About PriceTransparency.io

1. Discovery. We crawl each hospital's cms-hpt.txt file at the root of their website (45 CFR 180.50(b)(8), required since July 2024). That file lists the canonical MRF URL for the facility and any affiliated locations. Re-checked quarterly.

2. Download & parse. The MRF is fetched and parsed against one of three schemas: CMS v3.0.0 tall canonical (JSON or CSV), the legacy wide-format CSV/XLSX one hospital chargemasters still use, or our normalized internal shape derived from openpyxl/xlrd for XLSX/XLS files.

3. Filter. We extract rates for 209 target billing codes — the CMS-mandated 70 shoppable services plus top inpatient MS-DRGs by Medicare volume. Hospitals publish 5,000-50,000 codes per file; we filter aggressively because the long tail is dominated by chargemaster IDs and zero-priced line items.

4. Normalize payers. Raw payer names ("United", "UnitedHealthcare", "United HMO Inpatient", "commercial | united | all other plans") collapse to one canonical name per insurance brand via 33 regex rules. The original payer_name_raw field is preserved for audit.

5. Sanity-check. Rates outside $0.01 — $1M are discarded as ID / volume metric mis-extractions. The wide-format parser rejects any candidate "payer" column with a name matching code / chargemaster / item-ID patterns.

6. Compliance grade. Per-hospital A-F grade against 7 CMS-1717-F2 criteria: discoverable MRF, valid v2/v3 format, all 5 standard charge types, plain-language descriptions, updated within 12 months, ≥70 shoppable codes, payer diversity. Recomputed nightly.

It is what hospitals have agreed to accept from each payer for each procedure — the contracted rate before any patient cost-sharing. It's also what cash-paying patients are quoted (the "discounted cash" field).

It isn't a guaranteed bill. Your actual out-of-pocket depends on your specific plan's deductible, copay, and coinsurance, on whether the hospital is in-network for you, and on the actual mix of services delivered (a knee replacement typically includes lab, imaging, anesthesia, and PT services each with their own codes). Use this as a sanity check, not a binding quote.

Errata. Hospitals occasionally publish clearly-wrong rates (typos, placeholder values). We flag outliers but don't silently correct them — the raw URL is on every row so you can audit back to source.