Introduction

There were a significant number of excess deaths in Poland in 2020 and still more were registerd in 2021. The phenomena is analyzed using statistical exploratory analysis (Tukey 1980) and charts in particular. We do not pretend to provide answers, we try to find interesting or strange patters and ask questions…

Document URL: https://bookdown.org/rudolf_von_ems/excess_deaths_in_poland_in_202021_an_exploratory_analysis/exd.html

The scale of the problem

Assessing change in the number of deaths as a difference between 2015–2019 average and the current value one can easily find that approx 80,000 more people died in Poland in connection with or due to COVID19 (which present an unprecedented increase by 20%). At the same time 29095 people died in connection with COVID19 in 2020, with 27069 deaths in September–December alone (17 weeks or weeks number 36–52). In the first 14 weeks of 2021 more 29309 were registerd. This gives 1711 and 2039 deaths per week respectively (in 2021 on the average there were 20% more deaths than in september-december 2020. BTW there were 1254559 new cases in weeks 1–14 of 2021, and 1248165 new cases in weeks 36–52 of 2020 (89,5 vs 73,4/week or approx 20% increase)

The data

We analyse two datasets: deaths (highly reliable) and covid deaths (somehow reliable).

Deaths statistics come from the Eurostat database (table demo_r_mweek3: Deaths by week, sex, 5-year age group and NUTS 3 region). The last reported week is 14th (starting Monday 2021-04-05). COVID19 deaths statistics come from the announcements of the Ministry of Health published at https://www.gov.pl/web/koronawirus/. Polish HM do not provide any database (sic!) with COVID related data and until January 2021, the messages were not even archived. We archived them on our own. Our data differs from the official ones because our database is a copy of the daily messages and the Ministry of Health data is corrected in the so-called in the meantime, often without any explanation or backtracking. However the difference of number of deaths, is insignificantly small (according to our base number of deaths classified as COVID19 victims as of April 30, 2021 is 67575 and according to the MZ 67501 messages https://www.gov.pl/web/koronawirus/wykaz-zarazen-koronawirusem-sars-cov-2; difference 74 deaths or 0.1%).

Population data comes from Bank Danych Lokalnych (provided Central Statistical Office or CSOPL; https://bdl.stat.gov.pl/BDL/dane/podgrup/temat.) GDP data comes from Bank Danych Makroekonomicznych (provided by CSOPL; https://stat.gov.pl/obszary-tematyczne/rachunki-narodowe/rachunki-regionalne/wstepne-szacunki-produktu-krajowego-brutto-w-przekrój-regionow-w-2019-roku,8,3.html)

NOTE1: There are 16 provinces in Poland (according to National Register of Territorial Divisions of the Country or TERYT) but 17 regions (EU NUTS classification) because the biggest Polish województwo Mazowieckie (TERYT) is divided into two macroregions in NUTS. Deaths statistics are published by GUS/Eurostat according to NUTS but COVID according to TERYT. To get the common denominator we combined PL91 and PL92 macroregions (NUTS) back to województwo Mazowieckie. More information on NUTS is here https://stat.gov.pl/en/regional-statistics/classification-of-territorial-units/classification-of-territorial-units-for-statistics-nuts/the-nuts-classification-in-poland/

NOTE2: As one can expect the provinces differs with area, population size, GDP/per capita, health services availability, etc. The biggest is Mazowieckie for example and it has the highest GDP per capita. The short summary of Polish provinces (for foreigners) is here:

name population doctors/per 10ths GDP (% of average)
Polska 38382576 23.7 100.0
dolnośląskie 2900163 22.1 109.7
kujawsko-pomorskie 2072373 25.4 81.2
lubelskie 2108270 25.2 67.8
lubuskie 1011592 20.4 82.3
łódzkie 2454779 28.5 93.1
małopolskie 3410901 23.9 92.1
mazowieckie 5423168 27.6 160.5
opolskie 982626 20.1 79.4
podkarpackie 2127164 22.1 70.6
podlaskie 1178353 24.9 71.7
pomorskie 2343928 21.8 97.2
śląskie 4517635 24.5 103.9
świętokrzyskie 1233961 24.0 72.3
warmińsko-mazurskie 1422737 21.6 69.0
wielkopolskie 3498733 15.1 108.0
zachodniopomorskie 1696193 27.4 83.2

BTW: province in Polish is prowincja (due to both are from Latin) but actually Polish administrative provice is called “województwo”, from “wodzić” – ie commanding (the armed troops in this context). This is an old term/custom from the 14th century, where Poland was divided into provinces (every province ruled by a “wojewoda” ie chief of that province). More can be found at Wikipedia (cf Administrative divisions of Poland)

Weekly number of deaths in Poland (2015–2021)

We use the same trick (polar coordinates) as applied by Nightingale (1858) in her famous book on Crimean War deaths.

