Every denial letter you have ever received traces back to the same three words: medical necessity. Not coding errors, not clearinghouse hiccups, not "the payer lost it." Those are symptoms. The disease is almost always a test that was ordered, performed, and billed without a defensible clinical reason attached to it in the right place, in the right format, before the specimen ever hit the analyzer.

Labs that internalize this stop treating medical necessity as a billing department problem and start treating it as an operations problem. Because that is what it is.

What "medical necessity" actually means to a payer

Medical necessity is not a vibe. It is a specific, documentable assertion: this test was reasonable and necessary for the diagnosis or treatment of this patient's condition, at this point in their care. Medicare codifies this through National Coverage Determinations (NCDs) and Local Coverage Determinations (LCDs) published by the Medicare Administrative Contractors. Commercial payers layer their own medical policies on top, and while they are not identical to Medicare's, they borrow the same architecture.

The practical mechanism is coverage-by-diagnosis. An LCD or payer policy lists the CPT codes it governs and then enumerates the ICD-10 diagnosis codes that support coverage for those CPT codes. If the diagnosis on the claim is on the list, you are inside the guardrails. If it is not, the test is presumed not medically necessary until proven otherwise — and the burden of proof is on the lab, not the payer.

This is the single most misunderstood point in laboratory billing. The ordering provider's clinical judgment does not override the coverage policy. A physician can be completely correct that a patient needs a test and the claim can still deny, because "medically necessary in reality" and "medically necessary as documented against the applicable policy" are two different things. Your job as a lab is to close that gap.

The ICD-10-to-CPT linkage is where labs bleed

Here is the failure mode we see constantly. A provider orders a panel. The requisition has a diagnosis code — maybe Z00.00, "general adult medical exam without abnormal findings," or R68.89, or nothing at all. The panel contains a CPT code that requires a specific covered diagnosis. The two do not match. The claim goes out anyway. Thirty to sixty days later, it denies. Now someone has to chase the provider for a corrected diagnosis, resubmit, and hope it lands inside timely-filing limits.

That entire cycle is preventable at the point of order. The diagnosis-to-procedure linkage either exists at order entry or it becomes a collections problem later. There is no third option. Labs that push this validation upstream — flagging a mismatch on the requisition before the accession is even created — convert a 45-day denial-and-appeal loop into a five-second conversation with the ordering office.

The cheapest denial to fix is the one that never happens because the order was correct before the specimen was drawn.

We wrote more about where this revenue quietly disappears in our piece on revenue leakage — most of it is not fraud or big-ticket errors, it is thousands of small, correctable linkage failures.

Documentation lives in the chart, not on the claim

A clean claim with a covered diagnosis is necessary but not sufficient. When an auditor comes — and in high-scrutiny areas they do come — they do not look at your claim. They ask for the medical record. The documentation that supports medical necessity has to exist in the ordering provider's chart: the clinical indication, the relevant history, the signs and symptoms, and the physician's intent to use the result to manage the patient.

This creates an uncomfortable dependency. The lab is financially responsible for documentation it does not control and often cannot see. You performed the test correctly, you billed it correctly, and you can still lose the money on audit because the referring provider's note said "routine screening" instead of describing the actual clinical picture. The best-run labs treat their referring providers as partners in this, not just as a source of orders — they educate ordering offices on what supportable documentation looks like, because a denied claim hurts the lab far more than it hurts the physician who wrote the order.

Panels, reflex testing, and standing orders — the danger zones

Three ordering patterns generate outsized denial and audit risk:

Overbroad panels. A panel is a convenience, not a clinical judgment. When a panel bundles a dozen analytes and only three are supported by the patient's diagnosis, you are billing nine tests that will be scrutinized. Panels have to be built to match real ordering patterns and real coverage, not to maximize the number of billable components.
Reflex testing without a documented protocol. Reflex testing — automatically running a confirmatory or follow-on test when an initial result crosses a threshold — is legitimate and often good medicine. But the reflex has to be driven by a documented, medically justified algorithm that the ordering provider has authorized. A reflex that fires automatically to generate additional billable tests without clinical logic behind it is exactly what auditors are hunting for.
Standing orders that never expire. A standing order for recurring testing needs a defensible frequency and a real clinical rationale for each interval. "The patient is on our monthly panel" is not medical necessity. Each recurrence has to stand on its own.

Guardrails beat good intentions

You cannot solve medical necessity with training memos and hope. Providers are busy, requisitions get filled out fast, and the person entering the order is frequently not the physician. The only durable fix is to build the rules into the workflow so the correct behavior is the path of least resistance.

That means order-entry validation that checks the diagnosis against the coverage policy for each test in real time. It means an Advance Beneficiary Notice (ABN) workflow that triggers automatically when a Medicare test is likely to be non-covered, so the patient is informed and the lab is protected. It means reflex rules encoded as explicit, auditable algorithms rather than tribal knowledge in one tech's head. And it means every one of these decisions leaves a timestamped record, so when the audit comes you can reconstruct exactly why each test was performed.

This is the difference between a lab that operates from a system and a lab that operates from a stack of faxed requisitions and a billing team playing defense. The system-based lab catches the mismatch at order entry. The fax-based lab catches it in an appeal, forty-five days late, if it catches it at all.

Key takeaways

Medical necessity is a documentation problem, not a science problem. The test can be clinically perfect and still deny.
The ICD-10-to-CPT linkage is your highest-leverage control. Validate it at order entry, not on appeal.
Supporting documentation lives in the referring provider's chart. Educate your ordering offices — their notes are your audit defense.
Panels, reflex rules, and standing orders are audit magnets. Each billed component needs its own justification.
Build the rules into the workflow. Software guardrails convert 45-day denials into five-second corrections.

Medical necessity is not a compliance burden bolted onto the real work of running a lab. It is the real work — the discipline that determines whether the tests you run turn into revenue you keep. If you want to go deeper on how disciplined ordering protects margin, read the toxicology-specific playbook in documenting medical necessity for drug testing, and see the LabX test menu to understand how coverage-aware ordering looks in practice. Build the guardrails now, before the next denial batch teaches you the same lesson the expensive way.

Medical Necessity Is the Whole Ballgame: What Labs Still Get Wrong