How to Write a Behavioral Health Treatment Plan That Holds Up in an Audit

A behavioral health treatment plan is the documented clinical roadmap for a client’s care: the diagnosis, the measurable goals, the interventions tied to named clinicians, and the scheduled reviews that show the plan is current. It’s also one of the first documents a payer or surveyor turns to when reviewing whether care was medically necessary. This guide covers the required elements, review intervals, SMART goal anchors, and how the plan connects to your behavioral health clinical documentation standards.

TL;DR: There’s no federal 90-day rule for outpatient behavioral health treatment plans. Medicare sets the interval by level of care: 30 days in PHP, 60 in IOP. Everything else comes from your state Medicaid rules and payer contracts. Reviewers look for one traceable thread: diagnosis, a scored baseline, a measurable goal, the intervention, and the note that shows movement.

Key Takeaways

• Review cadence comes from your payer, not a national rule. Medicare requires recertification at least every 30 days in partial hospitalization and 60 days in intensive outpatient (42 CFR 424.24). Routine outpatient intervals are set by state Medicaid and payer contracts.
• An auditable goal names four things: a validated instrument, a numeric target, a date, and an owner. “Reduce PHQ-9 from 18 to 10 or below within 8 weeks, therapist” survives review. “Improve mood” doesn’t.
• Medical necessity needs three linked elements: a specific ICD-10 code, a dated functional-impairment statement, and at least one scored measure. Reviewers deny continued care when one of the three is missing, or when all three are present but don’t reference each other.
• The golden thread is what a reviewer follows. Every progress note should cite the goal it addresses and show the score change behind it, so the chart reads as one continuous record.
• Check your instruments and codes. CRAFFT, with two Fs, is the adolescent screener, not CRAFT. F50.01 stopped being billable on 1 October 2024. The ASAM Criteria 4th Edition renumbered its dimensions, so old dimension numbers no longer map.

What a behavioral health treatment plan must document

A compliant plan records eight things. Each has a home in the chart, and reviewers look for them in roughly this order.

Element	Why it matters	Audit-ready example
ICD-10 diagnosis	Establishes clinical need and payer coverage	“Primary: F33.1, major depressive disorder, recurrent, moderate. Confirmed 2026-05-03 by [clinician].”
Problem and functional impairment	Connects symptoms to daily function, which drives level of care	“Insomnia interfering with work 5 of 7 nights. Three missed shifts in the past month, confirmed by attendance record.”
Measurable goals	Shows progress and justifies continued care	“Reduce PHQ-9 from 18 to 10 or below within 8 weeks.”
Interventions with named owners	Shows planned care and who’s accountable	“Individual CBT weekly, [clinician name], LCSW.”
Outcome measures and schedule	Makes progress objective	“PHQ-9 at intake, then biweekly.”
Risk and safety planning	Required when indicated, and routinely examined on survey	“Safety plan completed and reviewed 2026-05-03. Crisis contacts confirmed.”
Review date and signatures	Creates the legal and authorization trail	“Reviewed 2026-06-01. Continued clinical instability. Signed [clinician].”
Discharge or transition criteria	Defines the endpoint and supports step-down decisions	“PHQ-9 at or below 10 sustained 4 weeks, safety plan in place, outpatient follow-up scheduled.”

The AHRQ Integration Academy publishes a behavioral health treatment plan template that maps closely to this structure. It’s public, free of vendor framing, and a useful reference when you’re rebuilding your own form.

Who owns the plan

The clinician who completes the initial assessment usually drafts the plan. A supervisor or interdisciplinary team reviews and co-signs it. The organization owns the record, and the clinician keeps the plan current as care changes.

How often do treatment plans have to be reviewed?

Review intervals are set by level of care and by payer, not by one national rule. Medicare requires physician recertification at least every 30 days in partial hospitalization and at least every 60 days in intensive outpatient. Community mental health centers review at least every 30 calendar days. For routine outpatient care there’s no federal interval at all.

What is the 3 month rule in mental health?

