HRSA AHRF · Census ACS · HPSA · CDC WONDER · GeoPandas · Statsmodels
A county-level analysis of prenatal care access gaps across North Carolina, with a focus on Western NC — a region whose pre-existing provider shortages were amplified by Hurricane Helene (September 2024). Integrates five federal data sources to build a composite burden index, identify zero-provider counties, and surface structural patterns that standard SDOH analyses miss.
- 27 of 100 NC counties have zero OB/GYN providers — statewide median is 0.9 per 10,000 women of reproductive age
- 96 of 100 counties carry active maternity care HPSA designations, reflecting near-universal shortage status across the state
- All 23 Western NC counties are suppressed in CDC WONDER natality data — the federal surveillance system is blind to the region most at risk, a finding consistent with Runkle & Sugg (2025, NC Medical Journal)
- WNC's crisis is not explained by standard SDOH: Mann-Whitney U tests show no significant regional difference in provider density, poverty, or vehicle access — only insurance coverage (p<0.001) and broadband (p=0.03) differ, both in WNC's favor. Geographic isolation and infrastructure loss are the more likely drivers
- Watauga County tops the composite burden index (0.90/1.0), driven by high uninsured rates among a large student-age population and zero OB/GYN subspecialists
27 of 100 NC counties have zero OB/GYN providers (gray). Care is concentrated in a handful of urban counties while Western NC and the eastern coastal plain are largely underserved.
Composite burden index combining inverse provider density, % uninsured women, and % poverty across all 100 counties. High-burden counties appear in both WNC and the eastern coastal plain — different geographies, similar structural gaps.
| Source | Variables | County Coverage |
|---|---|---|
| Census ACS 2023 (5-yr) | Insurance, poverty, vehicle access, broadband | 100/100 |
| HRSA AHRF 2024–2025 | OB/GYN provider counts (general + subspecialty) | 100/100 |
| HRSA HPSA Primary Care file | Maternity Care shortage scores (PC MCTA) | 96/100 |
| USDA Rural-Urban Continuum Codes 2023 | Rural classification (1–9 scale) | 100/100 |
| CDC WONDER Natality 2016–2024 | Prenatal care trimester initiation rates | 28/100 (suppression) |
Data pipeline: Multi-source integration across Census API, HRSA, USDA, and CDC; county-level FIPS-keyed master dataset; automated downloads for all sources except CDC WONDER (manual export required due to anti-scraping policy)
Provider density: OB/GYN providers per 10,000 women of reproductive age (15–44); source is HRSA AHRF tot_md_do_obgyn_gen_23 + md_nf_obgyn_subsp_23
Regional comparison: Mann-Whitney U (non-parametric) for Western NC vs. rest of state; 23 WNC counties defined by Appalachian Regional Commission designation
Composite burden index: MinMax-normalized equal-weight composite of inverse provider density, % uninsured women, and % poverty — built on all 100 counties to avoid CDC WONDER suppression excluding WNC entirely
OLS regression: Predictors of late/no prenatal care (28-county CDC WONDER subset); R²=0.436, F-stat p=0.025; interpret with caution given small n and high VIF on poverty/vehicle collinearity
| Figure | Description |
|---|---|
map_provider_density.png |
Choropleth — OB/GYNs per 10K women; zeros in gray, Blues scale for non-zero |
map_burden_index.png |
Composite burden tier map — High/Medium/Low burden across all 100 counties |
correlation_heatmap.png |
Spearman correlation matrix — SDOH and provider supply indicators |
regional_bar_charts.png |
WNC vs. Piedmont/Coastal mean comparisons across 5 variables |
provider_density_distribution.png |
Histogram + regional boxplot of provider density |
wnc_scatterplot.png |
Provider density vs. late/no prenatal care (28-county WONDER subset) |
Eight WNC counties (Buncombe, Haywood, Henderson, McDowell, Mitchell, Rutherford, Transylvania, Yancey) experienced significant healthcare infrastructure disruption from Hurricane Helene in September 2024. This analysis treats Helene as amplifying pre-existing structural gaps rather than creating them. Per Runkle & Sugg (2025), 6 labor and delivery units closed in WNC between 2015 and 2024, and only half of local facilities offered prenatal/delivery care before the storm.
Runkle, J. & Sugg, M. (2025). Rebuilding Maternal Health Access in Western North Carolina: Addressing Critical Gaps Amplified by Hurricane Helene. NC Medical Journal, 86(1), 8–11. https://doi.org/10.18043/001c.137497
nc-prenatal-care-access/
├── notebook/
│ └── NC_Prenatal_Care_Analysis_CLEAN.ipynb # Full analysis notebook
├── data/
│ └── processed/ # Merged county-level dataset
│ ├── master_county.csv # 100 counties × 32 variables
│ ├── burden_index.csv # Composite burden scores + tiers
│ └── wnc_county_profiles.csv # Western NC deep dive table
├── outputs/
│ ├── figures/ # All 6 visualizations (PNG)
│ └── tables/ # Summary stats, regression, Mann-Whitney
└── README.md
Note on raw data: Raw files from HRSA, USDA, and ACS are pulled automatically by the notebook at runtime. The CDC WONDER
.txtfile requires a manual export from https://wonder.cdc.gov/natality-current.html (see notebook Cell 5 for instructions).
git clone https://github.com/JohnApelJr/nc-prenatal-care-access.git
cd nc-prenatal-care-access
# Open notebook/NC_Prenatal_Care_Analysis_CLEAN.ipynb in Jupyter or Google Colab
# Run cells top-to-bottom — all data pulls automated except CDC WONDER (Cell 5)Requirements: Python 3.10+, geopandas, pygris, pandas, numpy, matplotlib, seaborn, scipy, statsmodels, scikit-learn, requests
- nc-pediatric-access-pipeline — Pediatric specialist supply vs. child population demand across NC's 100 counties
- chronic-disease-geospatial-pipeline — Chronic disease disparities across 2,956 U.S. counties
M.S. Applied Data Science · Syracuse University · February 2026
Contact: johnapeljr@gmail.com · Portfolio