The denial prevention field manual: twelve root causes.
First-pass denials run roughly 10 to 12 percent, and a large share, often cited around 65 percent, are never reworked and quietly become write-offs. This is the working manual for the biller, the coder, and the RCM director who own the clean-claim number. One root cause per card, one CARC reference, one front-end control, one KPI.
Executive summaryPrevention is cheaper than appeal.
Every denial has a root cause, and most of those causes are front-end failures that a control catches before the claim leaves the building. Read the CARC, find the control, watch the KPI. The cheapest denial is the one that never happens.
A denial that is never worked costs you the full claim value.
When roughly 65 percent of denials are never reworked, the denial rate understates the damage. The real loss is the abandoned tail. A prevented denial costs nothing to appeal, delays no cash, and never enters the write-off queue. That is the entire argument for moving spend from the back end to the front end.
The decisionEvery denial routes to one of two lanes.
Before you touch a denied claim, decide the lane. Preventable denials should never recur, so the fix is a control, not a rework. Appeal denials were not preventable, so the fix is a clean, evidence-backed appeal. Sorting by lane is how a denial program stops being a treadmill.
of denial volume traces to front-end failures that a control catches
- Eligibility not verified at registration
- Prior authorization missing or expired
- Claim submitted with missing information
- Coordination of benefits order unknown
- Provider not yet credentialed or enrolled
- Filed after the timely filing window
BY
CARC
of denial volume is a genuine dispute that needs documentation, not prevention
- Medical necessity dispute with clinical support
- Non-covered service the patient elected
- Bundling and NCCI edits worth challenging
- Payer processing error on a clean claim
- Downcoding that the record does not support
The centerpieceThe twelve root causes.
One card per root cause. Each carries its CARC reference, the front-end control that prevents it, and the single KPI that tells you the control is holding. Codes below are real CARC references; where a category has no single stable code, it is described without a fabricated number.
Eligibility & coverage terminated
CARC 27 signals expenses incurred after coverage terminated; CARC 26 is expenses before coverage began. The patient was not covered on the date of service.
Prior authorization missing
CARC 197 marks precertification, authorization, or notification absent. The service required an auth that was never obtained or had lapsed.
Non-covered service
CARC 96 flags a non-covered charge under the plan. The service is outside the benefit, not merely unauthorized.
Medical necessity
CARC 50 states the service is not deemed a medical necessity by the payer. This is the classic appeal-lane denial when the record supports the care.
Timely filing
CARC 29 marks the time limit for filing expired. The claim was correct but submitted after the payer window closed.
Duplicate claim or service
CARC 18 flags an exact duplicate claim or service. Often a resubmission sent before the original adjudicated.
Bundling & NCCI edits
CARC 97 flags a service included in another already adjudicated; CARC 236 flags an NCCI procedure-to-procedure conflict. Unbundling or a missing modifier.
Coordination of benefits
CARC 22 is care that may be covered by another payer per COB; CARC 23 is impact of prior payer adjudication. The payer order was wrong or the primary EOB was missing.
Missing or invalid information
CARC 16 flags a claim lacking information or containing a submission error, always paired with a RARC that names the exact missing field.
Not covered by this payer
The plan routes members to a specific network or the benefit sits with a carve-out vendor. Payers signal this with plan-specific CARC and RARC pairs rather than one universal code.
Provider not credentialed or enrolled
CARC 185 flags the rendering provider not eligible to perform the billed service. The provider was not yet credentialed or enrolled with the payer on the date of service.
Coding specificity
CARC 11 flags a diagnosis inconsistent with the procedure; coding-specificity gaps also surface as CARC 16 with a RARC naming the invalid or non-specific code.
The referenceTwelve causes in one table.
The same twelve causes condensed to a single reference row per cause: CARC reference, root cause, the control that prevents it, and the KPI that proves the control is holding. Hand this to the biller.
| CARC ref | Root cause | Front-end control | KPI to watch |
|---|---|---|---|
| 27 / 26 | Eligibility & coverage | 270/271 check at registration and day of service | Eligibility verify rate |
| 197 | Prior authorization missing | Auth-required flag; auth secured before the visit | Auth capture rate |
| 96 | Non-covered service | Benefit check plus signed ABN before service | ABN-on-file rate |
| 50 | Medical necessity | LCD/NCD check and documentation at point of care | Appeal overturn rate |
| 29 | Timely filing | Per-payer filing clock; clear aged-unbilled worklist | Days to submission |
| 18 | Duplicate claim or service | Duplicate scrub and status check before resubmit | Duplicate rate |
| 97 / 236 | Bundling & NCCI edits | NCCI scrubber pre-submission; correct modifiers | Scrubber pass rate |
| 22 / 23 | Coordination of benefits | COB order confirmed; primary EOB attached first | COB-verified rate |
| 16 + RARC | Missing information | Front-end edits enforce required fields | Clean claim rate |
| payer-specific | Not covered by this payer | Plan-and-network check; route carve-outs correctly | Registration accuracy |
| 185 | Provider not credentialed / enrolled | Enrollment roster current before first patient | Enrollment-current rate |
| 11 / 16 + RARC | Coding specificity | Code to highest specificity; validate dx-to-CPT link | Coding accuracy rate |
The playbookThe front-end controls checklist.
Eight controls that sit ahead of claim submission. Run them and the preventable lane empties. Each maps back to one or more of the twelve root causes above.
The scoreboardFour KPIs that prove prevention.
Two leading indicators tell you the front end is clean, one diagnostic points at the broken control, and one lagging financial number tells the board the money is landing. Targets are illustrative benchmarks.
When a single CARC category climbs, one control broke. Go fix the control.
Denial rate by CARC category is the most useful number in the whole scoreboard because it is diagnostic. It does not just tell you denials rose. It tells you which front-end control failed, so the fix is targeted, not a general call to work harder. Pair this manual with the 80 denial patterns reference for the pattern-level detail.
Common questionsFrequently asked: denial prevention.
What is a healthy first-pass denial rate?
What is a CARC code?
Is it better to prevent a denial or to appeal it?
Which denials are the most preventable?
What KPI tells me my denial prevention is working?
Want the twelve causes run against your denials?
Send ninety days of remittance data. Inside 30 days: a denial baseline by CARC category, the two or three front-end controls leaking the most dollars, and a preventable-versus-appeal split with the KPI targets to hold each one. Yours to keep.