Dene the PYLL rate for
([x,x+4],s)
For the 2013 European Standard Population (ESP) denote by
W([x,x+4],s)
the weight of
([x,x+4],s)
in the
standard population. Dene
where the sum is taken over all age intervals. For a given population of sex s this measure is interpreted as the
years of life lost per 100,000 people (of sex s) if the population has the same aged is tribution as the ESP. ASYR
allows for comparison of the pandemic impact on EU countries having dierent age distributions. Finally, we
derive total average and total standardized WYLL value approximations. To accomplish this we rst assume
people to be in the working age group if they are 15 to 64 years old, and thus exclude excess mortality for all age
groups over 65. To calculate the remaining years of working life, we further assume a mean age for each age group,
e.g. for the age interval 60–64 we assume a mean age at 62.5 years. is would leave this group with approximately
2.5 years until retirement. Limitations on this approach are discussed in the subsequent Limitations subsection.
Stringency index and mobility data. Metrics of population mobility were obtained from the Google
COVID-19 Community Mobility Reports25. ese datasets contain data on how visits and length of stay at dif-
ferent places change compared to a baseline by generating anonymized metrics from data of Google users who
have switched on “Location History” on their mobile devices.
To quantify governmental pandemic-response measures across countries, we used the Oxford COVID-19
Government Response Tracker26, which systematically collects information on several dierent common policy
responses that governments have taken to mitigate the eects of the pandemic27. is allows a comparison of
governmental measures between over 180 countries worldwide.
Limitations. Each of the presented data sources and approaches to analysis have their own limitations.
Below we discuss each one in detail.
Limitation of scope. e current time frame that is analyzed creates a hard boundary between week 10 and
week 53 of 2020. e exit conditions of dierent countries at these boundaries, however, are not equal. Some
countries experienced subsequent surges in January 2021 and later months. us the current research provides
a snapshot of the eects of the pandemic up to the end of 2020, not the totality of its eects.
Limitation of data. All cause mortality gures for 2020 are still provisional for most EU countries, so they are
subject to readjustment in future time. Even so, they can provide a good estimate of the eect of COVID-19 in
dierent countries up to this point.
Limitation of excess mortality and P‑scores. Inuenza outbreaks in the period 2015–2019 contribute to the
estimation for the expected mortality for 2020. us the expected mortality is an estimate of the “normal” death
rate in the presence of seasonal inuenza.
Since the P-score metric we compute is derived from the excess mortality gures we calculate for each indi-
vidual country, this metric also suers from the issues we outline for excess mortality.
Limitations of PYLL/ASYR/WYLL. Since PYLL, ASYR and WYLL data only take into account fatalities, these
metrics do not provide information about any worsened quality of life of surviving individuals, reduced life
expectancy of these individuals and working capacity. Metrics such as Disability-Adjusted Life Years (DALY),
Quality-adjusted life year (QALY) and Healthy Years of Life (HALE) metrics may illuminate further the total
disease burden on the European population, however, obtaining the necessary information for these measure-
ments is not yet possible.
As mentioned before, due to data availability limitations from Eurostat in our computations of PYLLs and
ASYRs we excluded the 90+ group. Given that countries like France, Italy and Spain have signicant excess
mortality in this age group we also present a computation of the ASYRs including the 90+ age by assuming 4
years of life expectancy (the average life expectancy for the 90+ age group for the European populationis 4.74,
according to the UNdata Data Service). By linear interpolation, 4 is approximately the life expectancy for the
age interval [90,94] assuming that the average age of deaths for this interval is in the range 92.7-93 (the average
age of the COVID-19 fatalities above 90 years of age is 92.7 in Czechia and 92.1 in Bulgaria, and likely it’s higher
in Western countries which overall have higher life expectancy).
is rough approximation gives an upper bound of how large the ASYRs can go. It leads to 5–14% and 14–22%
increase in the ASYRs for the
(0−89)
population of Eastern and Western European countries, respectively, but it
PYLL
rate(
[
x,x
+
4
]
,s)
:= PYLL([x,x+4])
P
([
x
,
x
+4],
s
)∗100, 000.