The 90-day cadence most programs run on is real practice, but it isn’t federal law for outpatient behavioral health. Two genuine federal 90-day rules exist and neither one applies here: Medicaid inpatient mental hospital plans of care (42 CFR 456.180(c), which applies to Medicaid mental hospitals), and outpatient physical, occupational and speech therapy recertification. The outpatient behavioral health convention comes from state Medicaid regulations and payer contracts, which vary by state and by plan.

Here’s what the rules require:

Setting	Requirement	Citation
Medicare partial hospitalization (PHP)	Physician recertification at day 18, then at least every 30 days	42 CFR 424.24(e)(3)(ii)
Medicare intensive outpatient (IOP)	Recertification at least every 60 days	42 CFR 424.24(d)(3)(ii)
Community mental health center	Plan within 7 working days, team review at least every 30 calendar days	42 CFR 485.916(d)
Psychiatric hospital	No fixed plan-review interval. Progress notes at least weekly for 2 months, monthly after	42 CFR 482.61(c) to (d)
Medicaid inpatient psychiatric, under 21	Team review every 30 days	42 CFR 441.155(c)
Opioid treatment program	Care plan within 14 days, updated as treatment response changes. No fixed interval	42 CFR 8.12(f)(4)
Routine outpatient behavioral health	No federal interval. Set by state Medicaid and payer contract	Varies

Accreditors don’t publish a number either. The Joint Commission states that how the plan of care is documented is up to each organization, that the required elements can live across the record including progress notes, and that compliance is judged against the organization’s own policies and procedures. No published Joint Commission standard sets a review interval. CARF requires periodic review appropriate to the person served. [Verify CARF’s current wording against the CARF Behavioral Health Standards Manual before publishing. That manual isn’t public, so this line is not independently sourced.] Both bodies expect you to define the interval, write it into policy, and then meet it. Survey citations come from missing your own stated interval.

Set your cadence from the strictest applicable source, usually your state Medicaid rule or your tightest payer contract, then let the EHR enforce it. See what CARF looks for in accreditation and how The Joint Commission surveys behavioral health programs for what each body examines.

Triggers that force a review early

Scheduled reviews are the minimum. These events open the plan wherever you are in the cycle:

• New or changed diagnosis
• Level-of-care change in either direction
• Medication change, including starting or stopping addiction medications
• Stalled progress against an active goal
• Safety event: overdose, hospitalization, suicide attempt
• Client request

Minor edits and major revisions

A minor revision clarifies wording, adjusts an intervention, or tweaks a goal without changing intent. A major revision changes the diagnosis, the level of care, the primary goals, or adds a treatment modality. Major revisions need documented clinical rationale, new measurable goals, a clinician signature with date, and any family or collateral involvement recorded.

Version the plan so a reviewer can reconstruct the sequence. A short convention works: sequential number, date, author initials. Keep prior versions read-only but retrievable, and reference the version in the progress note that accompanies the change, for example “updated to v2.0, new goals following hospitalization.” How EMR audit trails work covers the underlying mechanics.

What is the golden thread in clinical documentation?

The golden thread is the traceable line a reviewer follows from assessment to discharge. The assessment identifies a problem, the problem produces a measurable goal, the goal produces an intervention, and every progress note references that goal and shows movement against it.

When the thread breaks, the chart stops supporting medical necessity even though the care was delivered. It usually breaks in one of two places. Either notes stop citing goals and turn into narrative, or the plan never gets updated after a level-of-care change, so the notes describe care the plan doesn’t authorize.

Three habits keep it intact:

• Store goals as discrete records with an ID, so notes can reference them.
• Require a goal reference in every progress note template.
• Update the plan when care changes, without waiting for the next scheduled review.

A note that holds the thread looks like this:

Client reports decreased cravings. PHQ-9 12, down from 18 on 2026-05-03. Mood improved, medical necessity for continued medication management supported. Continue buprenorphine and weekly counseling. Addresses goal G-2026-007.

Four things are present: the subjective report, the scored change with dates, the clinical judgment, and the goal reference. Writing progress notes that reference a goal goes deeper on note structure, including BIRP (Behavior, Intervention, Response, Plan) and SOAP (Subjective, Objective, Assessment, Plan) formats.

