SUNDAY · 07 JUNE 2026

ClearRate for Patients — Data Methodology

Where the numbers come from, what they mean, and where they stop

ClearRate for Patients uses federal price transparency data to estimate what you’ll actually pay at a specific hospital. This page explains exactly how that works — including the parts that are estimated rather than sourced from a contract.

1. Primary sources

Every facility rate in ClearRate for Patients is derived from one of two sets of federal price transparency disclosures, both mandated by the Centers for Medicare & Medicaid Services (CMS).

Hospital Price Transparency Machine-Readable Files (MRFs)

Since January 2021, every hospital operating in the United States has been required to publish a machine-readable file listing every negotiated rate the hospital has accepted from every payer for every item and service. The files are published on each hospital’s own website, updated at least annually, and must conform to the CMS 2.0 JSON schema or an approved CSV schema. Compliance is enforced by CMS with civil monetary penalties.

Branmoor downloads these files directly from each hospital and parses them in-house. We do not use a third-party data aggregator. The rates you see are sourced from the hospital’s own published disclosure — the same file the hospital filed with CMS.

CMS Plans & Benefits Public Use File (PUF)

Plan-level benefit design data — individual deductibles, out-of-pocket maximums, coinsurance percentages, and copay structures — comes from the CMS Plans & Benefits Public Use File, published annually on data.healthcare.gov. The PUF covers all qualified health plans sold on the Marketplace in each state. We supplement it with CMS’s Plan Attributes file for plan marketing names and metal tier classifications.

Medicare Fee Schedule

For hospital and payer combinations where no contract rate is available in the MRF, we fall back to the CMS Medicare Physician Fee Schedule (MPFS) as the base for an estimated rate. The methodology for gap-filling is described in section 5.

2. How we parse hospital price files

File acquisition

Hospital MRF URLs are sourced from the CMS Hospital Price Transparency compliance database, supplemented by direct lookup from each hospital’s website. Tennessee’s current build covers 21 hospitals across 8 metro areas.

Format handling

Hospital MRF files come in several formats. We handle all of them:

  • CMS 2.0 JSON — the current standard schema, used by HCA (TriStar), Covenant (Parkwest, Fort Sanders), VUMC, and CommonSpirit (CHI Memorial, Parkridge). Files range from 70K to 400K billing items. We stream these with an incremental JSON parser that processes records without loading the full file into memory.
  • CSV (standard charges format) — used by Williamson Medical, Regional One, Baptist Memphis, UT Medical Center, and the Ballad Health system (Tri-Cities). One row per line item, often millions of rows.
  • Gzip-compressed JSON — used by CommonSpirit. We detect the \x1f\x8b magic bytes and decompress before parsing.
  • ZIP-wrapped CSV — used by Erlanger (Chattanooga). We extract the inner CSV without writing to disk before processing.

What we extract

For each hospital file, we filter to a defined set of CPT and HCPCS procedure codes corresponding to the services the tool covers (imaging, lab, office visits, procedures, and emergency care). The full CPT list is maintained in our services configuration. For each matching billing item, we extract:

  • The negotiated dollar amount (standard_charge_dollar in CMS 2.0 JSON; the relevant negotiated-dollar column in CSV files)
  • The payer name and plan name (used for payer matching — see section 4)
  • The billing class (facility vs. professional, where specified)

We exclude percentage-of-billed-charges rates and fee-schedule lookups. A rate of “150% of Medicare” is not a dollar amount; we do not convert it and do not display it as if it were.

232,204 negotiated rate rows extracted across TN
20 / 21 hospitals with usable dollar-rate data
8 metro areas covered in Tennessee

3. How rate ranges are constructed

The tool shows a low / mid / high range rather than a single rate. Here’s why, and how each bound is calculated.

Why a range?

Even within a single payer contract at a single hospital, a service category like “imaging” covers dozens of distinct CPT codes with different rates — a chest X-ray, a CT scan, and an MRI are all imaging, but their contracted rates can differ by 10×. Rather than show a single number that implies false precision, we show the distribution of what that category of service actually costs.

