Place-based structural disadvantage and mortality inequality across 379 Polish counties, 2010-2025: a spatial Mundlak analysis
- 1National Institute of Public Health NIH National Research Institute
Abstract
Place-based mortality inequalities may persist even as short-term associations with local conditions change. We examined age-sex-standardised all-cause mortality in 379 harmonised Polish counties (2010-2025), asking whether durable between-county differences or lagged within-county deviations better characterise mortality geography. Spatial Mundlak generalised additive models allowed non-linear associations and county-specific spatial modification; a pre-COVID model (2010-2019) was compared with overlapping models through 2022 and 2025. In the pre-COVID model, median county average partial slopes per standard deviation were -26.1 deaths per 100,000 for higher education, +29.4 for primary/lower-secondary education and +30.2 for PM2.5. Directions of these between-county gradients persisted across analytical windows. By contrast, within-county associations were less stable and sometimes reversed sign in pandemic-inclusive models. County-level uncertainty indicated stronger and more coherent pre-COVID spatial patterns for education, PM2.5 and population density than for several health-care and migration terms. Strict leave-one-year-out validation revealed weak temporal transportability, particularly for the 2010-2022 model, limiting predictive interpretation. Outcome-defined comparisons further showed that educational composition remained an important separator of high- and low-mortality counties after COVID-19, while the fitted PM2.5 contribution contrast diminished. These findings distinguish durable configurations of place-based disadvantage from period-dependent ecological co-movement. They support geographically sensitive, multi-domain approaches to reducing mortality inequalities, while not establishing individual-level causal effects.
Related
- Structural and Social Disadvantage and Post-Migration HIV Vulnerability among African, Caribbean, and Black Communities in Ontario
- Multimodal urban park service accessibility and spatial distributional inequality through a multiscale assessment
- Disability Inequality in Short- and Long-Term Mortality After Multidrug-Resistant Tuberculosis
- Exploring the spatial nexus between air quality, climate change, and socioeconomic determinants of maternal mortality across the Asia–Pacific region
- Mortality dynamics of West Nile virus and Lyme disease in the United States : Temporal, spatial, and socioeconomic dimensions (2003–2021)
BibTeX
@misc{madej2026placebased,
title = {{Place-based structural disadvantage and mortality inequality across 379 Polish counties, 2010-2025: a spatial Mundlak analysis}},
author = {T. Madej and B. Wojtyniak and D. Rabczenko and A. Smaga and S. Boguslawski},
howpublished = {medRxiv},
year = {2026},
month = nov,
doi = {10.64898/2026.11.02.26367044},
url = {https://www.medrxiv.org/content/10.64898/2026.11.02.26367044v1},
note = {CC BY-NC-ND},
}
RIS
TY - UNPB
TI - Place-based structural disadvantage and mortality inequality across 379 Polish counties, 2010-2025: a spatial Mundlak analysis
AU - T. Madej
AU - B. Wojtyniak
AU - D. Rabczenko
AU - A. Smaga
AU - S. Boguslawski
PY - 2026
DA - 2026/11/02
PB - medRxiv
DO - 10.64898/2026.11.02.26367044
UR - https://www.medrxiv.org/content/10.64898/2026.11.02.26367044v1
AB - Place-based mortality inequalities may persist even as short-term associations with local conditions change. We examined age-sex-standardised all-cause mortality in 379 harmonised Polish counties (2010-2025), asking whether durable between-county differences or lagged within-county deviations better characterise mortality geography. Spatial Mundlak generalised additive models allowed non-linear associations and county-specific spatial modification; a pre-COVID model (2010-2019) was compared with overlapping models through 2022 and 2025. In the pre-COVID model, median county average partial slopes per standard deviation were -26.1 deaths per 100,000 for higher education, +29.4 for primary/lower-secondary education and +30.2 for PM2.5. Directions of these between-county gradients persisted across analytical windows. By contrast, within-county associations were less stable and sometimes reversed sign in pandemic-inclusive models. County-level uncertainty indicated stronger and more coherent pre-COVID spatial patterns for education, PM2.5 and population density than for several health-care and migration terms. Strict leave-one-year-out validation revealed weak temporal transportability, particularly for the 2010-2022 model, limiting predictive interpretation. Outcome-defined comparisons further showed that educational composition remained an important separator of high- and low-mortality counties after COVID-19, while the fitted PM2.5 contribution contrast diminished. These findings distinguish durable configurations of place-based disadvantage from period-dependent ecological co-movement. They support geographically sensitive, multi-domain approaches to reducing mortality inequalities, while not establishing individual-level causal effects.
ER -