Answers are grounded in this paper and cite their sources.
Observational study reveals geographic redistribution of cardiovascular and respiratory hospitalizations following hurricanes, highlighting localized chronic disease vulnerabilities.
# Analysis code and derived results for: Hurricanes and the Burden of Chronic Disease — Cardiovascular and Respiratory Hospitalizations in Louisiana, 2001–2022 This repository contains the analysis code, derived (non-identifying) result files, and figures supporting the manuscript by Bushra et al. (submitted to *GeoHealth*). ## Important: source data are NOT includedThe primary data are the Louisiana Hospital Inpatient Discharge Database (LAHIDD), which contains protected health information and cannot be redistributed. They are available from the Louisiana Department of Health (https://ldh.la.gov/page/lahidd) under data-use agreement and approval. The encounter-level file produced by the build step is therefore **omitted** from this deposit. The Louisiana parish boundary shapefile is publicly available from the U.S. Census Bureau TIGER/Line files. ## Contents```code/ analysis scripts (Python 3) 01_build_and_storm_windows.py Reduce raw diagnosis records to one encounter per ID (principal diagnosis), classify cardiovascular/respiratory, and compute seasonally matched storm-window rate ratios. 02_storm_window_analysis.py Storm-window Poisson rate ratios + LOS/mortality from the built encounter file. 03_did_facility.py Coastal-vs-inland difference-in-differences (facility parish). 04_did_residence.py Difference-in-differences sensitivity using patient residence. 05_figure1_monthly.py Figure 1 (monthly cardiovascular/respiratory admissions). 06_figure2_map.py Figure 2 (coastal/inland parish exposure map).results/ derived, non-identifying outputs storm_window_results.csv Table 1 (per-storm observed vs baseline, rate ratios). did_results.csv Table 2 (facility DiD estimates). did_results_residence.csv Residence-based DiD estimates. did_descriptive_by_storm.csv Per-storm coastal/inland rate ratios (facility). did_descriptive_residence.csv Per-storm coastal/inland rate ratios (residence). los_mortality_pooled.csv Length of stay and in-hospital mortality. monthly_counts.csv Aggregate monthly admission counts (Figure 1 source). top_principal_codes_validation.csv Top principal diagnoses with ICD descriptions. summary.json Coverage, category shares, baseline-year set.figures/ Figure1_Monthly_Admissions.png, Figure2_Coastal_Inland_Map.png``` ## EnvironmentPython 3 with: pandas, numpy, scipy, statsmodels, pyarrow, matplotlib, geopandas.Install: `pip install pandas numpy scipy statsmodels pyarrow matplotlib geopandas` ## How to reproduce1. Obtain LAHIDD from the Louisiana Department of Health and place the merged CSV where the scripts expect it (edit the `SRC` path at the top of `01_build_and_storm_windows.py`).2. Run `code/01_build_and_storm_windows.py` to build the encounter-level file and storm-window results.3. Run `code/02`–`04` for the storm-window, facility-DiD, and residence-DiD analyses.4. Run `code/05`–`06` to regenerate the figures.File paths near the top of each script may need to be updated for your environment. ## Data definitions (summary)- Unit of analysis: one hospital encounter, classified by principal diagnosis.- Cardiovascular: ICD-9 402,404,410–438 / ICD-10 I11–I13, I20–I26, I30–I52, I60–I69; uncomplicated essential hypertension (ICD-9 401.x; ICD-10 I10) excluded.- Respiratory: ICD-9 460–519 / ICD-10 J00–J99.- Exposure: 60-day post-landfall window per storm; baselines are the same calendar window in covered non-storm years (2001, 2002, 2006, 2007, 2009, 2018, 2019, 2022).- Data coverage: 2001–2002, 2005–2009, 2018–2022 only (no 2003–2004 or 2010–2017). ## LicenseCode released under the MIT License. Figures and derived result files released under CC BY 4.0. ## CitationPlease cite the associated manuscript (Bushra et al.) and this software/data deposit by its archive DOI once minted.
No takes yet. Share an insight, caveat, or question.
Nazla Bushra (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: