Methodology & caveats

Supply (access)

Enhanced Two-Step Floating Catchment Area (E2SFCA) with a Gaussian distance-decay kernel (σ=3km): each facility's contribution to a community area falls off smoothly with distance and its capacity is shared across the populations that can reach it. Each facility is weighted by size × Medicaid-accessibility — a large hospital counts more than a storefront clinic, and supply only "counts" to the extent it's reachable by Medicaid/uninsured patients (FQHCs = 1.0 by mandate; hospitals by their Medicaid payer share; behavioral facilities by whether they accept Medicaid vs. cash/private only). Restricted to Chicago facilities. This moves behavioral facilities from 72% of raw count to 38% of weighted supply, and lifts FQHCs to 57% — correcting the over-count from sheer facility volume.

Need

Community-area prevalence + SDOH from Metopio (diabetes, hypertension, asthma, obesity, mental distress; poverty, uninsured), 2023 where available.

Capacity Balance

Standardized supply_z − need_z across the 77 areas, on a diverging scale: strongly negative = access gap (build), near zero = matched, strongly positive = excess capacity (consolidate). Excess isn't free — it fragments volume across too many small providers, the same thin-margin dynamic the financial modules measure.

Emergency care & Med/Surg (hospital-based)

These two ride on hospital supply, not clinics, so they behave differently. Emergency supply = ED-equipped hospitals weighted by capacity × trauma-center uplift (IDPH Region 11 Level I in-cohort: Stroger, Mount Sinai, UI Health), with no Medicaid down-weight — EMTALA means EDs can't turn patients away — and a wider σ≈5km. Med/Surg supply = staffed beds × Medicaid share × open-bed availability, σ≈8km (people travel further for inpatient). Demand is a modeled composite because all-payer ED/discharge utilization isn't published at community-area level: ED from uninsured + ACSC-condition prevalence + behavioral distress; Med/Surg from chronic-disease burden. Measured upgrades pending — observed utilization (IDPH/HCUP discharge data) and HCRIS med/surg-specific bed counts (the Lumen Medicaid claims table is provider-keyed with no patient residence, so it can't drive demand here).

Limitations

The weights are crude: hospital size from beds, FQHC from federal mandate, behavioral from SAMHSA payment + ownership (Medicaid payer mix per facility, and true throughput capacity — beds/slots/encounters — are refinements). Suburban facilities are excluded, so a handful of border areas may slightly undercount access from immediately-adjacent suburban providers. Very-low-population areas can still over-read as "excess" because per-capita supply inflates — area population is shown in the detail panel. Claims-based utilization (T-MSIS) is a planned second demand lens.