How to write SMART treatment plan goals

A goal is auditable when progress can be measured without interpretation. The SMART frame, meaning specific, measurable, achievable, relevant and time-bound, gets you there if you attach a real instrument to it.

Four steps:

1. Name the domain and pick an instrument. “Low mood” becomes depression measured by PHQ-9. “Wakes at night” becomes sleep measured by the Insomnia Severity Index.
2. Set a numeric anchor and a measurement schedule. A point change, a frequency count, or a percentage, plus how often you’ll check.
3. Check the target against baseline. Size the change to where the client actually started. An unmet target looks worse in review than a conservative one you hit.
4. Attach a timeframe and an owner. Both make the goal auditable and make follow-up somebody’s job.

Before and after

Domain	Vague goal	Auditable goal
Mood	Reduce depression	Reduce PHQ-9 from 18 to 10 or below within 8 weeks, checked biweekly by the therapist
Anxiety	Worry less	Reduce GAD-7 by 4 points within 6 weeks and complete 6 CBT sessions
Sleep	Sleep better	Reduce ISI by 6 points and reach 6 nights per week with at least 6 hours sleep, by week 6
Substance use	Use less	Maintain at least 21 of 30 substance-free days, confirmed by weekly urine drug screen
Substance use (harm reduction)	Cut back drinking	No more than 4 drinking days per month and no session above 3 drinks, over 12 weeks, logged daily
Functioning	Return to work	Attend 5 of 5 scheduled work days for 4 consecutive weeks, within 12 weeks
Safety	Feel safer	Safety plan reviewed weekly and crisis contacts confirmed by the next session
Medication	Take medication	Adherence at or above 80 percent of doses over 30 days, verified by pharmacy refill data
Coping skills	Improve coping	Use a coping skill during urges in at least 80 percent of episodes over 4 weeks, logged in the skills diary

Format goals for picklists

Use one structure so goals stay comparable and reportable: Domain | Measure | Target | Timeframe | Owner. Make the picklist the single source of truth and turn off duplicate free-text entry, which is where copy-paste drift starts.

Explaining a goal to a client

“We’ll track this with a short questionnaire, the PHQ-9. The goal is a 5-point drop over eight weeks, which tells us whether to keep going or change something.”

If a client declines to sign the goal section, document it plainly: what they disagreed with, the rationale you provided, the alternatives offered, and the refusal itself, per policy.

Which outcome measures to collect, and how often

Match instruments to the presenting problem and keep the battery short. Every added instrument costs clinician and client time, and a half-completed measure reads as a gap when the chart is reviewed.

Measure	Domain	Range and anchor	Typical cadence
PHQ-9	Depression	0 to 27. A 5-point reduction is the conventional threshold for meaningful change	Intake, then monthly outpatient, every 2 weeks IOP, weekly PHP when unstable
GAD-7	Anxiety	0 to 21. Commonly cited minimal clinically important difference is 4 points	Same as PHQ-9
AUDIT / AUDIT-C	Alcohol risk	AUDIT 0 to 40, 8 or higher indicates hazardous use. AUDIT-C 0 to 12, 4 or higher for men and 3 or higher for women	Intake, then monthly, weekly in PHP
DAST-10	Drug use severity	0 to 10. 3 to 5 moderate, 6 to 8 substantial, 9 to 10 severe	Intake, then monthly
CRAFFT	Youth substance use, ages 12 to 21	0 to 6. A score of 2 or higher indicates high risk in adolescents	Intake, then monthly in youth programs
Insomnia Severity Index	Sleep	0 to 28. A 6-point reduction is the recommended minimally important difference	Intake, then monthly, weekly in PHP if sleep is a target

Sources: PHQ-9 responsiveness, AHRQ, GAD-7 MCID, Toussaint et al. 2020, AUDIT, NIDA, CRAFFT provider manual.

A note on those thresholds. Published change scores vary by how they were derived. The PHQ-9 5-point figure comes from a distribution-based estimate in an older-adult sample, and anchor-based studies using patient-reported global improvement have produced smaller values. The ISI figure comes from a primary-insomnia drug trial, and other work puts moderate improvement closer to 8 points. Use these as documentation conventions your team applies consistently, and don’t treat any single number as a clinical certainty.

