Extreme heat and pneumonia mortality in Brazil, 2014–2024: aggregated mortality and meteorological data and code

Aggregated data and source code reproducing the analyses of "Extreme heat and pneumonia mortality in Brazil, 2014–2024: a nationwide case-crossover study of 877,000 deaths". The study asks whether pneumonia mortality responds disproportionately to extreme heat compared with other natural causes. It uses a time-stratified case-crossover design (location × year × month × day of week) with conditional Poisson regression and sandwich standard errors clustered by federative unit. The data are all 877,094 deaths with pneumonia (ICD-10 J12–J18) as the underlying cause in the Brazilian Mortality Information System, across all 27 federative units, 2014–2024. Exposure has two definitions:- Primary: the top decile of state-level maximum temperature within the same month.- Secondary: an INMET-based heatwave criterion (≥2 days ≥5 °C above a locally computed mean), applied per municipality with station data interpolated within 50 km.Both use extreme heat on the day of death or the three preceding days (lags 0–3). Results:- Monthly decile: the pneumonia RR was 1.047 (95% CI 1.034–1.060), versus 1.017 for other natural deaths. Ratio of RRs: 1.029 (1.015–1.044).- INMET-based criterion: 1.092 (1.058–1.128) versus 1.043, ratio 1.047 (1.011–1.084).- Timing: excess risk concentrated on lags 0–4, with no compensatory deficit up to 14 days.- Season: it was higher in November–March (1.177) than in June–August (1.021).- Age: among adults, it varied less with age than for other natural causes.Conclusions hold under a wild cluster score bootstrap. The deposit contains:- Unit-day and municipality-day death-count panels.- INMET temperature and precipitation series.- Python code producing every table, figure and number.- Reference outputs. Individual DATASUS microdata are not redistributed. A provenance file lists the 303 .dbc files and 12 annual INMET archives with SHA-256.