Temperature exposure and analysis code for a nationwide study of road-traffic mortality in Thailand, 2011–2025

Harmonised daily temperature exposure for Thailand, with the analytic panels and completecode of a nationwide time-stratified case-crossover study of ambient temperature androad-traffic mortality (283,847 deaths, 77 provinces, 1 January 2011 to 24 August 2025). The exposure data were assembled for this study and form the largest part of the deposit:- 4,514,684 daily station observations from 1,347 stations (1 January 2000 to 31 July 2026), harmonised from the Thai Meteorological Department synoptic network (via NOAA GSOD) and the Hydro-Informatics Institute telemetry archive, with per-station quality control;- province-day maximum, minimum and mean temperature, humidity and precipitation;- daily maximum temperature reconstructed on 0.1-degree cells by inverse-distance weighting from 1,145 stations (16,629,330 cell-days). The grid covers only the 3,965 cells that contain geocoded crash sites, from 2012 to 2023; median distance from cell to nearest station is 10.3 km. The analytic panels contain province-day death counts with all model covariates, a humiditypanel, and an individual-level crash-to-death panel from which per-death coordinates wereremoved. The conditional Poisson estimator was written for this study (Newton-Raphson with analyticgradient and Hessian, cluster-robust sandwich variance) and validated against a statsmodelsPoisson GLM with dense stratum fixed effects; coefficients agree to 5.8e-15. The 30 scriptscover parsing, exposure construction, the main models, five pre-specified sensitivityanalyses (spatial exposure resolution, crash-date versus death-date exposure assignment,holiday-period exclusion, fifteen consecutive-day heatwave definitions, and humid-heatmetrics), the attributable-burden estimand, and the figures. A file manifest with SHA-256checksums is included. Also included is a 608-line replication protocol, written in Portuguese, documenting thetraps encountered in this analysis, with an annex for applying it to Brazil (SIM/DATASUSmortality and INMET weather data). The road-traffic death records themselves are not redistributed. They are publicly availableunder CC-BY from the Thailand Open Government Data portal: https://data.go.th/dataset/rtddi