Methodology

Data sources

• NPPES: The National Plan and Provider Enumeration System is CMS's authoritative registry of every licensed physician in the United States, carrying NUCC taxonomy codes, practice addresses, and license states. Source: download.cms.gov/nppes/. The file is refreshed monthly; the full release runs approximately 8 GB uncompressed (11 GB raw). The pipeline filters to Mississippi and the four adjacent states (Louisiana, Alabama, Arkansas, Tennessee) and retains active providers only.
• CAH Provider of Services file: The federal registry of every Medicare-certified facility, including Critical Access Hospital designation. Source: data.cms.gov. The current quarterly CSV is resolved programmatically via the data.json catalog and filtered to Mississippi CAHs using CMS category codes (PRVDR_CTGRY_CD = 01, PRVDR_CTGRY_SBTYP_CD = 11).
• HRSA HPSA designations: Health Professional Shortage Area designations from the Health Resources and Services Administration, used to escalate gap severity for CAHs located in HRSA-designated counties.
• Census 2020 ZCTA gazetteer: ZIP Code Tabulation Area centroid coordinates, used for geocoding physician practice addresses.
• Census 2020 county boundaries: The cb_2020_us_county_500k cartographic file, filtered to Mississippi (STATEFP 28). Used to render the choropleth layer and to assign each CAH to its county via point-in-polygon intersection.

Specialty taxonomy mapping

NPPES encodes provider specialty using NUCC (National Uniform Claim Committee) taxonomy codes. The index tracks 15 core specialties selected to align with the HRSA HPSA shortage categories. Each specialty is defined as a set of NUCC codes that includes relevant subspecialties; the complete mapping is published in the taxonomy.yaml file in the public repository. The taxonomy is intentionally inclusive, meaning more codes are mapped per specialty than a strict definition would require. This approach produces a slight overcount of available physicians, which keeps gap claims conservative rather than inflated.

Why 30 miles

Specialty coverage is measured within a 30-mile radius of each hospital. That threshold is not arbitrary: HRSA applies 30 miles as its standard definition of "reasonable access" in non-metropolitan areas when designating Health Professional Shortage Areas, the federal mechanism that governs loan forgiveness placements, J-1 visa waivers, and related workforce interventions.

An earlier design used 60 miles. At that radius, the catchment areas of Jackson, Memphis, Mobile, and New Orleans begin to overlap with rural Mississippi CAHs, masking the access gaps that rural patients actually face. At 30 miles, the index produces a picture that is consistent with what rural CMOs report from operational experience.

Geographic distance

Hospital coordinates are resolved to precise latitude and longitude via Nominatim (OpenStreetMap). Approximately half of Mississippi CAHs carry rural highway-style addresses (for example, "25117 HIGHWAY 51") that Nominatim cannot geocode; those hospitals fall back to their ZIP code centroid. Within a 30-mile radius, this fallback introduces less than one mile of positional error. CAHs using the centroid fallback are marked with an outlined map symbol so users can identify which positions are approximate. Physician locations are geocoded using the Census 2020 ZCTA gazetteer. All distances are computed using the haversine great-circle formula.

Gap scoring

Each (CAH, specialty) pair receives one of four severity classifications based on the count of active physicians of that specialty within 30 miles of the hospital:

• CRITICAL: zero active physicians of this specialty within 30 miles, and the CAH's county carries an HRSA HPSA designation.
• HIGH: zero active physicians within 30 miles.
• MODERATE: one or two active physicians within 30 miles.
• COVERED: three or more active physicians within 30 miles.

Map severity buckets

Each CAH is assigned a gap count equal to the number of its 15 tracked specialties classified HIGH or CRITICAL. County shading on the map reflects the highest gap count among all CAHs within that county, using the maximum rather than the mean, so that a county containing one severely gapped hospital is represented accurately even if other facilities in the county are better resourced:

• 0-2 gaps
• 3-5 gaps
• 6-9 gaps
• 10+ gaps
• No CAH in county: counties without a federally designated Critical Access Hospital are not scored.

Verification

Prior to publication of each dashboard build, five hospital-specialty pairs are selected at random and manually reviewed against external sources to confirm or refute the underlying gap claim. Confirmations are logged in an internal verification record keyed to a build identifier derived from the gap matrix's claim columns. Signoffs issued against a prior build are rejected by the render gate, ensuring that verification is specific to the data version being published.

Known limitations

• NPPES is refreshed monthly. A physician who recently relocated to Mississippi may not appear in the index until the following release cycle.
• NPPES records practice addresses, not hospital staff privileges or rural call coverage. A physician whose address falls within 30 miles of a CAH is not guaranteed to hold privileges at that hospital or to serve its patients. Operationally verified placement data is what closes that gap; this dashboard represents the public-data baseline.
• Medicare Part B aggregate utilization has not yet been incorporated for demand weighting. That refinement is planned for V2.5.
• ZIP centroid geocoding is slightly conservative for physicians located near a ZIP boundary. Within a 30-mile radius, the resulting measurement error is less than 1%.
• One Mississippi CAH (Hancock Medical Center, Bay Saint Louis) carries a PO Box ZIP code that does not appear in the Census ZCTA gazetteer, and its street address could not be resolved by Nominatim; it is excluded from the current build.