ICD-10 Codes in Behavioral Therapy: What Therapists Are Getting Stuck On

Summary Table

Code	Category	What It Covers	Key Coding Note
Z03.89	Assessment	No diagnosis established after an encounter	Use when a first session confirms nothing, not “rule out”
F41.9	Anxiety	Anxiety symptoms, unspecified disorder	Interim code while a specific anxiety disorder isn’t yet confirmed
F41.1	Anxiety	Generalized anxiety disorder	Requires ongoing worry not tied to a single stressor
F41.0	Anxiety	Panic disorder without agoraphobia	Distinct from GAD, needs recurring panic attacks
F32.A	Depression	Depression, unspecified	MDD criteria not yet confirmed
F32.0 to F32.3	Depression	MDD, single episode, mild to severe	No prior episode documented in history
F32.9	Depression	MDD, single episode, unspecified severity	Severity not documented
F33.0 to F33.3	Depression	MDD, recurrent, mild to severe	Any prior episode with a full remission gap counts
F33.9	Depression	MDD, recurrent, unspecified severity	Severity not documented
F30.x	Bipolar	First manic episode, no prior mood history	Not F31 until a pattern is established
F31.x	Bipolar	Bipolar disorder, current episode	Code changes as episode type changes, not locked by history
F31.81	Bipolar	Bipolar II disorder	One flat code, no episode or severity character
F43.2x	Stress and adjustment	Adjustment disorder, by symptom type	Requires an identifiable stressor and full MDD or anxiety criteria not being met
F43.10	Trauma	PTSD, unspecified	Duration not yet established
F43.11	Trauma	PTSD, acute	Symptoms under 3 months
F43.12	Trauma	PTSD, chronic	Symptoms 3 months or longer
F43.81	Grief	Prolonged grief disorder	Grief impairing beyond 12 months
Z63.0	Relationship	Relationship distress with spouse or partner	Secondary code only, needs a supporting F-code
Z63.4	Grief	Death of a family member	Uncomplicated bereavement, not a diagnosis
F90.0	ADHD	Predominantly inattentive presentation	5 symptoms for adults, 6 for under 17
F90.1	ADHD	Predominantly hyperactive impulsive presentation	Same age based threshold applies
F90.2	ADHD	Combined presentation	Most commonly assigned ADHD code
F90.9	ADHD	Unspecified type	Placeholder only while assessment is incomplete
F84.0	Autism	Autism spectrum disorder	Correct code for a DSM-5 ASD diagnosis, any severity
F84.5	Autism	Asperger’s syndrome	Excludes1 against F84.0, cannot be billed together
F10.10	Substance use	Alcohol abuse	Maps to DSM-5 mild alcohol use disorder
F10.20	Substance use	Alcohol dependence	Maps to both moderate and severe alcohol use disorder
F10.180	Substance use	Alcohol abuse with induced anxiety disorder	Use instead of separate substance and anxiety codes

A question that comes up constantly, from both sides of the couch, is why the code on a bill doesn’t match what someone thinks their real diagnosis is. 

A client with a formal ADHD diagnosis sees PTSD and GAD codes on their claim instead and assumes something’s wrong, or that the biller just grabbed whatever was convenient. Therapists ask the mirror version of this just as often. 

Suppose one of your clients clearly has ADHD, but the anxiety and the trauma history are what’s driving this week’s session, so which one goes on the claim.

Both can be true without either being fraud. A diagnosis code documents the condition being treated in that encounter, not a client’s full diagnostic history. 

If a session was spent entirely on trauma processing, GAD or PTSD is the correct code for that visit even with ADHD sitting in the chart as an established, separate diagnosis. What turns this from a coding nuance into a real problem is when the code doesn’t match anything discussed in the note at all. That gap, not the choice between two legitimate diagnoses, is what draws attention during a payer’s review.

“What Do I Put Down Before I Know”

This is probably the single most common practical question in the field, and it comes up at every intake. There’s forty five minutes with a new client, some depression, some anxiety, and no confirmed picture yet. The instinct is to write ‘rule out bipolar’ and move on.

