Hospitals are publishing their prices. Almost nobody can read the files.
Compliance with price disclosure rules has risen sharply. The published files run to millions of rows, use inconsistent codes, and in most cases cannot be compared between two hospitals in the same city.
- Formal compliance is now above 90%, up from about 30% three years ago
- Files average 2.4 million rows; only 14% used the recommended schema fields consistently
- For the ten most common procedures, same-city prices varied by up to 8x
Hospitals are now overwhelmingly complying with rules that require them to publish negotiated prices. Formal compliance is above 90 percent, up from about 30 percent three years ago. Usable disclosure is a different measure and it has barely moved.
The published machine-readable files average 2.4 million rows. A researcher who wants to compare the price of a knee replacement at two hospitals across town must reconcile different procedure code sets, different definitions of what the price includes, and different naming for the same insurance plan.
The schema problem
Regulators publish a recommended schema. In an analysis of 340 hospital files, only 14 percent populated the recommended fields consistently enough for automated comparison. The rest were technically compliant and practically opaque.
Some of this is deliberate and some is not. Hospital finance departments building these files from a chargemaster and dozens of payer contracts genuinely struggle to express a negotiated rate that depends on volume tiers and quality adjustments as a single number.
They published everything, which turns out to be an excellent way of publishing nothing.— A health economist who works with the disclosure files
What the data does show
Where comparison was possible, the variation was large. For the ten most common elective procedures, negotiated prices at hospitals within the same metropolitan area varied by as much as eight times, with no consistent relationship to published quality measures.
The same procedure at the same hospital also varied by a factor of three or more between insurers, which is the finding that has drawn the most attention from employers who purchase coverage.
- Compliance: above 90%, up from about 30% three years ago
- File size: 2.4 million rows on average
- Usable schema: 14% of 340 files populated recommended fields consistently
- Variation: up to 8x for the same procedure within one metro area
For a patient, the practical route to a price is still to call the hospital’s billing department and ask for a written estimate for a specific procedure with a specific insurer, which the same rules also require. Consumer advocates point out, without much satisfaction, that a phone call remains more effective than a database with two and a half million rows in it.



Hope there is a follow-up. These stories tend to drop out of view after a few weeks.
Something similar happened in my county, and this matches what I saw.
Something similar happened in my county, and this matches what I saw.
Would be better with more detail on where the figures came from, but otherwise solid.
Sent this to my family. They said the summary was fair.
I agree with about half of it. The rest depends on how it is actually implemented.