Data and reproducibility package for "Socioeconomic and health-system indicators and infant mortality in Latin America and the Caribbean, 2000–2023: an auditable fixed-effects panel reanalysis

This repository contains the data, executable code, extended tables, figures, reporting materials, and integrity-verification files supporting the research article “Socioeconomic and health-system indicators and infant mortality in Latin America and the Caribbean, 2000–2023: an auditable fixed-effects panel reanalysis.” The study analyses a World Bank-derived panel of 408 country-year observations from 17 Latin American and Caribbean countries between 2000 and 2023. The reference fixed-effects model includes 303 complete country-year observations from 16 countries and examines conditional within-country associations between infant mortality and extreme poverty, income inequality, government health expenditure, universal health coverage service coverage, and gross domestic product per capita. Statistical inference includes country-clustered confidence intervals, a 19,999-draw Webb-weight wild cluster bootstrap, delete-one-country jackknife analyses, missing-data sensitivity analyses, alternative temporal specifications, covariance diagnostics, and leave-one-country-out influence analyses. The repository includes the frozen analytic datasets, data dictionaries, the single executable Python analysis script, pinned software requirements, machine-readable results, publication figures, extended-data tables, STROBE reporting materials, metadata files, and SHA-256 verification manifests. The package is designed to reproduce the reported tables, figures, sample dimensions, and statistical estimates from a clean computational environment. The findings did not provide robust confirmatory evidence that within-country changes in the analysed national socioeconomic and health-system indicators were associated with infant mortality. Estimates for government health expenditure and universal health coverage were generally inverse but imprecise and sensitive to missing-data and temporal assumptions. The study therefore supports cautious interpretation of aggregate expenditure and coverage indicators and highlights the importance of effective coverage, quality, financial protection, implementation, resource allocation, and subnational equity. All data are aggregate country-year indicators; no individual-level, identifiable, clinical, or confidential information is included. The original indicators are publicly available from the World Bank World Development Indicators and Poverty and Inequality Platform. One original raw JSON file for the extreme-poverty indicator was not available in the supplied research materials; however, the frozen derived values used in the analysis, the recorded indicator code and provenance information, the analytic datasets, and all executable outputs are included and transparently documented.