Methodology

Last updated: April 16, 2026

This page explains how the cost figures on CareCostIndex are produced, which sources we rely on, and what the published numbers do and do not mean. We publish this so readers can evaluate our estimates on their merits rather than take them on faith.

1. Primary data sources

• Centers for Medicare and Medicaid Services (CMS) — the Medicare Physician Fee Schedule (MPFS), Hospital Outpatient Prospective Payment System (OPPS), Ambulatory Surgical Center (ASC) payment system, and Inpatient Prospective Payment System (IPPS) files provide the allowed-amount baseline for many procedures. Medicare rates are a floor, not a ceiling — commercial insurers typically pay 1.5–3× Medicare for the same code. We pin each procedure page to the CMS file and version (for example, “MPFS CY2025 final rule”) that the figure derives from.
• Hospital Price Transparency disclosures (HPT MRFs) — under the federal Hospital Price Transparency rule (effective January 1, 2021), U.S. hospitals publish machine-readable files listing their standard charges, negotiated rates with commercial payers, cash (self-pay) prices, and minimum/maximum negotiated rates. We sample these files for the procedures we cover, preferring hospitals in the largest metros of each state for sample coverage. Where a hospital’s MRF is known to be non-compliant (missing negotiated rates, broken JSON, redacted payer names), we exclude it and note the exclusion in our internal source log.
• Transparency in Coverage (TiC) payer files — the companion federal rule applies to commercial health plans; we sample payer-published in-network negotiated-rate files for cross-validation when a hospital figure looks anomalous.
• State All-Payer Claims Databases (APCDs) — a growing number of states (Colorado, New Hampshire, Washington, Oregon, Massachusetts, Minnesota, Maine, Utah, Vermont, Rhode Island) publish APCD summaries that show what providers in the state were actually paid across commercial, Medicare, and Medicaid populations. APCD release cadence and variable definitions vary by state, so we document the release version used for each state-level figure.
• Genworth Cost of Care Survey — used for our elder-care figures (nursing home, assisted living, in-home care, adult day care). Genworth’s annual survey is the most widely cited source for long-term-care pricing. We cite the survey year used.
• Professional association surveys — for procedures where CMS and hospital transparency data are thin (common in dermatology, dentistry, fertility, vision, and mental health), we triangulate against published surveys from the American Academy of Dermatology, the American Dental Association, the Society for Assisted Reproductive Technology, the American Society of Plastic Surgeons, and similar bodies.
• Published academic and journalistic research — peer-reviewed studies and investigative reporting (for example, work published by ProPublica, KFF, Health Affairs, the Health Care Cost Institute, RAND Hospital Price Transparency Study) provide cross-checks on our estimates. Academic studies are used for context and cross-validation only; a published figure on CareCostIndex must also trace to a government or payer primary source.

2. How we compute a national average

For a given procedure we compile the most recent observations from each source above, weight them roughly in proportion to population coverage (Medicare rates are common but do not dominate the commercial market), and produce a single national “average cost.” This figure is intended as a realistic midpoint for an insured patient paying a negotiated rate — not the sticker price on a hospital bill, and not the lowest cash price a self-pay patient might negotiate.

We do not present a single number as a hard point estimate. Each procedure page also shows a low and high end of the plausible range to reflect the genuine variability in the market.

2.1 Data validation steps

Every published figure goes through the following checks before it appears on the site:

1. Source trace — the figure is attached to a specific CMS file version, APCD release, or hospital MRF hash in our internal source log. A figure without a traceable primary source is not published.
2. Outlier exclusion — observations above the 95th percentile or below the 5th percentile of the observed commercial rate distribution are excluded from the national average. These caps are documented per procedure; we do not apply a silent across-the-board trim.
3. Cross-source consistency — when CMS-based and HPT-MRF-based estimates diverge by more than 40% after trimming, we investigate before publishing. Common causes include facility-fee differences (hospital outpatient vs ASC vs freestanding), site-of-service shifts, or modifier misapplication.
4. Year alignment — mixing a CY2024 CMS rate with a 2021 APCD release distorts the average. We only blend sources within the most recent 24 months of observations, and prefer same-year data where available.
5. Geographic coverage — before publishing a state-level figure, we confirm that at least three observations cover that state across sources. States with thinner coverage carry a “lower confidence” internal flag and are reviewed more frequently.

2.2 Price transparency compliance — how we handle non-compliant hospitals

The federal Hospital Price Transparency rule has been enforced unevenly. Our compliance assumption is that roughly one in five MRFs a user might open in the wild is either missing, stale, or structurally malformed. Our internal pipeline flags MRFs that are:

• Older than 12 months from their listed effective date
• Missing the required de-identified minimum/maximum negotiated rate fields
• Listing a single payer under a placeholder code (for example, “contracted” without named payer)
• Published in an undocumented custom schema that cannot be mapped to the CMS guidance

Flagged files are excluded from our sample, not silently smoothed over. Because excluded files skew our sample toward larger, better-resourced health systems, we explicitly note this bias where it materially affects interpretation (for example, for small-town procedures where the compliant-MRF population is thin).

3. How we compute state-level figures

State-level averages start from the national average and apply a geographic cost index derived from CMS’s Geographic Practice Cost Index (GPCI) for the physician component and the Medicare Wage Index for the facility component. Where state-specific APCD data is available, we use it directly and cross-check the derived adjustment.

Metro-level figures (where shown) apply an additional metro-area factor based on published commercial-insurance benchmarks for that market.

4. What is included in a “procedure cost”

Unless a page states otherwise, our cost figures are intended to reflect the total allowed amount for a standard, uncomplicated episode of the procedure — that is, the facility fee plus the professional (physician) fee plus routine ancillary charges (imaging contrast, standard lab work, post-op visits that are part of a global surgical package).

We exclude pre-existing chronic care, inpatient stays unrelated to the procedure, complications, and out-of-network balance billing. These exclusions are the single biggest reason an individual patient’s bill can exceed our estimate.

5. Update cadence

Each procedure page carries a last-reviewed date. Our editorial team re-reviews the top procedures quarterly and the remainder at least annually, or sooner when CMS or a major payer announces a rate change we believe will move the average by more than 5%.

6. Known limitations

• Hospital transparency data is incomplete. Compliance with the federal rule has improved but remains uneven; some filings are poorly structured or rounded.
• Cash and self-pay prices vary enormously. A self-pay patient who asks for a discount can sometimes pay less than the insured negotiated rate, and sometimes much more.
• Emergency and complicated cases can cost far more than our averages suggest.
• Dental, vision, and elective procedures often have thin public data and are more susceptible to estimation error.
• We do not model balance billing or surprise bills in the averages. The No Surprises Act limits these for many emergency and in-network settings, but not all scenarios.

7. Corrections

If you believe a cost figure is materially wrong, please send the URL, the figure in question, and a source we can check (a hospital transparency file, an APCD extract, a peer-reviewed study, or an itemized bill with PHI removed) to info@carecostindex.com. Reader corrections are the single most important input to each revision. We document every material change in the page’s last-reviewed date.

8. What our estimates are not

Our figures are not price quotes, insurance-benefits predictions, or legal or medical advice. See our medical disclaimer for the full scope note, and our Terms of Service for limitations on your reliance.