Guide10 min readUpdated August 24, 2026

How to document medical necessity in behavioral health

The four elements payers look for when testing medical necessity in behavioral health, the language that satisfies them, and the phrases that quietly undermine an otherwise strong note.

Clinician reviewing documentation and insurance paperwork

The short answer

Medical necessity in behavioral health is documented by showing four things in the record: a covered diagnosis, current symptoms with severity and frequency, functional impairment in daily life, and a clinical rationale explaining why this service at this intensity is required. Missing any one of the four is the most common reason behavioral health claims are denied on review.

Key takeaways

  • Medical necessity is an argument, not a checkbox. The chart has to make it in writing.
  • Four elements carry it: diagnosis, symptoms, functional impairment, and clinical rationale.
  • Functional impairment is the element clinicians most often leave implicit, and reviewers most often cite.
  • Continued-stay justification is a separate argument from initial necessity, and needs its own language.

Medical necessity is the single most cited reason behavioral health claims fail on review. Not fraud, not coding errors, not missing signatures. The note simply never argued that the care was needed.

This surprises clinicians, because to them the necessity is obvious. They sat with the client. They watched the panic attack. The problem is that necessity obvious in the room is invisible on the page unless someone writes it down, and reviewers only ever see the page.

The four elements

Payer definitions vary in wording, but they converge on the same four questions. A note that answers all four is defensible almost anywhere in the United States.

1A covered diagnosis

Coded, current, and consistent with what the rest of the chart describes. A diagnosis that appears only on the claim and nowhere in the clinical record is a finding waiting to happen.

2Symptoms with severity and frequency

Not 'anxious' but 'panic episodes three to four times weekly, lasting 20 minutes, with anticipatory avoidance in between.' Severity and frequency are what separate a diagnosis from a description.

3Functional impairment

What the symptoms stop the client doing. Work, school, parenting, self-care, relationships, safety. This is the element most often left implicit and most often cited when a claim is overturned.

4Clinical rationale for this service

Why individual therapy weekly, rather than monthly, or a group, or nothing at all. The rationale connects the impairment to the intervention and the intensity.

The impairment sentence

If you write only one extra sentence per note, make it the impairment sentence: what the client cannot currently do because of these symptoms. It is the shortest path from a note that describes care to a note that justifies it.

Language that works, and language that does not

Symptom documentation
Weak

Client reports ongoing depression and low motivation.

Defensible

Client reports depressed mood most days over the past three weeks, sleeping 4 to 5 hours nightly, and has not left the apartment except for appointments in 11 days. PHQ-9 administered today: 19, up from 14 at intake.

Rationale for continued treatment
Weak

Client will continue weekly therapy to work on coping skills.

Defensible

Weekly frequency remains indicated. Symptom burden has not reduced across four sessions, the client remains unable to sustain employment, and reduced frequency at this stage risks decompensation given a prior hospitalization in the same pattern.

Initial necessity and continued-stay necessity are different arguments

Most clinicians document the first well and the second not at all. Establishing that someone needed treatment in March is not the same as establishing that they still need it, at this intensity, in September.

Continued-stay justification has to answer a harder question: given the progress documented so far, why is this level of care still the right one? There are only a few honest answers, and each needs to appear in writing.

  • Symptoms persist at a level that continues to impair function, with current evidence.
  • The client is progressing, and the current intensity is what is producing the progress.
  • Risk factors require the current frequency to monitor safely.
  • A step down was attempted or considered, and there is a clinical reason it is not yet appropriate.

Notice that 'the client finds the sessions helpful' is not on that list. It is true, it matters clinically, and it will not survive a utilization review on its own.

Common failure patterns

PatternWhy it failsFix
Copy-forward notesIdentical language week to week suggests no reassessment occurredChange at least the symptom and impairment detail each session
Diagnosis without criteriaA code with no supporting findings in the recordReference the specific findings that met criteria at least at intake and on any change
Progress with no measure'Improving' cannot be auditedAnchor to a scale, a frequency count, or an observable behaviour
Plan never updatedSuggests care is running on autopilotReview and date the plan on the cadence your payer requires
Impairment only at intakeNecessity is assessed per date of serviceCarry a current impairment statement into each note

Where to put it in the note

Format does not matter as much as presence. In SOAP, symptoms and impairment live in Subjective and Objective, and rationale lives in Assessment and Plan. In DAP, all four have to fit into Data and Assessment. In BIRP, the Response and Plan sections carry most of the weight. What matters is that a reviewer can find all four without hunting.

Frequently asked

Medical necessity means the service provided is clinically required to diagnose or treat a covered condition, at an appropriate level of care. In behavioral health it is demonstrated through a covered diagnosis, current symptoms with severity and frequency, documented functional impairment, and a written rationale for the service and its intensity.

See what a payer would flag.

Run your first clinical review in about two minutes.