CRAFFT is not CRAFT

The names are close enough that the two get confused in charts. CRAFFT, with two Fs, standing for Car, Relax, Alone, Forget, Family/Friends, Trouble in the current 2.1 version, is a validated screening tool for youth aged 12 to 21. CRAFT, with one F, is Community Reinforcement and Family Training, a family-directed intervention for engaging a treatment-refusing loved one. Charting the wrong one misrepresents what was done.

One more note on CRAFFT: the cut-off of 2 or higher is validated in adolescents. The instrument’s own manual cites studies in 18 to 20 year olds that found an optimal cut point above 3, so don’t apply a single threshold flatly across the whole 12 to 21 span.

Store scores where you can report on them

Put every score in a discrete, named field instead of in note text. Discrete fields let you trend, report and audit without re-reading narrative, and they let the goal reference the score directly.

Measurement-based care is well supported for symptom outcomes when scores get fed back to the clinician close to the encounter. A meta-analysis in Psychiatric Services found meaningful effects against usual care, with the caveat that screening alone doesn’t improve outcomes. Collecting scores without returning them to the clinician won’t move anything.

How do you document medical necessity?

Medical necessity is documented when three elements sit in the record and point at each other: a specific ICD-10 diagnosis, a dated statement of functional impairment in concrete terms, and at least one scored standardized measure. Reviewers deny claims when any one is missing, or when all three are present but unconnected.

Tie symptoms to function, with a source

Translate symptoms into measurable effects on work, school, family, legal, medical or housing functioning, and name where the evidence comes from. Pair them in one sentence:

Daily panic attacks (GAD-7 = 17 on 2026-05-14) prevent the client from driving to work. Missed 60 percent of required shifts in the past 30 days, confirmed by employer attendance record.

Symptom, objective score with a date, functional consequence, corroboration.

Map to level of care using the ASAM Criteria 4th Edition

For substance use cases, document the level-of-care rationale against the ASAM dimensions. The 4th Edition renamed and renumbered the dimensions, so a 3rd Edition dimension number no longer refers to the same construct. The current six:

1. Intoxication, Withdrawal, and Addiction Medications
2. Biomedical Conditions
3. Psychiatric and Cognitive Conditions
4. Substance Use-Related Risks
5. Recovery Environment Interactions
6. Person-Centered Considerations

Readiness to change no longer contributes independently to the level-of-care recommendation, and is integrated across dimensions. If your templates still carry “Dimension 4: Readiness to Change,” they’re on the 3rd Edition and will read as out of date to a reviewer working from the current criteria. The ASAM Criteria dimensions covers how the dimensions get applied, and what changed in the 4th Edition covers the transition.

Withdrawal scales belong in Dimension 1. Define them on first use in your templates: CIWA (Clinical Institute Withdrawal Assessment for Alcohol) and COWS (Clinical Opiate Withdrawal Scale).

Reusable sentences for notes and payer letters

Keep short editable statements on hand so the language stays consistent across clinicians:

• “Primary diagnosis: [ICD-10 code and full description], onset [date].”
• “PHQ-9 total = [score] on [date], indicating [severity] symptoms limiting [specific function].”
• “COWS = [score] on [date] with [signs], consistent with [severity] withdrawal requiring [level of monitoring].”
• “Requested level of care is supported by: [symptom], [objective score with date], [documented impairment], and [time-limited goal].”
• “ASAM dimension findings: [dimension and finding]. Recommended service: [service].”

For a payer letter, mirror the clinical note. Same dates, same scores, same wording. Discrepancies between the letter and the chart invite a second request for information. Billing for mental health services covers the claim side of the handoff.

Get ICD-10 specificity right. A retired or insufficiently specific code can get a claim rejected before anyone reads the clinical detail. One current example: F50.01 stopped being a valid billable code on 1 October 2024. It survives in the tabular list as a non-billable subcategory header, so billing now requires one of the five-character codes carrying a severity or remission specifier, F50.010 through F50.019. Pull codes from a validated picklist your team keeps current against the annual release.

