Temperature and non-violent mortality among young people in Mexico, 2015–2024. Replication data and R pipelines

This repository contains the data, model outputs and R code needed to reproduce a Bayesian spatiotemporal analysis of weekly ambient temperature and non-violent mortality among people aged 15 to 24 years in 2,469 Mexican municipalities, from 2015 to 2024. Death records come from the national vital statistics of the National Institute of Statistics and Geography (INEGI), population and the 2020 Index of Marginalization from the National Population Council (CONAPO), and weekly mean temperature from the ERA5-Land reanalysis. The models are negative binomial distributed lag non-linear models fitted with R-INLA, with lags of 0 to 3 weeks, sex-specific temperature effects, a BYM2 spatial effect, temporal trend and seasonality. The primary outcome is all non-violent deaths, decomposed into medical deaths and injury deaths excluding X59, and the package includes the outputs of the sensitivity analyses and of an exploratory screening of 17 cause groups. Death counts are aggregated by municipality, epidemiological week and sex and contain no individual-level information. A master README describes the package, each dataset folder has its own README, and a file index describes every file.