ASYR
(s)
:= 
PYLLrate(
[
x,x
+
4
]
,s)
∗
W(
[
x,x
+
4
]
,s
)
Content courtesy of Springer Nature, terms of use apply. Rights reserved

15
Vol.:(0123456789)
Scientic Reports | (2022) 12:6333 | https://doi.org/10.1038/s41598-022-09790-w
www.nature.com/scientificreports/
does not yield a decrease between the inequalities of the countries from the two groups or any signicant change
in their ranks (see Supplementary Fig.3).
e WYLL measure we present has some additional limitations. e rst comes from the assumption that
retirement age across European countries is 65. While it is most oen assumed as a standard between European
countries, there is actually some variation between individual member states30. Furthermore, we assume that the
mean age of people who have died in a given age group is the middle of the given range, e.g. for the age group
60–64 -
mean age =62.5
. It may well be a fact that a majority of the fatalities are concentrated in the upper part
of the age bracket. However, since we do not have data about the dierent causes of mortality, but rather an
aggregate total, we cannot be certain that this trend will hold true for all age groups and across dierent countries.
Google COVID‑19 community mobility reports. Bulgaria is below the EU average when it comes to use of
mobile devices in the 16–74 age group. Still, a majority of the population within that group (~64%) utilized
mobile devices to access the internet in 201931. However, it is possible that there might be a skew towards the
younger half of this age range of users who are supplying data.
Data availability
All datasets and associated code can be found at https:// github. com/ Mlad- en/ COV- BG. git.
Received: 30 April 2021; Accepted: 29 March 2022
References
1. Wang, C., Horby, P. W., Hayden, F. G. & Gao, G. F. A novel coronavirus outbreak of global health concern. Lancet 395(10223),
470–473 (2020).
2. Zhou, P. et al. A pneumonia outbreak associated with a new coronavirus of probable bat origin. Nature 579(7798), 270–273 (2020).
3. ...Huang, C. et al. Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China. Lancet 395(10223), 497–506
(2020).
4. Li, Y., Wang, X. & Nair, H. Global seasonality of human seasonal coronaviruses: A clue for postpandemic circulating season of
severe acute respiratory syndrome coronavirus 2?. J. Infect. Dis. 222(7), 1090–1097 (2020).
5. Smit, A. J. et al. Winter Is Coming: A Southern Hemisphere Perspective of the Environmental Drivers of SARS-CoV-2 and the
Potential Seasonality of COVID-19. Int. J. Environ. Res. Public Health 17(16), 5634 (2020).
6. Karlinsky, A. & Kobak, D. Tracking excess mortality across countries during the COVID-19 pandemic with the World Mortality
Dataset. eLife 10, e69336 (2021).
7. Bustos Sierra, N. et al. All-cause mortality supports the COVID-19 mortality in Belgium and comparison with major fatal events
of the last century. Arch. Public Health 78(1), 117 (2020).
8. Kontopantelis, E., Mamas, M. A., Deaneld, J., Asaria, M. & Doran, T. Excess mortality in England and Wales during the rst wave
of the COVID-19 pandemic. J. Epidemiol. Community Health jech-2020-214764. (2020).
9. Fouillet, A., Pontais, I. & Caserio-Schönemann, C. Excess all-cause mortality during the rst wave of the COVID-19 epidemic in
France, March to May 2020. Euro Surveill. 25(34), 2001485 (2020).
10. Morfeld, P. et al. COVID-19: Spatial resolution of excess mortality in Germany and Italy. J Infect S0163–4453(20), 30678 (2020).
11. Eurostat. Deaths by week, sex, 5‑year age group. https:// ec. europa. eu/ euros tat/ datab rowser/ view/ demo_r_ mwk_ 05/ defau lt/ table?
lang= en.
12. Eurostat. Deaths by week, sex, 5‑year age group and NUTS‑3 region. https:// ec. europa. eu/ euros tat/ datab rowser/ view/ demo_r_
mweek3/ defau lt/ table? lang= en.