Percentile construction

For each hospital + payer + service category combination, we collect all matching negotiated dollar amounts from the MRF. We then compute:

low = 25th percentile of matched rates
mid = 50th percentile (median) of matched rates
high = 75th percentile of matched rates

The mid figure is the number we use for out-of-pocket cost estimation. The low and high bounds show the realistic range for services in that category at that facility with that payer.

What the range does not mean

The range does not represent what you will be charged. Hospitals bill at the chargemaster rate; the negotiated rate is what the payer contractually pays after adjudication. Your actual bill will be the negotiated rate (or a portion of it, depending on your deductible and coinsurance). The tool estimates that patient share — see section 6.

4. Payer and plan matching

MRF files list payer names as free text strings. The payer name for BCBS Tennessee might appear as “BlueCross BlueShield of Tennessee,” “BCBS TN,” “BCBST,” or any number of variations depending on the hospital’s billing system. We maintain a pattern library that maps these variations to a canonical payer key.

Matching is done by normalizing both the MRF payer string and our pattern library to lowercase, stripping punctuation, and checking for substring matches and exclusion patterns (to avoid false positives — e.g., “Cigna HealthSpring” is Medicare Advantage, not commercial Cigna). Rates that do not match any known payer are pooled into an “unknown” bucket and used as a fallback.

On payer matching completeness

Not every payer in our display list has a named rate at every hospital. When a hospital’s MRF doesn’t include a specific payer (for example, TriStar Centennial may not list named Ambetter rates), we fall back to an estimated rate rather than show nothing. See section 5.

5. Where rates are estimated

This is the most important section to read if you want to understand the limits of what the tool shows.

Each hospital’s MRF contains rates for only the payers with whom it has a named contract. For a given hospital, this might be 4 to 12 named payers. The tool displays rates for up to 11 payers in Tennessee. For payer + service category combinations where no direct MRF rate exists, we use an estimate.

How estimates are constructed

The fallback estimate is calculated as:

estimated rate = Medicare facility rate × market segment multiplier × service category adjustment

# Market segment multipliers are derived from CMS actuarial data
# and validated against observed MRF rates in the same metro.
# Example: commercial payers in mid-urban TN markets ≈ 1.7–2.0× Medicare

The Medicare rate used is the CMS Physician Fee Schedule facility rate for the representative CPT code in each service category, at Tennessee’s applicable geographic payment locality.

Coverage breakdown

Based on our Tennessee build, here is the approximate breakdown of how displayed rates are sourced:

Source Approx. share Description
MRF direct ~15% Named payer rate extracted directly from the hospital’s machine-readable file for the matched CPT codes
MRF (unknown bucket) ~5% Real dollar rate from the MRF, but the payer name did not match our pattern library — pooled with unmatched payer contracts
Estimated ~80% Medicare rate × market multiplier. No direct contract rate available for this hospital + payer + service category combination

The 80% estimate share reflects a structural limitation of the MRF format: hospitals are required to publish the rates they have contracted, but a hospital that primarily contracts with 5 payers will only publish 5 payers’ rates. Smaller payers, newer market entrants, and out-of-state plan variants are routinely absent from MRFs.

We do not distinguish between MRF-sourced and estimated rates in the current UI. This is a known limitation we intend to address. Until then, treat all rates as estimates with varying degrees of primary-source backing.

6. Patient cost calculation

The tool converts a facility negotiated rate into an estimated patient out-of-pocket cost using your plan’s benefit design. The calculation is:

if deductible_remaining > 0:
  patient_pays = min(negotiated_rate, deductible_remaining)
  above_deductible = max(0, negotiated_rate − deductible_remaining)
  patient_pays += above_deductible × coinsurance_rate
else:
  patient_pays = negotiated_rate × coinsurance_rate

patient_pays = min(patient_pays, oop_remaining)

Deductible, out-of-pocket maximum, and coinsurance are pulled from the CMS Plans & Benefits PUF for your selected plan. The remaining deductible and OOP you enter are used as-is — we do not have access to your actual claims history.

