An independent read of a public record

When a California health plan refuses a treatment as not medically necessary or as experimental, the enrollee can ask the state for an Independent Medical Review. A physician who does not work for the plan decides, and the plan is bound by the result.

Those decisions are public because Health & Safety Code § 1374.33(g) requires the Department to publish them, and each one carries a written summary of the standards and evidence behind it because § 1374.33(h)(1)(K) requires that too. We did not obtain this record by asking nicely or by scraping anything: the state publishes it, and the same statute strips the health plan’s name from every decision before it appears.

Reading reviewers’ reasoning by machine is the part that can quietly go wrong, so every category is scored against a 250-decision hand-read before it is used. The results, in full:

A category that failed is absent, not shown with a warning next to it. Every score, passed and failed, is on the methodology page, including the two rules we withdrew after the evidence showed they were measuring the wrong thing.

• No percentage below 30 cases

  With fewer than 30 decisions a rate moves several points every time one case changes. Those groups show their decisions and no percentage. It is applied before we look at the numbers, not after.

• No prediction about anyone's case

  This describes what reviewers decided in groups of past disputes. It is not medical advice, not legal advice, and it cannot tell you what will happen to your appeal.

• The same product to both sides, at the same price

  Patients, provider appeal teams and health plans get identical figures on identical terms. A record that only serves one side stops being a record.

• No health plan named against its reasoning

  California removes the plan's name before publishing, so this cannot be built from the source at all — and we do not attempt to infer one. There are no insurer league tables here.

• Nothing quoted that could identify a person

  The reviewers' written decisions stay in our build database. What you see are counts and cited guidelines. Every one of those decisions is a real person's medical history.

• Corrections in public, with a date

  If a figure here is wrong, it gets fixed and logged where anyone can see it.

Every figure here counts denials that were appealed all the way to independent review. Fewer than 1% of denials are appealed internally at all, so “65.2% overturned” is not the share of denials that are wrong — it is the share of contested denials that reviewers did not sustain. It also covers California only, and only plans the state regulates: not self-funded employer plans, Medicare or Medicaid.