Deaths during the first year of the COVID-19 pandemic: insights from regional patterns in Germany and Poland

The COVID-19 pandemic imposed major economic, health and social burdens worldwide, with considerable variation between and within countries. Germany and Poland sit at opposite ends of excess deaths in 2020 among 29 high-income countries. Using EUROSTAT data, Poland’s excess deaths rate in 2020 (vs. 2016–2019 average) was 194 per 100,000 inhabitants, compared to Germany’s 62. Despite Germany’s larger population (2.2 times Poland’s), officially recorded COVID-19 deaths in 2020 were 50,447 in Germany and 28,479 in Poland. The study investigates whether regional spatial patterns of mortality can shed light on reasons behind these stark country-level differences, considering the two countries’ differing pre-pandemic demographics, healthcare and public health systems, and substantial within-country regional heterogeneity in risk factors.

Prior research highlights the spatial nature of infectious disease spread and COVID-19 outcomes, implicating factors such as age structure, comorbidities (cardiovascular, pulmonary, cancer), population density, overcrowded housing, transportation hubs, workplace exposure, environmental conditions (e.g., air pollution), healthcare accessibility, and non-pharmaceutical interventions. European regional studies showed early importance of transportation hubs, variable links between hospitalizations and excess mortality, and associations of intensive care capacity with mortality. For Italy, excess deaths were highly uneven geographically. Early evidence from Poland found higher incidence and mortality in Silesia (partly due to testing), higher risks for older men and those with chronic diseases, and associations with air pollution. In Germany, mortality increased in certain regions during 2020, life expectancy drops were larger in the East among older men, and case/death counts were related to nursing home personnel numbers and age structure. Excess mortality in Germany was low overall for 2020 due to summer troughs offsetting winter peaks, with higher incidences during the second wave in denser, lower-income, higher-employment, more industrial regions. Despite numerous spatial studies, comprehensive spatial analyses linking mortality to combined demographic, economic, and healthcare factors for Germany and/or Poland had been lacking.

Study design: Comparative spatial analysis of mortality across regional units in Germany (401 Kreise, NUTS-3) and Poland (380 powiaty, LAU-1) during 2020. Outcomes (per 1,000 population, using 2019 population denominators): (1) Officially recorded COVID-19 deaths (aggregated daily county-level counts for 2020), (2) Total excess deaths = total deaths in 2020 minus the average annual deaths in 2015–2019, and (3) Difference = excess deaths – official COVID-19 deaths. Conceptualization: Deaths are classified as direct (immediate consequence of SARS-CoV-2 infection, whether recorded as COVID-19 or not) or indirect. Indirect consequences are split into type 1 (premature deaths arising from strain on healthcare due to local spread, e.g., bed shortages, limited procedures, staff shortages) and type 2 (deaths stemming from broader pandemic effects not necessarily tracking local contagion, e.g., reduced access to primary care, isolation, reduced support, mental health effects). Spatial analysis: Initial OLS regressions were estimated, with residual spatial correlation tested via Moran’s I. Then spatial autoregressive models were fit: spatial error models (spatially lagged errors; ML estimator) and spatial lag models (spatially lagged dependent variable; ML estimator). Multiple spatial weight matrices (SWMs) were examined (first/second-order contiguity, fixed distance, and inverse-distance with various truncation thresholds). Preferred SWM: row-normalized inverse-distance matrix truncated at 70 km (shortest threshold covering all counties in both countries). Global Moran’s I tests for the three outcomes indicated similar significance across tested matrices; detailed results in supplementary materials. Covariates: Pre-pandemic regional characteristics included demographics (age structure, population density), economic structure (e.g., employment in agriculture), and healthcare infrastructure (hospital beds per 1,000 inhabitants; 2016 for beds). Data sources: Official geodata and statistics from German Federal/state offices and Robert Koch Institute; Polish Central Statistical Office (GUS), Ministry of Health, regional/local Sanitary Inspectorates, and curated datasets, as detailed in data statement. Analytical focus: Identify and compare spatial dependence in mortality measures within each country, and assess associations between mortality measures and regional characteristics in multivariate SAR frameworks.

