Package for INLA model fitting and temporal decomposition of heatstroke mortality risk in Japan

This archive provides a peer-review reproducibility package for INLA model fitting and temporal decomposition analysis of national relative risk (RR) changes in heatstroke mortality risk in Japan (model m4). Purpose The package is prepared to allow reviewers to reproduce: INLA model fitting (model m4) to obtain: fixed_effects_m4.csv year_effects_m4.csv Temporal decomposition of national RR changes (baseline year = 1995) into: warming effect (Temperature_anomaly) ageing effect (Elderly_ratio) residual temporal effect (RW1 year effect from INLA) This package does not include restricted death records. Included files covariates.csvNon-restricted municipality-year covariates used for model fitting and decomposition.Required columns: Year, Keycode_str, Temperature_anomaly, Elderly_ratio, Log_pop. fit_inla_bym2.RINLA model fitting script for model m4.This script generates the model outputs used in the decomposition: fixed_effects_m4.csv year_effects_m4.csv fixed_effects_m4.csvINLA fixed effects output (coefficients used for Temperature_anomaly and Elderly_ratio). year_effects_m4.csvINLA year effect output (used as the residual temporal component in decomposition). decompose_contributions_m4.RScript to decompose national RR changes relative to the baseline year and generate: decomposition CSV contribution shares CSV optional figure PDF Reproducibility note The package supports two levels of reproducibility: Full workflow: run the INLA fitting script, then run the decomposition script. Decomposition-only check: run the decomposition script directly using the included INLA output CSV files. The decomposition script is designed to run without restricted mortality records, using only the non-restricted covariates and INLA output CSVs.