What are machine-readable files?

What is an MRF?

A machine-readable file (MRF) is a large, standardized data file that a health plan publishes to disclose the prices it pays for care. The files list the negotiated rates a plan has agreed with every in-network doctor, hospital, lab and clinic, along with historical payments for out-of-network care.

“Machine-readable” means the data is published in a structured format, JSON, that follows a public schema maintained by CMS. It’s built so software can process it, not so a person can open it in a spreadsheet.

Why MRFs exist

For decades the prices insurers paid providers were confidential. Patients learned what something cost only after the bill arrived, and employers paying for their workers’ care often couldn’t see the rates either.

The Transparency in Coverage rule, finalized in 2020, changed that. It requires plans to publish their rates publicly, for free, without a login. The MRF requirement took effect for plan years beginning on or after January 1, 2022, and enforcement began July 1, 2022.

Who publishes them

Most non-grandfathered group health plans and health insurance issuers in the individual and group markets must publish MRFs. That includes national carriers, regional Blue Cross Blue Shield plans and self-funded employer plans, which usually publish through the insurer or third-party administrator that runs their claims. The organization posting the files is called the reporting entity.

Files must be updated monthly and must show the date they were last updated.

The types of MRF

• In-network rate files list every negotiated rate, by billing code and provider. They’re by far the largest and most useful files.
• Allowed-amount files report what the plan allowed for out-of-network services over a historical period, with billed charges.
• Table-of-contents (index) files map each plan to the rate files that apply to it, so many plans can share one network’s file.
• Prescription drug files were also part of the rule, but their enforcement was deferred, so most plans don’t publish them.

What’s inside an in-network file

At the top level, an in-network file names the reporting entity and the date it was last updated, then lists every covered item or service. Each item looks roughly like this:

{
  "billing_code_type": "CPT",
  "billing_code": "70551",
  "description": "MRI brain without contrast",
  "negotiated_rates": [{
    "provider_references": [42],
    "negotiated_prices": [{
      "negotiated_type": "negotiated",
      "negotiated_rate": 538.00,
      "billing_class": "professional",
      "service_code": ["11", "22"],
      "expiration_date": "9999-12-31"
    }]
  }]
}

The key fields are:

• Billing code: the CPT, HCPCS or MS-DRG code for the service.
• Negotiated rate and type: a dollar amount, a percentage of billed charges, a fee schedule or a per diem.
• Billing class: professional (the clinician’s fee) or institutional (the facility’s fee).
• Service code: the place of service the rate applies to.
• Provider references: pointers to provider groups, each a TIN and a list of NPIs.

For a full walkthrough, read how to read an in-network rate file.

Why they’re hard to use

• Size. A large carrier can publish thousands of files each month, many terabytes uncompressed. Single files can exceed what a laptop can open.
• Repetition. The same rate is often repeated across dozens of plans and files that share a network.
• Indirection. Rates point to provider groups that are sometimes stored in separate files, and providers are identified by number, not name or address.
• Variation. Carriers interpret the schema differently, and file layouts and URLs change from month to month.

Making MRFs useful means streaming them, de-duplicating rates, resolving provider references, and joining NPIs to the public NPPES registry to get names, specialties and locations. That’s the work US Health.io does.

MRFs vs. hospital price transparency files

Hospital price transparency is a separate rule, in effect since 2021, that requires each hospital to publish its own standard charges. The two datasets complement each other:

Using MRF data

Clean MRF data answers questions that used to be unanswerable: what an MRI costs at each imaging center in a city, how one hospital’s rates compare to its competitors, or whether an employer’s plan pays more than the market. See who uses MRF data and how, or browse the glossary for any term you didn’t recognise.