What this does not include

  • Professional fees. Physician, radiologist, anesthesiologist, and other professional charges are billed separately and are not in the hospital MRF. A CT scan at the hospital facility has a separate read fee from the radiologist that can range from $50 to $400+.
  • Copays. Some plans use a fixed copay for outpatient services rather than (or in addition to) coinsurance. The tool displays the copay structure from your plan benefits but applies it only where the data is unambiguous.
  • Network determination. The tool assumes all displayed hospitals are in-network for your plan. Whether a specific facility is actually in your plan’s network is not verified. Check your plan’s provider directory before scheduling.
  • Bundled payments and case rates. Some services are priced as bundles across facility and professional components. The tool does not model bundled payments.

7. What’s covered, what isn’t

Service categories

The tool currently covers five service categories, each mapped to a set of high-frequency CPT and HCPCS codes:

  • Imaging — X-ray, CT, MRI, ultrasound, nuclear medicine scans
  • Labs — comprehensive metabolic panel, CBC, lipid panel, HbA1c, thyroid, urinalysis, blood draw
  • Office visits — established and new patient evaluation & management (E/M) codes, levels 3 and 4
  • Procedures — colonoscopy, upper endoscopy, joint injection, skin biopsy
  • Emergency care — ED evaluation and management codes, levels 3 through 5

The hospital that doesn’t have rates

Methodist Le Bonheur Healthcare (Memphis) publishes only a chargemaster (CDM) file rather than a payer-specific negotiated rate file. CDM rates are the hospital’s list price before any payer discount — not the contracted rate. We do not use CDM rates for this tool.

Insurers in the dropdown

The tool shows only the medical insurance companies available on Tennessee’s 2025 ACA Marketplace. Dental-only issuers (Delta Dental, DentaQuest, Guardian) are filtered out. Marketplace plans are used as the benefit-design source; employer-sponsored and Medicaid/Medicare plans are not currently modeled.

8. Update cadence

Hospital MRFs are re-downloaded and re-parsed when a new version is detected or when a hospital’s published compliance date updates. CMS requires hospitals to update their MRFs at least once per year; in practice, most large hospital systems update quarterly or when contracting cycles close.

Plan benefits data is refreshed when CMS publishes a new plan year PUF, typically in November for the following plan year.

The current Tennessee build was last parsed in May 2026. Rates reflect contracts as published in each hospital’s MRF at that date.

9. Limits of this tool

ClearRate for Patients is the most transparent cost estimation tool we know of for Tennessee hospital services. It is also genuinely incomplete in several ways you should know before acting on its numbers:

  • ~80% of displayed rates are estimates, not contracts. See section 5. A rate tagged to your payer at a given hospital is more likely derived from a Medicare multiplier than extracted from a named contract.
  • MRF data accuracy depends on the hospital. Hospitals are required to publish accurate rates, but data quality varies. Rates that look anomalous (very high or very low) may reflect publishing errors in the hospital’s own MRF.
  • Your actual cost will differ. Actual charges depend on what the provider actually bills, which modifiers apply, whether pre-authorization was obtained, whether services are bundled, and how your plan adjudicates the claim. Estimates here are a directional guide, not a quote.
  • Networks are not verified. A facility appearing in the tool is not a guarantee it is in-network for your specific plan. Verify with your plan’s provider directory.
  • Tennessee only. This build covers Tennessee. Other states are not yet available.
  • No employer-sponsored or Medicaid/Medicare plans. The benefit design data comes from the ACA Marketplace. Employer plan benefits vary significantly from Marketplace plans.

10. Medical and financial disclaimer

ClearRate for Patients provides estimates for informational purposes only. It is not a guarantee of charges, a determination of your benefits, or medical advice. Nothing on this tool should be used as the basis for a clinical decision.

Cost estimates are based on publicly available data that may be outdated, incomplete, or inaccurate. Always verify your out-of-pocket cost with your insurance company and the facility’s billing department before scheduling a service.

Branmoor is not a licensed insurance agent, broker, or financial advisor. We do not receive compensation from hospitals, insurers, or payers for rate data or tool placement.

Corrections

If you find a rate that looks wrong — dramatically inconsistent with what you paid or were quoted — email corrections@branmoor.com with the hospital name, service, payer, and the discrepancy. We will check the source MRF and update or flag accordingly.