The official guidelines for ICD-10-CM are direct that a diagnosis written as probable, suspected, questionable, or rule out cannot be coded. What’s left instead.

• If the assessment produced no diagnosis at all, Z03.89, encounter for observation for suspected condition ruled out, is the code for that visit. State Medicaid behavioral health programs use it specifically for assessment sessions that don’t result in a diagnosis
• If some symptoms are confirmed but the full clinical picture isn’t, an unspecified code that matches only what’s documented, like F41.9 for anxiety symptoms without a more specific anxiety disorder confirmed, is the right call
• A provisional diagnosis is expected to change and is billable. A rule out is not the same thing, even though clinicians use the two terms interchangeably in casual conversation

“Will Changing My Client’s Diagnosis Trigger an Audit”

This fear shows up constantly in how people talk about diagnosis coding, and it’s worth addressing directly because it changes how people document. A diagnosis given after one session isn’t supposed to be permanent. 

If a client presented with adjustment disorder in week one and by week six the picture clearly supports major depressive disorder instead, updating the code is the correct move, not a red flag. What draws scrutiny isn’t a diagnosis that evolved with more information. It’s a diagnosis that stays static for years on a condition that’s structurally defined by change, or one that shifts without any documentation explaining why.

Bipolar disorder is the clearest example of a diagnosis that’s supposed to move. F31, bipolar disorder, is coded to the current episode, not the lifetime pattern. A client with ten documented depressive episodes who comes in manic today gets a manic episode code that visit, full stop.

• A first ever manic episode, with no prior mood history, is coded under F30, not F31
• The current episode determines the code every visit, whether manic, hypomanic, depressed, or mixed, each with its own severity and psychotic feature branches, listed out on CMS’s site
• Bipolar II breaks this pattern entirely and is coded as one flat code, F31.81, with no episode, severity, or remission character attached at all

Depression runs on the opposite logic, and mixing the two up is a common error. F32 covers a single lifetime episode, F33 covers recurrent. Once any prior episode is documented, even something a client mentions offhand from college years ago, the current episode moves into F33 and stays there. That’s history locking the code in place, the opposite of how bipolar disorder works next to it in the same chapter.

“My Client Doesn’t Meet Criteria for Anything, But They Need Support”

This is the question that pushes a lot of therapists toward a workaround worth naming plainly. Z codes exist for exactly this situation, circumstances affecting health that aren’t a diagnosable disorder, but many practices avoid them because commercial payers frequently won’t reimburse a Z code as the primary diagnosis on a claim. Rather than risk the denial, a common practice pattern is to default to an adjustment disorder code instead, since it’s reimbursable and only requires an identifiable stressor rather than full symptom criteria.

That workaround holds up when the presentation fits adjustment disorder. It becomes a real problem when it’s used purely because the accurate code won’t get paid. This shows up most concretely in couples work. Z63.0, relationship distress with spouse or partner, describes exactly what’s happening in most couples sessions, and it cannot carry a claim by itself.

• One partner needs a documented DSM-5 diagnosis, an F-code such as generalized anxiety disorder or major depressive disorder
• That partner is the identified patient on the claim
• Session notes frame the couples work as treatment for that partner’s diagnosed condition
• CPT 90847 is billed, with Z63.0 added only as a secondary code for context

Grief work runs into a version of the same structural gap. A client whose grief is following an expected course after a loss gets Z63.4, disappearance and death of a family member, which has existed in the code set since ICD-10-CM’s original rollout. But normal grief isn’t a mental disorder, which means straightforward bereavement support has historically sat outside what insurance considers treatable. 

That gap is why F43.81, prolonged grief disorder, was added in 2022, as the APA coverage of that update explains. It gave clinicians a diagnosable code for grief that’s stalled and impairing well past the one year mark, instead of forcing a choice between an unreimbursable Z code and stretching adjustment disorder to fit.

“How Long Does Trauma Have to Last Before the Code Changes”

Trauma coding runs on a third kind of logic entirely, neither episode history like depression nor current state like bipolar disorder, but elapsed time. 

