Medical Necessity Denials: The Coverage Policy, Not the Care

Payers do not decide medical necessity claim by claim from first principles. They publish coverage policies — written criteria that say, for a given service, the conditions under which the plan considers it necessary and payable — and adjudicate the claim against them. A medical-necessity denial is the outcome when the claim, as submitted, does not meet those criteria. On the remittance it usually arrives as CARC 50, the reason code for a service the payer does not consider a medical necessity, often with a remittance remark pointing to the specific policy behind the decision.

The distinction that matters most is that this is a coverage standard rather than a clinical judgment. The payer is not overruling the treating clinician about what the patient needed; it is deciding whether the claim and the record together satisfy the policy the plan adjudicates against. The care can have been exactly right and the claim still fall outside the policy — most often because the reported diagnosis does not support the service under the criteria, or because the record does not carry the specific finding the policy asks for. What is being questioned is the description of the service, not the service.

It is also worth separating from two denials it is easily confused with, because each has a different answer. A non-covered service denial says the plan never covers this service for anyone — the policy question is not whether it was necessary but whether it is a benefit at all. An experimental-or-investigational denial (CARC 55) says the service is not yet established enough to cover. A medical-necessity denial is the middle case: a service the plan does cover, judged not necessary as reported for this patient. Reading the reason code and remark closely is what tells them apart, and it is the same skill set out in reading a denial.

Before deciding whether to fight the denial, settle who owns the balance, because the answer decides whether a patient can be billed at all. The rule is one this cluster returns to often: it is the group code, not the reason code, that assigns responsibility. A medical-necessity denial can carry a contractual-obligation group code, which makes it a provider write-off, or a patient-responsibility group code, which makes it billable — and the same CARC 50 can appear either way.

On Original Medicare, which of the two applies is not a matter of discretion; it is set by statute. A service denied as not reasonable and necessary under Social Security Act §1862(a)(1)(A) falls under the limitation-on-liability rule of §1879, which turns on knowledge: the beneficiary can be held responsible only if they knew, or could reasonably have been expected to know, that Medicare would not pay. In practice that knowledge is established by giving the patient an Advance Beneficiary Notice of Noncoverage (ABN) before the service. With a valid ABN on file, the denied amount can be billed to the patient and generally posts as patient responsibility; without one, the practice cannot collect from the beneficiary, must refund anything already collected, and absorbs it as a contractual write-off.

Commercial plans are governed by the contract rather than by statute, and there is no single rule: some agreements make a medical-necessity denial a provider write-off, others allow it to be billed, and many condition billing on a signed acknowledgment comparable to an ABN. The durable point is the same across payers — read the group code the payer actually returned, then read the contract, before assuming the amount can go to the patient.

Because a medical-necessity denial is the claim measured against a written policy, the response cannot start until that policy is in hand. There is no general medical-necessity standard to argue in the abstract; there is only the specific document the payer adjudicated against, and it names the conditions, the supporting diagnoses, and the documentation the plan expects. Where to find it depends on the payer.

Medicare — a National or Local Coverage Determination

For Medicare, coverage of a service is set either by a National Coverage Determination (NCD), which applies everywhere, or by a Local Coverage Determination (LCD) issued by the Medicare Administrative Contractor for its jurisdiction — so the criteria for the same service can differ by region and change over time. How the two relate, and how they operationalize the §1862(a)(1)(A) reasonable-and-necessary standard, is the subject of national and local coverage determinations.

Commercial — the payer's own medical policy

Private payers publish their own medical or coverage policies, the commercial analog of an NCD or LCD, defining when they consider a service medically necessary. Each payer maintains its own, they differ from one another, and they are revised on each payer's own schedule — so the only policy that matters is the current one for the plan that denied the claim.

The reported diagnosis is what the policy is tested against

Most coverage policies turn on the link between the service and the patient's condition: they list the circumstances — commonly expressed through ICD-10 diagnoses — under which the service is covered. A claim can draw a medical-necessity denial because the reported diagnosis does not support the service under the policy, even when the record does. That is why the first read of the policy is often a read of the diagnosis reporting.

With the policy in hand, the claim splits into two very different cases, and mistaking one for the other is where time is lost. Either the record already meets the policy and the claim simply failed to report it correctly, or the claim reported the case accurately and the payer's determination is what is in dispute. The disposition follows entirely from which one it is.

When the case is a genuine coverage dispute — the claim was right and the record supports necessity — the appeal is a documentary argument, and its shape is set by the policy rather than by how strongly anyone believes the service was needed. The task is to show, point by point, that the record satisfies the criteria the payer applied.

1. 01

   Pull the exact policy the payer cited

   Start from the remark on the remittance and retrieve the specific NCD, LCD, or commercial policy — the version in force on the date of service, which is not necessarily the current one. The appeal is measured against that document, so working from any other version undermines it before it is read.
2. 02

   Map the record to the criteria, not to the conclusion

   Take each condition the policy states and point to where the record establishes it — the finding, the failed prior treatment, the clinical indication the policy asks for. An appeal that asserts the service was necessary without tying the record to the criteria gives the reviewer nothing to act on; one that walks the policy line by line does.
3. 03

   Attach the documentation, do not summarize it

   The supporting record goes in with the appeal — the note, the results, the order — not merely referenced. A reviewer decides on what is in front of them, and a medical-necessity appeal that describes documentation it does not include is an appeal the reviewer cannot grant.
4. 04

   Use a peer-to-peer where the payer offers one

   For a clinical dispute, a peer-to-peer review between the treating provider and a payer clinician can resolve a denial faster than a written appeal, where the payer makes one available post-service. Whether it comes before or during the appeal, and what it can and cannot settle, is payer-specific.

If the first-level decision upholds the denial, the case does not stop there — it moves up a defined ladder of review that differs between Medicare and commercial plans. Which body decides each level, and where a case goes when the payer keeps saying no, is the subject of the levels of appeal. What to gather before filing, so the argument goes in complete the first time, is laid out in the denial appeal readiness checklist.

A medical-necessity denial is expensive to work and often avoidable, and almost none of them originate in billing. They are produced upstream — at the point where a service is ordered, documented, and coded — which is where prevention has to live.

The coverage policy was not checked before the service

The most preventable source. When a service has a known NCD, LCD, or payer policy, checking it before the service tells the practice what the record and the diagnosis will need to show. On Medicare, it is also what tells the practice whether an ABN is warranted so the patient can make an informed choice.

The record did not capture what the policy asks for

A service can be necessary and the note still omit the specific finding the policy requires. The gap is between what the clinician knew and what the record shows, and it closes at the point of documentation — long before the claim.

The diagnosis reporting did not reflect the record

Where the record supports necessity but the claim reports a diagnosis that does not, the denial is a translation error, not a coverage problem. Reporting the diagnosis that the record actually establishes is what prevents it.

Medical necessity is one of the recurring categories mapped in why claims get denied, and like most of them the controls that stop it are the front-end ones set out in preventing denials — here, checking coverage and documenting to it before the claim is ever built. The rest of this cluster is indexed on the Denials & Appeals pillar.