13. Eurostat. Population on 1 January by age group and sex region. https:// ec. europa. eu/ euros tat/ datab rowser/ view/ demo_ pjang roup/
defau lt/ table? lang= en.
14. UNdata Data Ser vice. Population by age, sex and urban/rural residence. http:// data. un. org/ Data. aspx?d= POP&f= table Code% 3A22.
15. Bulgarian National Statistical Institute. Mortality and life expectancy by sex and place of residence. https:// www. nsi. bg/ en/ conte nt/
6643/ morta lity- and- life- expec tancy- sex- and- place- resid ence.
16. Czech Statistical Oce. Life Tables for the Czech Republic, Cohesion Regions, and Regions-2018–2019, https:// www. czso. cz/ csu/
czso/ life- tables- for- the- czech- repub lic- cohes ion- regio ns- and- regio ns- 2018- 2019
17. World Health Organization. Expectation of life at age X (Mortality and global health estimates. https:// apps. who. int/ gho/ data/ node.
imr. LIFE_ 00000 00035? lang= en.
18. Bulgarian Ministry of Health. https:// www. mh. gover nment. bg/ en/.
19. Statistical information of the Ministry of Health. Open Data Porta. https:// data. egov. bg/ covid- 19? secti on= 8& subse ction= 16&
item= 36.
20. Onemocnení Aktuálne od MZCR. COVID‑19: Overview of deaths according to reports from regional hygienic stations. ht t ps:// onemo
cneni- aktua lne. mzcr. cz/ api/ v2/ covid- 19
21. Our World In Data. Excess mortality during the Coronavirus pandemic (COVID‑19)https:// ourwo rldin data. org/ excess- morta lity-
covid# how- is- excess- morta lity- measu red.
22. Our World In Data. Total COVID‑19 tests per 1,000 people, Our World in Data. htt ps:// ourwo rldin data. org/ graph er/ full- list- cumul
ative- total- tests- per- thous and? time= 2020- 03- 01.. 2020- 12- 31
23. Preedy, V. & Watson, R. (Eds.) Handbook of Disease Burdens and Quality of Life Measures. Springer. (2010).
24. Martinez, R., Soliz, P., Caixeta, R. & Ordunez, P. Reection on modern methods: years of life lost due to premature mortality—A
versatile and comprehensive measure for monitoring non-communicable disease mortality. Int J Epidemiol. (2019).
25. Google COVID‑19 Community Mobility Reports. https:// www. google. com/ covid 19/ mobil ity/.
26. Oxford Blavatnik School of Government. Covid‑19 Government Response Tracker. https:// www. google. com/ covid 19/ mobil ity/.
27. Hale, T. et al. A global panel database of pandemic policies (Oxford COVID-19 Government Response Tracker). Nat Hum
Behavhttps:// doi. org/ 10. 1038/ s41562- 021- 01079 (2021).
28. Eurostat. Population change ‑ crude rates of total change, natural change and net migration plus adjustment. https:// ec. europa. eu/
euros tat/ datab rowser/ view/ tps00 019/ defau lt/ table? lang= en.
29. Eurostat. Population change ‑ Demographic balance and crude rates at national level. https:// ec. europa. eu/ euros tat/ datab rowser/
view/ demo_ gind/ defau lt/ table? lang= en.
30. Finnish Center for Pensions. Retirement Ages. https:// www. etk. / en/ work- and- pensi ons- abroad/ inter natio nal- compa risons/ retir
ement- ages/.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

16
Vol:.(1234567890)
Scientic Reports | (2022) 12:6333 | https://doi.org/10.1038/s41598-022-09790-w
www.nature.com/scientificreports/
31. Eurostat. Individuals using mobile devices to access the internet on the move. https:// ec. europa. eu/ euros tat/ datab rowser/ view/ tin00
083/ defau lt/ table? lang= en
32. Bhopal, S. S. & Bhopal, R. Sex dierential in COVID-19 mortality varies markedly by age. Lancet 396(10250), 532–533 (2020).
