Reproducibility package for "Programmatic Transition to Post-Intervention Surveillance in Ghana's Lymphatic Filariasis Elimination Programme: A National Discrete-Time Analysis"

This reproducibility package accompanies the manuscript "Programmatic Transition to Post-Intervention Surveillance in Ghana's Lymphatic Filariasis Elimination Programme: A National Discrete-Time Analysis" (Adda, 2026).

Contents:

1. Derived analytic datasets:
   - iu_year_panel.csv: Implementation Unit (IU) by year panel, 2014-2024, with programmatic classification, cumulative MDA rounds, reported coverage, and environmental covariates.
   - hazard_intervals.csv: Person-interval dataset for the discrete-time hazard analysis of transition to post-intervention surveillance (218 intervals, 95 IUs, 92 events).
   - regional_summary.csv: Regional distribution of endemic classification and 2024 programmatic status, 2021-2024.

2. Analysis code:
   - 01_build_panel.R: Constructs the IU-year panel from the raw ESPEN extract.
   - 02_panel_regression.R: Fixed-effects and random-effects panel models (plm).
   - 03_hazard_analysis.py: Discrete-time life table, logistic hazard models, Moran's I diagnostics (statsmodels, libpysal, esda).
   - 04_figures.R and 04_figures.py: Scripts that regenerate every table and figure in the manuscript.

3. Data dictionary:
   - data_dictionary.xlsx: Variable definitions, units, and sources for all derived datasets.

4. README.md: Instructions for reproducing the analysis from the raw data.

Raw data source: The underlying lymphatic filariasis data are publicly available from the WHO Expanded Special Project for Elimination of Neglected Tropical Diseases (ESPEN) portal (https://espen.afro.who.int/). Raw ESPEN data are not redistributed here; the derived datasets and code allow full reproduction once the raw extract is downloaded from the portal.

Environmental covariates: Elevation and distance to stream were derived from the HydroSHEDS Conditioned DEM for Ghana, available from UNESCO IHP-WINS (https://ihp-wins.unesco.org/).

Software: R 4.3.3 (plm); Python 3.12 (statsmodels, libpysal, esda, geopandas, rasterio, rasterstats).

No individual-level or personally identifiable data are included. All data are aggregated at the Implementation Unit level.