The Ghost Rate Report.
Payers publish rates under clinicians’ names for procedures those clinicians never perform. We are counting every one of them, and the count publishes as Edition 001. No number appears before the frozen methodology run completes. That is the point of this report.
A published rate is not proof of a real contract line.
The federal Transparency-in-Coverage mandate makes every payer publish its negotiated rates, per NPI, per code. The files arrive at scale, and they arrive polluted. A rate can be printed under a clinician who could not have negotiated it.
Every rate, per NPI, per code.
Transparency-in-Coverage requires each payer to publish the negotiated rate it holds for each provider and each procedure. The intent is disclosure. The output is a machine-readable file the size of a small country’s tax record.
Rates for procedures no one did.
Those files arrive with rates attached to implausible procedures under real clinicians: line items for work the named provider does not and cannot perform. The rate is in the file. The contract line behind it is not.
A filing artifact, not a market fact.
Illustration, not a measurement: a psychotherapist with published rates for joint injections. That is an artifact of how the file was assembled. Reading it as a real market rate is the mistake the ghost report exists to correct.
The example above is a hypothetical framing of a filing artifact. It is not a statistic and no count is implied by it.
The count ships with the method, never before it.
The methodology is fixed and public before a single ghost is counted. It cannot be tuned to produce a headline. Each stage below has a permanent anchor so any finding can be traced to the rule that produced it.
The Ghost Rate Method
Six stages, applied in order to the full Transparency-in-Coverage corpus. A rate is only called a ghost when it survives every stage. Nothing is estimated; a stage that cannot resolve a rate leaves it uncounted rather than assumed.
#1 · Taxonomy-plausibility mapping
Every clinician in the registry carries an NPPES taxonomy. We map each taxonomy to the code families it can plausibly bill. A published rate for a code outside its provider’s plausible families is a candidate ghost, not yet a confirmed one.
Rule
NPPES specialty → plausible CPT and HCPCS families. Candidate = a published NPI×code pair where the code family falls outside the provider’s taxonomy. Mapping is authored, versioned, and published with the edition.#2 · Never-billed verification
Implausible is not the same as impossible. We check each candidate against Medicare utilization: has this code ever been billed by this provider, or by any provider in this specialty at real volume. A code with no billing footprint is a stronger ghost.
Rule
Candidate NPI×code tested against CMS utilization. Never-billed by the provider, and negligible within-specialty volume nationally, both raise confidence. Present billing history clears the candidate.#3 · NCCI edit sanity
National Correct Coding Initiative edits define which code pairings and contexts are valid. A rate that only exists inside a bundling or mutually-exclusive edit is a filing artifact by construction, not a negotiated line.
Rule
Candidate cross-checked against NCCI procedure-to-procedure and medically-unlikely edits. A rate that cannot survive its own coding edits is disqualified from being a real contract line.#4 · Per-payer league table
Ghosts are not evenly distributed. The edition will rank payers by the density of ghost rates in their published files, so the count attaches to who filed it. The plan is fixed now; the table fills only when the run completes.
Plan
Confirmed ghosts aggregated per payer, normalized by that payer’s published NPI×code volume. Ranked as a league table. No payer is named on this page before the frozen run produces its row.#5 · Independent re-verification before print
The top findings are re-verified by a second pass that does not share the first pass’s inputs, before anything is printed. A finding that does not reproduce does not ship.
Rule
The highest-impact ghosts are re-derived independently ahead of publication. Reproduction is a gate, not a courtesy. A non-reproducing finding is held out of the edition.#6 · Publication into the record
The edition publishes as a public comment in the Transparency-in-Coverage rulemaking record. The count does not just describe the filings; it enters the file about the filings.
Rule
Edition 001 is submitted, with its full methodology, as public comment on the TiC rule. The number and the method are one document. Neither is released without the other.When it publishes. Edition 001 publishes when this run completes against the full corpus. The number ships frozen with the methodology, never before it. Until then this page carries the method and the corpus it runs against, and no count.
What the count runs against.
These are not ghosts. They are the corpus the method will run across: the rate pairs indexed and the providers in the registry the mapping resolves against. The ghost count is derived from this, and it is not shown until the run completes.
The corpus under audit
LiveCount your own ghosts now.
You do not have to wait for the edition to see your own filings. The Mirror runs the same registry join on a single NPI and returns the published-rate docket filed under your name.
The Mirror
Put an NPI in. Get the published-rate docket out: every rate a payer prints under that number, per code, per payer. No signup. It is the loop the report closes at national scale, run one provider at a time.
Run your NPI in the MirrorThe share card carries counts, never dollars.
What the Mirror shares is a count of the docket, not the rates inside it. The report and its companion both refuse to leak the number ahead of the edition. Counts travel; dollars stay in the docket.
Documented reimbursement opportunity, modeled not guaranteed. No PHI required.
Edition 001, when it ships.
Reporters and researchers can arrange access under embargo. The edition is payer-independent and citable, and it ships whole: number, method, and cuts together.
What Edition 001 will carry
- The per-payer league table of ghost density
- The full methodology, frozen at v1.0
- State cuts of the count
No figure from the edition is released before the frozen run completes.
Republish terms
The count and the method are free to republish with attribution required: cite Reddenda, The Ghost Rate Report, and link the permanent page. Do not detach a number from its methodology.
Embargo contact
David@Reddenda.com
As cited in
the ledgerBring us a payer file. We will read it.
Publishers, provider groups, and researchers who want the audit run against a specific corpus, or want Edition 001 under embargo, start on a call.