Number of deaths in Poland by province

Before 2020 slow linear increase in (almost; łódzkie/lubelskie are exceptions probably due to migration) all provinces. Sharp increase in 2020.

Excess deaths by province

As mentioned earlier we measure excess deaths (or mortality) as the difference between 2015–2019 average and the number of deaths registered in relevant weeks in 2020 and 2021. This simple measure is commonly used, ie https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Excess_mortality_-_statistics (On the right covid deaths for easy comparison.)

Provinces with problems (PwP): świętokrzyskie, donośląskie, małopolskie, kujawsko-pomorskie, warmińsko-mazurskie.

There is a theory that is number of cases (and deaths) is related to (high) population density and reversely related to separation but obviously they did not know about that in Podkarpackie and Warmińsko-mazurskie…

Question#1: Is excess deaths phenomenon in 2020–21 in Poland caused by COVID or only somehow related to?

Number of deaths in Poland 2020/21 by age groups

We can observe the pattern by various age gropus in the following chart (note different Y-scales)

Overall number of deaths in Poland 2020/21

Weekly total deaths as percent of 2015–2019 province average. Green vertical line denotes week #47 which was particularly disastrius (22-28/11/2020, why?)

Augmented version of the above. Weekly Covid-related deaths as percent of province average for 2020W36–2021W14 vs weekly total deaths as percent of 2015–2019 province average. Green vertical line denotes week #47 (22-28/11/2020)

Question#2: why peaks of COVID/general deaths curves are shifted horizontally?

Question#3: Podkarpackie and Małopolskie recorded highest relative excess deaths in fatal late November weeks (over 250% of 2015-2019 average) yet both rank low in all-time-loses table (Podkarpackie #6, Małopolskie #15). Małopolskie even ranks last in the number of excess deaths table. HC failure or measurement artefact?

COVID/excess deaths ratio

Covid related deaths as % of excess deaths (values of this ratio are increasing significantly. Because better tests are applied? :-) )

Question#4: why rate of COVID-related deaths to total of excess deaths increases with no obvious relation to the stages/levels of COVID epidemics (as depicted below)? To be precise: the rate was well below 50% in 2nd wave, very high between 2nd–3rd waves (2021W7) and during 3rd wave…

Excess Deaths vs GDP/access to healthcare

Any correlation with GDP or health care availability? The measure of HC availibility is number of doctors per 10 ths population (lekarze pracujący wg podstawowego miejsca pracy na 10 tys. ludności doctors employed by the primary workplace per 10 thous. population; cf: https://bdl.stat.gov.pl/BDL/metadane/cechy/3173)

Doctors vs ED: coefficient = 14.874 (probab 0,053 barely significant and R² = 24%) GDP vs ED: coefficient -2,39 (prob 0,23 and R² = 10%)

One additional doctor/10 ths or 100 additional doctors per 1 mln population causes extra 14 deaths (per 1 mln population), so at least statistically it is not worth to invest in HC 😜😜

Technical remarks

This document is an abridged version of a report prepared as R-markdown document (Xie 2021). The source file and the data are available at author’s github repo here: https://github.com/hrpunio/Papers/tree/main/PSW_2021

Bibliography

Nightingale Florence, Mortality of the British Army, at Home and Abroad, and During the Russian War as Compared with the Civil Population in England, Harrison&Sons, London 1858, (https://archive.org/details/mortalityofbriti00lond)

Tukey John W., We Need Both Exploratory and Confirmatory The American Statistician Vol. 34, No. 1 (Feb., 1980), pp. 23-25 (3 pages)

Xie Yihui, Allaire J.J, Grolemund Garrett, R Markdown: The Definitive Guide https://bookdown.org/yihui/rmarkdown/ 2021-04-09