System Maintenance Notice

Please note that maintenance work is scheduled between 4 AM and 5 AM - CET. During this time, you may experience reduced performance. Thanks for your patience.

Publications

Journal Article Ppublish

PubMed: 40499277 DOI: 10.1016/j.ebiom.2025.105788

Exploring the relationship between medications and heat-related community deaths during the 2021 heat dome: a hybrid approach using machine learning.

EBioMedicine 117:105788

Authors: Jeremie Boudreault, Kathleen E McLean, Sarah B Henderson

Affiliations

  • Environmental Health Services, British Columbia Centre for Disease Control, 655 West 12th Ave, Vancouver, BC, V5Z 4R4, Canada; Centre Eau Terre Environnement, Institut national de la recherche scientifique, 490 de la Couronne, Québec, QC, G1K 9A9, Canada.
  • Environmental Health Services, British Columbia Centre for Disease Control, 655 West 12th Ave, Vancouver, BC, V5Z 4R4, Canada.

Abstract

BACKGROUND

Extreme heat events (EHEs) are a growing threat to health worldwide. To date, only a limited number of studies have evaluated medications as risk or protective factors for mortality during EHEs.

METHODS

We explored the relationship between dispensed pharmaceuticals and heat-related community deaths using linked administrative health data and both logistic regression (LR) and machine learning (ML) models. We conducted a case-control study during the 2021 EHE in British Columbia, Canada, including 504 community deaths from heat exposure as cases and 2520 similar controls who survived the EHE. We used medications dispensed 30, 60 and 90 days prior to death (or 30, 60 and 90 days before the end of the EHE for controls) as predictors, grouped by Anatomical Therapeutic Chemical (ATC) classification at level 2 for LR (28 classes) and level 4 for ML (270 subclasses). Models were adjusted for multiple covariates, including common chronic diseases.

FINDINGS

Results from LR showed increased odds of mortality associated with dispensations of antiepileptics, anti-Parkinson drugs, psycholeptics, diuretics, drugs for diabetes, beta blocking agents, analgesics, urologicals and drugs for treatment of bone diseases. We observed a protective association with dispensations of calcium channel blockers and ophthalmologicals. Results varied by sex, age, and other covariates. The ML model highlighted the most computationally important subclasses of medications within each of the ATC level 2 classes.

INTERPRETATION

This study leveraged both LR and ML to generate insights about medications and mortality during EHEs. The results add to the existing evidence on pharmaceutical risks during EHEs and provide new avenues for further research. They can be used to help develop more targeted messages to inform individuals whose medications put them at greater risk during EHEs.

FUNDING

BC Centre for Disease Control and Ministère de l'Enseignement supérieur du Québec.

MeSH

Mortality, Biomedical Research, Pharmaceutical Preparations, Machine Learning, Therapeutics, Learning