33. Aburto, J. M., Schöley, J., Zhang, L., Kashnitsky, I., Rahal, C., Missov, T. I., Mills, M. C., Dowd, J. B. & Kashyap, R. Recent Gains in
Life Expectancy Reversed by the COVID-19 Pandemic. medRxiv 2021.03.02.21252772. (2021).
34. Pifarré, I. et al. Years of life lost to COVID-19 in 81 countries. Sci. Rep. 11(1), 3504 (2021).
35. Oce for National Statistics. Comparisons of all-cause mortality between European countries and regions: 2020. https:// www.
ons. gov. uk/ peopl epopu latio nandc ommun ity/ birth sdeat hsand marri ages/ deaths/ artic les/ compa rison sofal lcaus emort ality betwe
eneur opean count riesa ndreg ions/ 2020. (2021).
36. Liang, L. L., Tseng, C. H., Ho, H. J. & Wu, C. Y. Covid-19 mortality is negatively associated with test number and government
eectiveness. Sci. Rep. 10(1), 12567 (2020).
37. Wang, Y., Liu, Y., Struthers, J. & Lian, M. Spatiotemporal characteristics of the COVID-19 epidemic in the United States. Clin.
Infect. Dis. 72(4), 643–651 (2021).
38. Gao, S. et al. Association of mobile phone location data indications of travel and stay-at-home mandates with COVID-19 infection
rates in the US. JAMA Netw. Open 3(9), e2020485 (2020).
39. C andido, D. S. et al. Brazil-UK Centre for Arbovirus Discovery, Diagnosis, Genomics and Epidemiology (CADDE) Genomic Net-
work, Ferguson NM, Costa SF, Proenca-Modena JL, Vasconcelos ATR, Bhatt S, Lemey P, Wu CH, Rambaut A, Loman NJ, Aguiar
RS, Pybus OG, Sabino EC, Faria NR. Evolution and epidemic spread of SARS-CoV-2 in Brazil. Science 369(6508), 1255–1260
(2020).
40. Islam, N. et al. Physical distancing interventions and incidence of coronavirus disease 2019: Natural experiment in 149 countries.
BMJ 370, m2743 (2020).
41. Goolsbee, A. & Syversonab, C. Fear, lockdown, and diversion: Comparing drivers of pandemic economic decline. J. Public Econ.
193, 104311 (2021).
42. Stokes, A. C., Lundberg, D. J., Elo, I. T., Hempstead, K., Bor, J. & Preston, S. H. Assessing the Impact of the Covid-19 Pandemic
on US Mortality: A County-Level Analysis. medRxiv 2020.08.31.20184036 (2020).
43. Eurostat. Cardiovascular Statistics. https:// ec. europa. eu/ euros tat/ stati stics- expla ined/ index. php/ Cardi ovasc ular_ disea ses_ stati
stics# Deaths_ from_ cardi ovasc ular_ disea ses.
44. Bulgarian Ministry of Economy. Economic map of Bulgaria.
45. Central Europe, Spared in the Spring, Suers as Virus Surges. New York Times, Oct. 14th (2020).
46. Bulgaria’s health system on brink of collapse from coronavirus crisis. Politico (2020).
47. It’s Like the Second World War. Covid-19 Is Tearing Into the Parts of Europe at Lack Doctors. Wall Street J. (2020).
48. Adovor, E., Czaika, M., Docquier, F. & Moullan, Y. Medical brain drain: How many, where and why?. J. Health Econ. 76, 102409
(2021).
Author contributions
A.R, M.M. collected data; A.R, and M.M. analyzed data and A.R., M.M., and G.K.M. wrote the manuscript.
Competing Interests
e authors declare no competing interests.
Additional information
Supplementary Information e online version contains supplementary material available at https:// doi. org/
10. 1038/ s41598- 022- 09790-w.
Correspondence and requests for materials should be addressed to G.K.M.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional aliations.
Open Access is article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the
Creative Commons licence, and indicate if changes were made. e images or other third party material in this
article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.
© e Author(s) 2022, corrected publication 2022
Content courtesy of Springer Nature, terms of use apply. Rights reserved

1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:
use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at
onlineservice@springernature.com