Association between temperature and natural mortality in Belgium: Effect modification by individual characteristics and residential environment

Claire Demoury; Katrien De Troeyer; Finaba Berete; Raf Aerts; Bert Van Schaeybroeck; Johan Van der Heyden; Eva M De Clercq

doi:10.1016/j.scitotenv.2022.158336

Association between temperature and natural mortality in Belgium: Effect modification by individual characteristics and residential environment

Sci Total Environ. 2022 Dec 10;851(Pt 2):158336. doi: 10.1016/j.scitotenv.2022.158336. Epub 2022 Aug 26.

Authors

Claire Demoury¹, Katrien De Troeyer², Finaba Berete³, Raf Aerts⁴, Bert Van Schaeybroeck⁵, Johan Van der Heyden³, Eva M De Clercq⁶

Affiliations

¹ Risk and Health Impact Assessment, Sciensano, Brussels, Belgium. Electronic address: claire.demoury@sciensano.be.
² Family Medicine and Population Health, University of Antwerp, Belgium.
³ Lifestyle and Chronic Diseases, Sciensano, Brussels, Belgium.
⁴ Risk and Health Impact Assessment, Sciensano, Brussels, Belgium; Division Ecology, Evolution and Biodiversity Conservation, KU Leuven, Leuven, Belgium; Center for Environmental Sciences, University of Hasselt, Hasselt, Belgium.
⁵ Department of Meteorological Research and Development, Royal Meteorological Institute of Belgium, Brussels, Belgium.
⁶ Risk and Health Impact Assessment, Sciensano, Brussels, Belgium.

PMID: 36037893
DOI: 10.1016/j.scitotenv.2022.158336

Abstract

Background: There is strong evidence of mortality being associated to extreme temperatures but the extent to which individual or residential factors modulate this temperature vulnerability is less clear.

Methods: We conducted a multi-city study with a time-stratified case-crossover design and used conditional logistic regression to examine the association between extreme temperatures and overall natural and cause-specific mortality. City-specific estimates were pooled using a random-effect meta-analysis to describe the global association. Cold and heat effects were assessed by comparing the mortality risks corresponding to the 2.5^th and 97.5^th percentiles of the daily temperature, respectively, with the minimum mortality temperature. For cold, we cumulated the risk over lags of 0 to 28 days before death and 0 to 7 days for heat. We carried out stratified analyses and assessed effect modification by individual characteristics, preexisting chronic health conditions and residential environment (population density, built-up area and air pollutants: PM_2.5, NO₂, O₃ and black carbon) to identify more vulnerable population subgroups.

Results: Based on 307,859 deaths from natural causes, we found significant cold effect (OR = 1.42, 95%CI: 1.30-1.57) and heat effect (OR = 1.17, 95%CI: 1.12-1.21) for overall natural mortality and for respiratory causes in particular. There were significant effects modifications for some health conditions: people with asthma were at higher risk for cold, and people with psychoses for heat. In addition, people with long or frequent hospital admissions in the year preceding death were at lower risk. Despite large uncertainties, there was suggestion of effect modification by air pollutants: the effect of heat was higher on more polluted days of O₃ and black carbon, and a higher cold effect was observed on more polluted days of PM_2.5 and NO₂ while for O₃, the effect was lower.

Conclusions: These findings allow for targeted planning of public-health measures aiming to prevent the effects of extreme temperatures.

Keywords: Air pollution; Effect modification; Mortality; Preexisting conditions; Temperature; Vulnerability.

Publication types

Multicenter Study

MeSH terms

Air Pollutants* / analysis
Air Pollution* / analysis
Belgium / epidemiology
Carbon
Cross-Over Studies
Hot Temperature
Humans
Mortality
Network Meta-Analysis
Nitrogen Dioxide / analysis
Particulate Matter / analysis
Temperature
Time Factors

Substances

Air Pollutants
Carbon
Nitrogen Dioxide
Particulate Matter