Why this project
The ECG toolkit answers a clinical question with a model. This project is the other half of the work: taking messy public datasets, joining them, and answering a question a health system executive actually asks on a Monday morning.
I have spent four years inside a cardiac service line. The questions asked in that room are financial as often as they are clinical. Which payers reimburse well for the services we run? Are our contracts below what comparable hospitals negotiated? Where is the gap between the care we deliver and the revenue it produces?
I narrowed the work to the services a clinical exercise physiologist runs every day: cardiac rehab sessions and exercise, echo and nuclear stress testing.
What the dashboard shows
- Price. Commercial plans pay Washington hospitals a typical $363 for a monitored cardiac rehab session, second among states with enough price files to compare (Oregon is first at $382). The typical US commercial rate is $166. Medicare and Medicare Advantage pay about $132.
- Enrollment. In Washington’s typical county, 32% of eligible Medicare patients start cardiac rehab, against 36% nationally.
- Retention. 26% of Washington starters finish all 36 sessions, against 33% nationally. The average starter attends 23.5.
- What it is worth. For every 100 eligible patients, a Washington program brings in roughly $150,000 a year at a typical payer mix. Matching a top-quarter US county on enrollment and attendance adds roughly $105,000.
The dashboard has three parts: a US map that compares states on price, enrollment, completion and unbilled sessions; a Washington view that places one hospital’s contracts against its peers, insurer by insurer; and a program model that prices referral, enrollment and attendance improvements, or forecasts a program from its own monthly numbers.
How it works
↓
Keep commercial fee-schedule contracts — plan names classified, physician fees and unit errors dropped
↓
Match facilities to CMS — 94 Washington facilities reviewed by hand; ERs and clinics rejected
↓
Benchmark — insurer × service × hospital type, each price file counted once
↓
Join CDC Medicare data — who starts, who finishes, sessions attended, by county
↓
Dashboard and program model
Price files carry no certification number, so every Washington facility was matched to its CMS record by name and address and checked by hand. Twenty of 94 were rejected: freestanding emergency rooms, clinics, a psychiatric hospital, and two locations that one rural district’s file lists by mistake.
What changed from the plan
The original plan multiplied rates by CMS outpatient volumes. That file is now published by payment group rather than procedure code, and it leaves cardiac rehab out entirely, so per-hospital volumes are not available for free. The dashboard measures volume a different way, per 100 eligible patients from CDC data, and lets a program bring its own monthly numbers for a forecast.
What the data cannot support
- A published rate is a contract term, not proof of what was paid or that the service is offered.
- Rates are self-published and hospitals interpret the schema differently. Some states’ rates are implausibly low, which points at filing practice rather than payment.
- The CDC publishes county figures only, so state values are the typical county, not a patient-weighted average. They cover Medicare fee-for-service patients in 2019–2020.
- This estimates revenue. It does not estimate cost, so it cannot speak to margin.
Status
Beta. Next: a second price snapshot to start a price time series, pulmonary rehab, and splitting results by hospital system.
I work at UW Medicine. Every input is public data, and UW Medicine facilities are not excluded.