Methodology
How we turn raw federal data into actionable hospital intelligence.
1. Data Ingestion
Each of our 60+ pipelines downloads source files, validates schema, and loads records into a normalized Postgres database. We deduplicate across sources using CMS Certification Numbers (CCNs) for facilities and National Provider Identifiers (NPIs) for clinicians.
2. Normalization
Raw measures vary widely in scale and direction (lower readmissions = better, higher star ratings = better). We normalize each measure to a 0–100 scale using min-max normalization within peer groups, adjusting for hospital type and size where appropriate.
3. Composite Scoring
Normalized measures are grouped into domains — quality, safety, patient experience, financial health, and technology. Each domain receives a weighted composite score. Domain weights reflect the relative importance to patient outcomes based on published health services research.
4. Ranking
Hospitals are ranked nationally and within state using their composite scores. Rankings are recalculated each time underlying data is refreshed (typically quarterly). Ties are broken by the quality domain score, then by patient experience.
5. Pricing Analysis
Machine-readable price transparency files are parsed to extract negotiated rates by payer, plan, and procedure code. We compute percentiles, national medians, and hospital-to-hospital comparisons. All prices are standardized to a common CPT/HCPCS code set for apples-to-apples comparison.
6. Linkage
Physician-to-hospital affiliations are built from CMS Provider Enrollment data and Medicare claims. Open Payments records are linked to hospitals via the affiliated physician's NPI, enabling facility-level views of industry payment patterns.
Our methodology evolves as new data sources become available and health-services research advances. Major methodology changes are noted in our release announcements.