The COVID-19 pandemic caused substantial variation in mortality rates across regions and countries. This study investigates the spatial patterns of mortality during the first year of the pandemic in Germany and Poland, two neighboring countries with vastly different excess death rates in 2020. Poland's excess death rate was 194 per 100,000 inhabitants, compared to 62 in Germany. While Germany had 1.7 times more COVID-19 deaths than Poland (50,447 vs 28,479), the overall excess mortality was far greater in Poland. This disparity motivates a comparative analysis to understand the underlying causes. The pre-pandemic conditions in both countries, including socio-demographic structures, healthcare systems, and public health infrastructure, varied significantly, both at the national and regional levels. This regional heterogeneity, combined with the dramatically different aggregate mortality outcomes, presents a unique opportunity to investigate the regional dimension of the pandemic's impact and gain insights into the reasons behind the observed differences. The study explores the spatial nature of mortality in both countries using county-level data, focusing on the spatial patterns and their relationship with pre-pandemic characteristics. Three measures of death are analyzed: officially recorded COVID-19 deaths, total excess deaths, and the difference between the two. These are linked to regional characteristics like population, healthcare, and economic conditions using multivariate spatial autoregressive models. The goal is to determine if spatial patterns of mortality offer clues to explain the significant differences in the pandemic's consequences between Germany and Poland.

Existing research highlights the spatial nature of COVID-19 risks, emphasizing the roles of factors like population age structure, pre-existing health conditions (cardiovascular and pulmonary diseases, weakened immune systems, cancer), and access to healthcare infrastructure. Studies in various European countries show the impact of factors such as regional spread of the virus, accessibility to healthcare, effectiveness of non-pharmaceutical interventions, and socioeconomic factors. In Poland, early findings indicated higher incidence and mortality in the Silesia region, partly due to mass testing of miners. Air pollution also correlated with higher incidence and deaths in Poland. In Germany, increased mortality was observed in several regions during the first wave of 2020, with life expectancy decreases particularly impacting older men in the east. Regional studies emphasized the uneven distribution of excess deaths and the importance of factors such as healthcare infrastructure, population density, and economic conditions. However, a comprehensive analysis linking these factors in a detailed spatial analysis of mortality for both Germany and Poland was lacking before this study.

The study uses county-level data (Kreise in Germany and powiaty in Poland) to analyze the spatial correlation of deaths during the first year of the pandemic. Three outcome variables are examined: officially recorded COVID-19 deaths, total excess deaths (difference between 2020 deaths and the 2015-2019 average), and the difference between excess deaths and officially recorded COVID-19 deaths. The researchers employ spatial autoregressive models to account for spatial correlation in both the dependent variables and the residuals. Different spatial weight matrices (SWM) are used, considering nearest neighbors and inverse distance weights, to test for spatial correlation. The preferred model uses a row-normalized inverse distance weight matrix truncated at a 70km threshold. The significance of global Moran's I test is assessed to determine spatial dependence. The relationship between deaths and the pandemic is categorized into direct (immediate consequences of SARS-CoV-2 infection) and indirect (premature deaths not directly resulting from infection). Indirect consequences are further divided into type 1 (deaths resulting from the spread of COVID-19, such as hospital bed shortages) and type 2 (deaths that would not have occurred without the pandemic but are not directly related to the virus spread, such as those due to reduced access to healthcare or mental health issues). Pre-pandemic regional characteristics, including population density, age structure, employment in agriculture, and hospital beds per 1000 inhabitants, are used as explanatory variables in the multivariate spatial autoregressive models. OLS and SAR models are estimated, and diagnostic tests for spatial correlation are performed to evaluate the models' validity and interpret the results.

The study finds that excess mortality was considerably higher in Poland than in Germany. The spatial distribution of officially recorded COVID-19 deaths shows distinct clusters in southeastern Germany and central-eastern Poland. However, a strong spatial correlation for excess mortality is only observed in Germany, not in Poland. In Germany, excess deaths are spatially correlated with COVID-19 deaths, suggesting that a significant proportion were either direct or indirect consequences (type 1) of the pandemic. In contrast, the lack of spatial correlation in Poland for excess deaths above the official COVID-19 count indicates that these deaths were not primarily due to the direct or type 1 indirect effects of the pandemic. Moran scatterplots support these findings, showing higher spatial autocorrelation in Germany than in Poland for all three mortality measures. In Poland, spatial autocorrelation was only statistically significant for officially reported COVID-19 deaths, not for excess mortality overall or the difference between excess mortality and COVID-19 deaths. Multivariate spatial autoregressive models confirm the spatial patterns of mortality. In Germany, all three measures show significant spatial correlation, while in Poland, only COVID-19 deaths exhibit this pattern. The analysis of pre-pandemic regional characteristics shows some expected correlations with mortality in Germany (e.g., higher COVID-19 deaths associated with a larger proportion of the oldest age group), but few consistent relationships in Poland. This suggests that factors beyond pre-existing regional conditions played a more dominant role in shaping mortality patterns in Poland.

The contrasting spatial patterns of excess mortality in Germany and Poland highlight the importance of considering not only direct pandemic impacts but also indirect effects, particularly type 2 indirect effects, in understanding the overall consequences. The lack of spatial correlation in excess deaths in Poland, despite significantly higher overall numbers, suggests that policy failures and other factors unrelated to the direct spread of the virus contributed heavily to the excess mortality. This might be due to reduced access to healthcare, fear of seeking treatment, and mental health issues resulting from isolation and lockdowns. While the study cannot definitively distinguish between these hypotheses, the data strongly suggest the existence of significant policy deficits in Poland's healthcare and public health response to the pandemic. The findings in Germany support the notion that a substantial proportion of excess deaths were directly or indirectly linked to the pandemic's spread and its direct and type 1 indirect consequences.

This study demonstrates that spatial patterns of mortality can reveal important insights into the factors driving differences in pandemic outcomes across countries. The contrasting spatial patterns observed in Germany and Poland highlight the critical role of policy responses and healthcare systems in mitigating pandemic mortality. The significantly higher excess mortality in Poland, without the expected spatial correlation, indicates potential policy failures and the need for further investigation into the indirect consequences of type 2. Future research should explore these policy aspects in more depth and examine the specific mechanisms behind the observed differences in mortality patterns. This study emphasizes the need for robust healthcare systems and appropriate policy responses to effectively manage future global health crises.

The study acknowledges several limitations. The analysis is largely exploratory, and the relationships between mortality and regional covariates should be interpreted cautiously, as unobserved regional characteristics might bias the estimates. The definition of excess deaths, while common, relies on simple historical averages and might not capture all complexities. Local responses to the pandemic and variations in testing, tracing, and lockdown policies are not directly accounted for in the analysis, but are acknowledged as potential confounding factors influencing mortality patterns. Further research is warranted to address these limitations and provide a more comprehensive understanding of the observed patterns.