This is a code example for the implementation presented in the paper "Estimating the heat-related mortality and morbidity burden in Quebec, Canada" to compute the heat burden for multiple health outcomes.
- By Jérémie Boudreault, Éric Lavigne, Céline Campagna and Fateh Chebana
Datasets are located in data/ folder :
health_data_synthetic.csv: Daily count* in each health region (RSS) for each health outcome (HO)weather_data.csv: Daily lagged values of weather and air pollution in each health region (RSS)meta_predictors.csv: Meta-predictors for each health region (RSS) and each health outcome (HO)heat_thresholds.csv: Extreme temperature thresholds (Q95) by health region (RSS)
*Real health data cannot be shared. So we replaced the count of each health outcomes with simulations from a Poisson distribution with λ=100. Obtained results will not make sense, but at least the user can run the code and understand the data structure.
The analyses are all performed in the main.R script. The script is separated in three main steps :
- Step 1 : DLNM fitting by region and heat burden computation
- Step 2 : BLUP from meta-regression and heat burden computation
- Step 3 : Pooled effect across the province and heat burden computation
Helper functions are located in the R/ folder.
Enjoy !