# State concentration across 13 organization NPI profiles

Group each retained organization CSV by the primary practice state field. Count NPI rows in each group, preserve the source state code, and divide by the total NPI rows in the same organization file. Emit one row per organization and state-code pair. The summary table shows the largest group per organization. No secondary address, mailing address, corporate headquarters or external ownership data enters the calculation.

A state-code count measures where the selected registrations list their primary practice addresses. It is not a count of unique facilities, active offices, customers or headquarters. Codes can include the District of Columbia and territories; a count above 50 must not be presented as more than 50 states. Selected legal names do not necessarily cover a whole brand, and the cohort is not representative of all providers.

Reproduce: python3 scripts/build_research_drafts.py organization-npi-state-footprints; then python3 scripts/build.py. Inputs and SHA256 hashes are in provenance.json.