@pipeworx/insurance-denials

Connect: https://gateway.pipeworx.io/insurance-denials/mcp · Install: one-click buttons

Tools: 4

US health-insurance claim denial rates and appeal outcomes, per insurer — which companies deny the most claims, and whether appealing one actually works.

Tools

Auth

None. No key, no account.

Data sources

The CMS Transparency in Coverage Public Use File, plan year 2026, published by the Centers for Medicare & Medicaid Services:

• Landing page: https://www.cms.gov/marketplace/resources/data/public-use-files
• File: https://download.cms.gov/marketplace-puf/2026/transparency-in-coverage-puf.zip

Issuers report these figures to CMS themselves under 45 CFR 156.220. CMS publishes the file once a year, as an XLSX inside a ZIP.

What this covers, and what it does not

Two limits, both easy to mistake for a data bug:

HealthCare.gov states only — 30 of them. An issuer on a state-run exchange (California, New York, Massachusetts, Washington, Colorado, Pennsylvania, New Jersey and others) reports to that state, not into this federal file. A caller asking about California gets an explicit state_not_in_source refusal naming the reason, rather than an empty list that would read as “California insurers deny nothing”.

Self-reported. These are the issuers’ own counts, and CMS says so in the file’s own disclaimer. They are not an audit.

Reading the numbers

• Dental denies far more than medical. Standalone dental plans (Individual SADP) sit at the top of any unfiltered denial ranking — several above 70%. Pass market: "medical" when the question is about medical cover, or the answer will be true and misleading.
• Suppressed is not zero. CMS marks a figure it cannot publish with * or **. Those are null here, never 0 — a suppressed denial count reported as zero would describe a perfect insurer.
• Internal and external appeals differ. Internal appeals are decided by the insurer that issued the denial; external appeals go to an independent reviewer. The external number is the arm’s-length one.
• One filing per state. A multi-state insurer reports separately in each state, so insurer_denial_profile returns one entry per state rather than a single national figure.

Refreshing

Annual, when CMS publishes the next plan year:

node scripts/refresh-insurance-denials.mjs   # rewrites src/data.json

The script re-reads the CMS file and rolls it up to issuer level. It dedupes by issuer id and never sums: the Issuer_* columns repeat identically on every plan row an issuer has, so summing multiplies each figure by that issuer’s plan count. Verified when the pack was built — 0 of 185 issuer ids varied across their own rows.

Tools

• insurer_denial_rates — Which health insurers deny the most claims — US marketplace issuers ranked by the share of in-network claims they denied, for plan year ${PLAN_YEAR}. Answers “what percentage of claims does my insuran
• insurer_appeal_outcomes — Does appealing a denied health-insurance claim work? Marketplace issuers ranked by the share of appeals that were OVERTURNED in the patient’s favour, plan year ${PLAN_YEAR}. Answers “is it worth appea
• insurer_denial_profile — Everything reported for ONE named health insurer: claims received and denied in and out of network, the denial rate, and both internal and external appeal outcomes, for plan year ${PLAN_YEAR}. Use whe
• insurer_denial_coverage — What this data covers before you rely on it: the plan year, how many issuers and states are present, which states are included, and the counts of issuers whose denial or appeal figures CMS suppressed.