Adapting the plan by diagnosis and level of care

The framework stays constant while the instruments, interventions and discharge criteria change. Short-term crisis stabilization focuses on immediate safety, rapid symptom reduction and daily review. Longer-term recovery planning focuses on function, relapse prevention and monthly or quarterly outcome tracking.

Plan type	ICD-10 example	Example goal anchor	Core disciplines	Review cadence
Major depressive disorder	F33.1, recurrent, moderate	PHQ-9 from 18 to 10 or below in 8 weeks	Psychiatry, psychotherapy, nursing, case management	PHQ-9 weekly, medication review at 2 weeks
Generalized anxiety disorder	F41.1	GAD-7 down 4 points in 6 weeks with CBT	Psychotherapy, psychiatry, occupational therapy	GAD-7 biweekly, function monthly
Substance use disorder	F11.20, opioid dependence, uncomplicated	30 consecutive abstinent days, or documented engagement in addiction medication treatment, within 60 days	Addiction medicine, counseling, case management, nursing	UDS (urine drug screen) weekly, ASAM review at placement decisions
Eating disorder	F50.010 to F50.019, specifier required	Meal plan adherence at or above 90 percent, with weight-restoration targets set individually by the treatment team	Medical, nutrition, psychotherapy, family therapy	Vitals and labs per medical protocol, EDE-Q (Eating Disorder Examination Questionnaire) biweekly
Crisis stabilization	Varies	Safe disposition within 72 hours with a documented follow-up appointment	Acute or emergency team	Daily

The eating disorder row deliberately gives no numeric weight target. The APA practice guideline for eating disorders requires individualized weekly gain and target weight, so a published number would be both clinically wrong and out of step with the governing guideline. That row also shows a code range because the severity specifier has been mandatory since the FY2025 release.

For worked examples at depth, see a worked depression treatment plan and planning for substance use disorder.

A one-page audit rubric

Score a chart 0 to 2 on each item. 0 means missing or noncompliant, 1 means partial or inconsistent, 2 means complete and clinically useful. Maximum 20.

#	Item	Score
1	Diagnosis or presenting problem documented with a current ICD-10 code	0 to 2
2	Objective baseline measures or clinician-rated scores present with dates	0 to 2
3	Client strengths and barriers listed and linked to goals	0 to 2
4	Goals are specific, measurable, achievable, relevant and time-bound	0 to 2
5	Each intervention is tied to a goal and to a named clinician	0 to 2
6	Session frequency, length and modality specified	0 to 2
7	Risk and safety plan documented where indicated	0 to 2
8	Referrals, consents and care-team coordination recorded	0 to 2
9	Progress notes reference active goals and show measured change	0 to 2
10	Discharge or transition criteria stated	0 to 2

How to use it. Run it on a 30-chart sample to establish a baseline. Triage the results: 8 or below is immediate, 9 to 12 is action, above 12 is monitor. For each immediate chart, meet the clinician within 48 hours, rewrite the missing goal together, and resubmit for review within 7 days. Attach a note to the chart describing the coaching, the timeline and the next review date. Repeated noncompliance moves to a formal performance improvement plan.

Common findings and fixes:

• Vague goals. Rewrite from the picklist with a target metric and 30 and 90-day markers.
• No progress measures. Add a baseline instrument and make the score field required.
• Incomplete risk plan. Complete a safety plan addendum, notify the clinical director, schedule same-day follow-up.
• Unsigned reviews. Enforce signature and version capture in the workflow itself.

Preparing for a records audit covers the wider audit picture.

What to require from your EHR

The features that cut treatment-plan admin time are the ones that make structure mandatory instead of optional. When you demo a system, ask which fields can be made required.

Requirement	What it changes
Configurable, accreditation-aligned plan templates	Clinicians start from the right structure instead of rebuilding it
SMART goal picklists as discrete records with IDs	Goals become referenceable, reportable and comparable
Validated instruments that score into discrete fields	No transcription step, and scores can be trended and audited
Review reminders configurable by level of care	Your 30, 60 or contract-specific interval gets enforced
Versioned audit trail with signer and timestamp	You can show who changed what and why during a survey
Role-based routing	Tasks land with the right person without a handoff conversation
Verification of benefits and claims-ready documentation	Eligibility and authorization state is visible before care

Platforms built for behavioral health tend to ship with these workflows already configured. Generic multi-specialty systems can be adapted, though it’s worth asking whether “flexible” means a configuration screen or a development project, and getting the answer before you sign. For a longer look at evaluating options, see our guide to choosing an outpatient behavioral health EHR.

