Polish Journal of Public Health

Original Article, Pol J Public Health, Vol. 132 (2022): 67-72

Mieczysław Szyszkowicz¹, Eugeniusz Porada²

¹ Environmental Health Science and Research Bureau, Health Canada, Ottawa, Canada
² An independent consultant

 10.2478/pjph-2022-0014

© 2022  Medical University of Lublin. This is an open access article distributed under the Creative Commons Attribution-NonComercial-No Derivs licence (http://creativecommons.org/licenses/by-nc-nd/3.0/)

Abstract

Introduction. Estimating the impact of short-term exposure on health outcomes needs knowledge of both the profile and magnitude of the relative risks. This motivates constructions of practical and reliable concentration-response functions (C-RFs).

Aim. To define a practical method of finding concentration-response parametric function whose adjustable parameters can be tuned by data-driven well established routines.

Material and methods. Mortality data for the period from 1987 to 2015 (10,592 consecutive days) in Montreal, Canada, are used for illustrative purposes. Exposure to ambient ozone measured by its concentration levels is considered health risk. Concentration-response function is built using statistical modelling, conditional Poisson regression, natural spline technique, and a rudimentary hierarchical data clustering. The case-crossover design is applied to fit the model of C-RF to the mortality data consisting of daily counts of non-accidental deaths.

Results. Log-linear models of the concentration-response functions were computed for the concentrations and cofactors data lagged by 0 to 7 days; the results were statistically significant within this range of lags. The effectiveness of fitting was confirmed by reliable statistical tests. Digital routines were created to perform all computational tasks; software codes (written for R soft­ware platform) are included. The C-RF specifying the current responses to the cumulative exposure in several previous days can be obtained from the responses to lagged exposures.

Conclusions. The proposed method of concentration-response function estimation appears practical and effective in producing reliable results. The constructed function is a parametric and monotonic non-decreasing.

Keywords: ambient air pollution; concentration-response function; log-linear modelling.