Piotr Oskar Czechowski

Poland has one of the worst air quality in the European Union, particularly regarding concentrations of particulate matter (PM). This study aimed to evaluate short-term effects of air pollution and weather conditions on all-cause mortality and pneumonia-related hospitalizations in three Polish agglomerations. We investigated data from 2011-2018 on a number of health outcomes, concentrations of PM2.5, PM10, nitrogen dioxide (NO2), ozone (O3), and selected meteorological parameters. To examine the impact of air pollutants and weather conditions on mortality and pneumonia burden, we identified optimal general regression family models for each agglomeration. GRM uses the concept of general linear models (GLM), enabling to capture the non-linearity of the impact of cause and effect relationships at the stage of nonlinear link function as well as through the interactions and lags of independent factors. The identification of interactions and lags up to 3 days is particularly important element of the presented research. The final models explained <25,3% of variability in all-cause mortality. The models with interactions, O3 concentration in Warsaw, NO2, O3, and PM2.5 concentrations in Cracow and PM10 and O3 concentrations in the Tricity explained >13% of variability in the number of deaths. Up to 48,6% of daily variability in the number of pneumonia-related hospitalizations was explained by the combination of both factors i.e. air quality and meteorological parameters. The impact of NO2 levels on pneumonia burden was pronounced in all agglomerations. We showed that air pollution profile and its interactions with weather conditions exert a short-term effect on all-cause mortality and pneumonia-related hospitalizations. Our findings may be relevant for prioritizing strategies to improve air quality.

Keywords: air pollution; ambient particulate matter; nitrogen dioxide; ozone; weather; modelling; mortality; morbidity; GRM; GLM

Audience Take Away:

• Audience will learn synthetically the situation of Poland and partly Europe in the area of selected air pollution.
• Audience will be able to see an outline of the stochastic data processing methodology (stochastic, exploration and author’s), including diagnostics of measurement data based on air pollution automatic monitoring system.
• Audience will have the opportunity to learn about the factors (air pollution and meteorology) with lags and interactions affecting selected diseases and deaths (among others pneumonia, asthma and COPD) over a long period of time (2011-2018) identified by GLM family statistical models.
• The presentation is primarily aimed at expanding the knowledge about the air pollution situation in Europe and in Poland and about statistical methods and models allowing to identify in detail the causes of selected diseases. The cognitive aspect of statistical methodology, relatively rare in this scale of research, its advantages and weaknesses also seems to be important.
• The key benefit from the point of view of conducting own research and job seems to be getting to know the classic and new statistical methodology of measuring data diagnostics, learning about solutions that ensure high quality of both data and the results of model works. An interesting aspect is also the approximation of the statistical models themselves, their assumptions and problems that were solved in the identification and estimation process. Important are the conclusions in the form of identified factors influencing selected disease and long-term deaths, which allow for comparison with other regions of the world. The presented methodology and results allow to include this component in the regional policy of sustainable development of urban and industrial agglomerations.