At Alleva, treatment plan goals are stored as discrete records that progress notes reference directly, and validated instrument scores are written into reportable fields instead of note text. InCheck, our compliance and accreditation module was built by a behavioral health compliance specialist around CARF, Joint Commission and state requirements, so review cadences and evidence collection are configured for the standards you’re surveyed against. And Echo, our ambient documentation assistant drafts the note while the clinician stays in the conversation.

Software won’t write the plan. It can make the required structure the default, so nothing depends on memory.

Where to start

1. Run the 10-item rubric on a 30-chart sample and record the baseline.
2. Confirm your review intervals against your state Medicaid rule and your tightest payer contract, then write them into policy.
3. Rewrite the goals in the lowest-scoring charts from a standard picklist.
4. Audit your ICD-10 picklist against the current annual release, starting with the eating disorder codes.
5. Check whether your ASAM templates use 4th Edition dimension names, and update them if not.
6. Schedule weekly quality review rounds and track remediation time to closure.

To see how this works inside a system built for behavioral health, book a demo of Alleva and we’ll walk your team through goal records, review reminders and the audit trail using your own workflows.

Frequently Asked Questions

What is the 3 month rule in mental health?

There’s no federal 3-month rule for outpatient behavioral health treatment plans. The 90-day cadence many programs follow comes from state Medicaid regulations and payer contracts, not federal law. Medicare sets its own intervals by level of care: at least every 30 days in partial hospitalization and every 60 days in intensive outpatient.

What is a behavioral health treatment plan?

A behavioral health treatment plan is a documented clinical roadmap recording the diagnosis, presenting problems and functional impairment, measurable goals with target dates, the interventions assigned to named clinicians, outcome measures, and scheduled review dates. It guides care day to day and evidences medical necessity for payers and accreditors.

What is considered behavioral health treatment?

Behavioral health treatment covers mental health and substance use services across levels of care, from routine outpatient therapy through intensive outpatient, partial hospitalization, residential and medically monitored withdrawal. Treatment plan requirements tighten as intensity rises, which is why review intervals and documentation depth are set by level of care.

What are some examples of treatment plan goals?

An auditable goal names an instrument, a target, a timeframe and an owner. Examples: reduce PHQ-9 from 18 to 10 or below within 8 weeks, monitored biweekly by the therapist; attend 5 of 5 scheduled work days for 4 consecutive weeks; maintain at least 21 substance-free days in 30, confirmed by weekly urine drug screen.

What makes a treatment plan goal audit-ready?

A goal is audit-ready when progress can be measured without interpretation. That means a validated instrument or an observable behavior, a numeric baseline and target, a date, and a named responsible clinician. Every progress note addressing the goal should reference it and show the current value.

Which standardized measures should we collect at intake?

Match instruments to the presenting problem. PHQ-9 for depression and GAD-7 for anxiety, AUDIT or AUDIT-C and DAST-10 for substance use risk, the Insomnia Severity Index when sleep affects functioning, and CRAFFT for adolescents aged 12 to 21. Keep the battery short to limit clinician and client burden.

Is CRAFFT the same as CRAFT?

No. CRAFFT, with two Fs, is a validated screening tool for substance use among youth aged 12 to 21, scored 0 to 6. CRAFT, with one F, is Community Reinforcement and Family Training, a family-directed intervention for engaging a treatment-refusing loved one. The two are frequently confused in documentation.

What should a payer letter include to justify level of care?

Mirror the clinical note. Include the ICD-10 diagnosis with onset date, standardized scores with the dates they were collected, one or two concrete functional-impairment statements with corroborating documentation, ASAM dimension findings for substance use cases, and the time-limited goals and services requested with responsible clinicians named.