# Additional file 3 — Data and code availability ## Public source data The study used de-identified public data from Colombian government agencies: 1. SIVIGILA event 113 notification microdata, 2024: Instituto Nacional de Salud, . 2. DANE non-fetal death microdata, catalogues 807 (2022), 876 (2023), and 878 (2024): . 3. DANE municipal population projections by area, sex, and single year of age, 2018–2042: . 4. DANE census-source municipal Multidimensional Poverty Index, 2018: . 5. DANE National Geostatistical Framework (MGN), 2024: . The exact filenames, direct download endpoints, retrieval dates, sizes, and SHA-256 hashes are recorded in `source_manifest.csv` in the repository. ## Reproducibility deposit The derived non-identifiable municipal analysis file, executable Python code, complete model and sensitivity output, figures, environment specification, and source manifest are deposited in Zenodo: **DOI: [ZENODO DOI — reserve and insert before journal submission]** The Zenodo record must be published or at least reserved before submission so a stable DOI can be entered in the manuscript's data-availability statement. The deposit contains no individual-level notification or death records. ## Reproduction command After installing `requirements.txt` and downloading the public inputs: ```bash python code/rebuild_analysis.py --raw raw --out reproduced ``` The pipeline validates the audited source counts and writes every reported table, figure, model, spatial diagnostic, and sensitivity result. Randomised Moran tests use the fixed seed 20260731. ## Licences and attribution Derived data and figures are released under CC BY 4.0; code is released under the MIT licence. Source data remain governed by their producing agencies. DANE requests the attribution: “Fuente: Departamento Administrativo Nacional de Estadística: www.dane.gov.co”.