The F43.1 block covers F43.10 for unspecified, F43.11 for acute, and F43.12 for chronic PTSD. The line sits at three months. Under three months of symptoms is acute, three months or longer is chronic, and the clock runs on when the symptom picture met full criteria, not on when the traumatic event happened, which matters for delayed onset presentations where the trauma predates the symptoms by months. 

F43.10 is fine at intake while that duration history is still being established, but once enough sessions have passed to know which side of three months the client is on, staying on the unspecified code stops being appropriate caution and starts looking like an unfinished chart.

Adjustment disorder sits in this same F43 block, and it’s worth naming precisely where it stops applying, since it comes up constantly as a workaround code. DSM-5’s criteria for adjustment disorder specifically exclude it once full criteria for another disorder, like MDD, are met. So the clinical call has to happen before the coding call. If MDD criteria are met, that’s the diagnosis, not adjustment disorder sitting next to it as a softer label for the same symptoms.

“Does an Adult ADHD Diagnosis Work the Same Way as a Kid’s”

Adult ADHD assessments have grown enough that this question comes up regularly now, often from clinicians whose training was mostly pediatric. 

F90.0 covers predominantly inattentive presentation, F90.1 covers predominantly hyperactive impulsive presentation, and F90.2 covers combined presentation. The DSM-5 symptom threshold behind each of those shifts by age, six or more symptoms in a cluster under 17, but only five or more at 17 and older. 

An adult intake that applies the pediatric six symptom threshold can undercount and miss a diagnosis the correct adult threshold would have confirmed. F90.9, unspecified type, is fine while an evaluation is in progress, but staying there once a full assessment should have identified the subtype reads as an incomplete workup.

“Is Asperger’s Still a Diagnosis I Can Code”

This question comes from clinicians working with adults who were diagnosed decades ago under criteria that no longer exist in the DSM. DSM-5 eliminated Asperger’s syndrome as a separate diagnosis in 2013, folding it into one autism spectrum disorder diagnosis with severity levels. ICD-10-CM never made that same move. 

F84.0 lists autism spectrum disorder as an inclusion term, even though the code’s title still reads “autistic disorder,” and it’s the correct code for a DSM-5 based ASD diagnosis at any severity. F84.5, Asperger’s syndrome, still exists as its own code, and it carries an Excludes1 note against F84.0, meaning the two can never be billed together. 

An adult reassessed under DSM-5 as ASD without accompanying intellectual impairment belongs on F84.0, not left on F84.5 out of habit or deference to a label the client has used for themselves for twenty years.

“The Drinking and the Anxiety Are the Same Thing, So Why Two Codes”

Substance use and anxiety or mood symptoms showing up together is common enough that it’s worth knowing ICD-10-CM handles this differently than DSM-5 does. DSM-5 uses one severity scale for substance use disorder based on symptom count, two to three is mild, four to five is moderate, six or more is severe. ICD-10-CM never adopted that scale. 

The official guidelines still run the mental and behavioral disorders chapter on the older use, abuse, and dependence hierarchy instead. Mild maps to the abuse code. Moderate and severe both map to the same dependence code, because ICD-10-CM has no separate severe tier at all.

When drinking is severe enough to be producing anxiety symptoms directly, the answer to “why two codes” is that it shouldn’t be two codes. There’s a single combined code inside the substance block itself, F10.180, alcohol abuse with alcohol induced anxiety disorder. 

Billing the substance code and a standalone F41.1 for what’s really one induced condition counts the same clinical event twice, and it’s exactly the kind of chart pattern that AI assisted coding tools built for behavioral health, like the ones inside OmniMD, are designed to catch before the claim goes out rather than after a payer sends it back.

Conclusion

None of these codes are hard once the underlying logic is clear. Depression locks onto history, bipolar disorder resets with every episode, trauma runs on a clock, and a handful of common presentations, couples work and uncomplicated grief among them, sit outside what insurance treats as a diagnosis at all. The mismatches people run into almost always trace back to one of those patterns being applied where a different one governs instead. Getting that pairing right the first time is what keeps a claim moving instead of coming back.