- Country-level contrasts (2020): • Excess deaths per 100,000 (EUROSTAT): Poland 194 vs Germany 62. • Official COVID-19 deaths: Germany 50,447 vs Poland 28,479. - Regional averages (per 1,000 population): • COVID-19 deaths: Germany mean 0.63; Poland mean 0.86 (greater regional variance in Germany). • Excess mortality: Germany mean 0.64; Poland mean 2.19. • Difference (excess – COVID-19): ~0 in Germany; 1.33 in Poland. - Spatial patterns: • Germany shows strong and significant spatial autocorrelation for all three outcomes (official COVID-19 deaths, excess deaths, and difference), with notable clusters (e.g., Sachsen, Bayern, Brandenburg). • Poland shows significant spatial autocorrelation only for official COVID-19 deaths; excess deaths and the difference exhibit little to no spatial dependence, despite higher overall excess mortality. - Interpretation of spatial results: • Germany’s excess deaths (both recorded as COVID-19 and not) are consistent with direct COVID-19 impacts or indirect type 1 effects tracking local contagion. • Poland’s lack of spatial dependence in excess deaths suggests a large share of indirect type 2 fatalities (broader pandemic-related consequences not spatially aligned with virus spread). - Associations with regional characteristics: • Germany: COVID-19 deaths positively associated with share aged 85+; excess deaths and the difference negatively associated with hospital beds per 1,000; the difference negatively associated with population density and positively with agricultural employment. • Poland: COVID-19 deaths higher in less densely populated, more agricultural regions; associated with share aged 70–84 and with greater hospital bed availability (possibly due to more testing/recording or fewer unrecorded home deaths). Few consistent associations for excess deaths or the difference in Poland. - Overall: Pre-pandemic demographic and healthcare variables do not explain Poland’s non-spatial excess mortality pattern; findings point to policy and healthcare system factors.

The study demonstrates that spatial patterns of mortality during 2020 reflect different underlying dynamics in Germany versus Poland. In Germany, strong spatial autocorrelation across COVID-19 deaths, excess deaths, and their difference suggests that both direct COVID-19 fatalities and indirect type 1 effects (e.g., healthcare strain) predominated, consistent with clusters tracking contagion. Negative associations of excess deaths with hospital bed availability underscore the mitigating role of healthcare capacity. In Poland, excess mortality was much higher but did not follow expected spatial patterns for pandemic-driven deaths; instead, the lack of spatial dependence for excess deaths and the difference indicates substantial indirect type 2 effects, likely tied to broader disruptions: reduced healthcare seeking due to fear of infection, insufficient care for non-COVID conditions, excessive or misdirected limitations on admissions/procedures, mobility restrictions, reduced social contact/support, and mental health deterioration. These suggest significant shortcomings in healthcare and public health responses in Poland. The results emphasize that spatial analyses can reveal whether excess mortality aligns with infection-driven dynamics or broader systemic and behavioral impacts, informing the evaluation of policy effectiveness.

Spatial analysis of regional mortality in 2020 in Germany and Poland reveals starkly different patterns. Germany exhibits strong spatial clustering for COVID-19 deaths and excess mortality, consistent with infection-driven and type 1 indirect effects mediated by healthcare capacity. Poland, despite far higher excess mortality, shows little spatial dependence for excess deaths beyond official COVID-19 counts, pointing to a predominance of type 2 indirect consequences not tied to local contagion. These findings indicate that beyond demographic and healthcare infrastructure, policy reactions and public health system performance critically shape mortality outcomes. Further research should probe specific policy and healthcare mechanisms underlying Poland’s elevated non-spatial excess mortality to guide effective responses in future health crises.

- Potential omitted variable bias: County-level pre-pandemic indicators may correlate with unobserved regional factors, biasing coefficient estimates. - Excess deaths definition: Uses simple historical average (2015–2019) rather than age- or composition-adjusted or model-based counterfactuals; while common and assumption-light for spatial dependency, alternative models may yield different baselines. - Policy heterogeneity not fully captured: Temporal and regional differences in testing intensity, tracing, NPIs, hospital bed allocation, and access restrictions in Germany and Poland are not explicitly modeled. - Data nuances: Continuous updates to official COVID-19 death counts and differences between cause-of-death registries vs. broad COVID-19 death definitions may introduce measurement discrepancies. - Exploratory nature: The study focuses on spatial dependence and associations, not causal identification; vaccine rollout (early 2021) does not affect 2020 results.