Methodology
How we verify every number.
This is what a skeptical CFO checks before a call, not after. Every rate figure we cite comes from CMS hospital price transparency machine-readable files (MRFs) — the standardized files hospitals are federally required to publish. We don’t estimate, model, or extrapolate rates. If a number is in one of our reports, it’s because we pulled it directly from a hospital’s own published file.
CMS Hospital Price TransparencyHow we build a peer set
For Michigan Critical Access Hospital comparisons, we cross-check candidate peers against independent CAH status — ownership, system affiliation, and bed count — before including them. Hospitals that look independent but are owned by or merged into a larger regional system get excluded, because that changes their payer leverage and makes the comparison misleading. Where a peer has a real but limited affiliation (shared clinical or EMR systems without shared payer contracting, for example), we keep them in the set but flag it explicitly in the report.
Michigan Center for Rural Health — Critical Access HospitalsWhat we check before a number ships
MRF data is public, but it’s also inconsistent — hospitals report it in different formats, with different levels of granularity, and errors are common. Before any figure ships, we check for:
Component mismatches
Professional-only rates (modifier 26) getting averaged in with global rates for the same code, which distorts the comparison. We catch these and note it when we do.
Plan-type ambiguity
Where a payer category (like “BCBS Commercial”) bundles distinct products — PPO and HMO, for example — without the hospital’s own disclosure separating them, we say so and treat the number as directional, not precise.
Thin peer support
If a code is only backed by a handful of peers, or the spread between peers is wide, we flag it as a weaker data point rather than presenting it with the same confidence as a number backed by the full peer set.
Where AI fits in
We use AI-assisted tools to process large, inconsistently formatted MRF files quickly — pulling relevant codes, flagging outliers, and catching the kinds of component and formatting errors described above. Every figure that makes it into a client-facing report is manually checked against the source file before it’s included. AI speeds up the first pass; it doesn’t replace verification.
Verify it yourself
You don’t have to take our word for any of this. Every number we cite is traceable to a public source:
- —Pull the hospital’s own MRF file from CMS Hospital Price Transparency or directly from the hospital’s website (federally required to be posted).
- —Cross-check CAH status and independence against the Michigan Center for Rural Health or the Michigan Health & Hospital Association.
- —Ask us for the specific line item and file we pulled any figure from — we’ll point you to it directly.
If a number in one of our previews doesn’t hold up when you check it, we want to know before you do anything with it, not after.