Investigators examined municipality-level data in Austria to assess factors associated with COVID-19 mortality between January 2020 and June 2023. The dataset included 18,430 reported deaths across the national population. Using univariable zero-inflated negative binomial models, the research team observed that municipalities with larger shares of elderly residents aged 75-84 and ≥85 years experienced higher mortality risks, with pooled relative risks (RR) of 1.18 and 1.29 respectively. Socioeconomic characteristics intertwined with mortality outcomes: areas with more residents holding only compulsory education had elevated mortality (RR 1.15), whereas higher proportions of commuters and employed individuals aged 15-64 correlated with reduced mortality risk (RRs of 0.79 and 0.89). Interestingly, homeownership negatively associated with mortality (RR 0.91), suggesting potential protective effects related to housing stability. In multivariable models adjusting for confounders, the proportion of residents aged 85 or older remained an independent risk factor (RR 1.30), while homeownership continued to exhibit a modest protective association (RR 0.92).
The study team analyzed annual data separately and then pooled the results via random-effects meta-analysis to explore whether associations shifted over the pandemic course. Mortality predictors demonstrated relative stability across calendar years from 2020 through mid-2023. This finding indicates consistent demographic and socioeconomic vulnerabilities driving COVID-19 mortality throughout successive pandemic phases despite evolving viral variants and public health measures. However, predictors of SARS-CoV-2 infection rates showed more variability over time, reflecting the dynamic nature of viral spread and testing practices. Such temporal differences highlight complexity in understanding infection dissemination versus mortality drivers at the municipality level.
Researchers incorporated annually averaged environmental factors, including air pollution exposures and UV radiation metrics, into their models. In univariable analyses, certain climatic variables correlated with mortality: precipitation exhibited a small inverse association (RR 0.94) and the average time with mean temperature between 5-15°C also trended negative (RR 0.95). Despite these signals, environmental variables did not retain significance in multivariate frameworks, suggesting limited independent influence relative to population composition and socioeconomic indicators. The models did not provide detailed dosing or threshold data for these exposures. The interaction between ecological environmental factors and COVID-19 outcomes merits cautious interpretation because of potential residual confounding and the complexity of exposure assessments.
The dataset encompassed 5,963,882 documented SARS-CoV-2 infections. A negative binomial model identified demographic and employment-related predictors of detected infections. Higher employment shares among individuals aged 15 to 64 and increased part-time employment rates both associated with slightly higher infection counts (RRs of 1.01, with upper ranges suggesting minimal effect size). Contrastingly, higher district-level nursing care allowance recipiency linked to lower detected infections (RR 0.97). Other variables, including overnight stays as a proxy for tourism, the proportion of elderly residents, income levels, educational attainment, and air pollution showed multiple conflicting signals with overall non-significant pooled associations. These fluctuating patterns underscore the challenge in uncovering stable predictors for infections due to variations in testing access, behavioral changes, and reporting over the pandemic.
To strengthen validity, the study employed two alternative approaches for selecting smaller sets of predictors, acting as sensitivity analyses. The outcomes from these refined models closely mirrored the main analysis results, reinforcing confidence in the identified associations between demographic, socioeconomic factors, and COVID-19 mortality and infection trends. This methodological triangulation suggests that the core findings are not artifacts of variable selection but reflect underlying community-level characteristics influencing pandemic impacts.
This ecological analysis utilized aggregated municipal data with requisite approvals from the Austrian Agency for Health and Food Safety and Statistics Austria. Patient privacy was protected by restricting data access and anonymizing identifiers. The research team secured ethical clearance from the Medical University of Graz ethics committee (approval no. 33-144 ex 20/21). All relevant patient and participant consents were obtained as mandated. The authors disclosed no competing interests. Their adherence to ethical guidelines ensured responsible conduct and reporting of population-based COVID-19 research in Austria.
The study did not provide detailed methodological aspects of dosing or individual-level exposures, and the ecological approach limits causal inference. Future work might explore mechanisms underlying socioeconomic vulnerabilities that persist as risk factors across pandemic waves.