ADHD ICD 10 code: picking between F90.0, F90.1, and F90.2

TL;DR: There is no single ADHD ICD 10 code. You choose by presentation, with F90.2 covering combined type, F90.0 and F90.1 covering the two single-domain presentations, and F90.8 and F90.9 catching the rest. The evidence problem is cross-setting: the code requires impairment in more than one place, and most charts document only the one you can see.

Key takeaways

• F90.2 is "Attention-deficit hyperactivity disorder, combined type" and is billable.
• Presentations can change over time, and the code should change with it.
• Impairment must be present in two or more settings, which usually means collateral report.
• Self-report is systematically unreliable in ADHD, and that is a clinical feature rather than a client failing.
• Prevalence differs sharply by sex, which is worth knowing when you audit your own diagnostic pattern.

The client cannot tell you how often it happens. Not because she is evasive, but because time estimation and retrospective recall are part of what is impaired. She thinks she was late "a couple of times" this month. Her manager has a different number.

That gap is the whole documentation problem with ADHD, and it is why a chart built on session self-report alone rarely supports the code it carries.

Which ICD-10 code do you use for ADHD?

You select by presentation type within category F90. The combined presentation is F90.2, officially "Attention-deficit hyperactivity disorder, combined type," and it is billable (AAPC). The two single-domain presentations sit at F90.0 and F90.1, with F90.8 and F90.9 available for other specified and unspecified presentations.

We have separate guides going deeper on the two single-domain codes, and they are the better starting point if you already know which presentation you are dealing with: F90.0 covers the predominantly inattentive presentation, and F90.1 covers predominantly hyperactive-impulsive. This piece is about choosing between them and evidencing whichever you choose.

The selection rule is simpler than it looks. Count which symptom domain carries the impairment. If both do, F90.2. If one clearly dominates, use the matching single-domain code. Reserve F90.9 for genuine uncertainty rather than for haste.

One structural point people miss. Presentation is not a fixed trait. A child coded F90.1 at seven, whose hyperactivity attenuates while inattention persists, may genuinely be F90.0 at fourteen. The code should follow, and the record should show why it moved.

Why does the two-setting requirement cause documentation problems?

Because you only witness one setting. ADHD requires evidence of impairment across more than one context, and your consulting room is a single, atypical, highly structured environment that frequently suppresses the symptoms you are trying to document.

That makes collateral information load-bearing in a way it is not for most diagnoses. A teacher report, a partner's account, a manager's feedback relayed by the client: these are not optional colors. They are how the second setting enters the record.

Two disciplines make this defensible. Record the source explicitly, naming who provided the information. And record that consent for the contact exists before you make it, since a collateral conversation without documented consent creates a different problem entirely.

Structured measures carry real weight here too, because they gather multi-informant data by design. Our guides to the ASRS v1.1 for adults and the NICHQ Vanderbilt scales cover administration, and the Vanderbilt in particular exists precisely to capture the parent and teacher perspectives your session cannot.

What do the prevalence figures tell you about your own coding?

They give you a base rate to check your pattern against, and the sex difference is the most striking part. Among US adults aged 18 to 44, current ADHD prevalence is 4.4%, with 5.4% among males and 3.2% among females (NIMH).

In children the gap is much wider. Among ages 4 to 17, 11.0% had ever been diagnosed, splitting to 15.1% of males and 6.7% of females (NIMH). The adolescent figures follow the same shape, with lifetime prevalence of 13.0% in males against 4.2% in females.

Here is an honest reading of that. The male-to-female ratio narrows considerably from childhood to adulthood, which is consistent with inattentive presentations in girls being identified later or missed entirely. If your own adult caseload skews heavily toward F90.0 in women diagnosed after twenty-five, you are probably seeing the downstream effect of that, and it is worth documenting the developmental history carefully rather than treating a late diagnosis as a weaker one.

What should the note contain?

Impairment described by setting, symptoms tied to concrete incidents, the collateral source, and any measure with its score and date. Adjectives do almost no work here.

Weak: Client continues to struggle with focus and organization. Reports ongoing difficulties at work.

Defensible: Client missed a project deadline on Tuesday, the third this quarter, after losing track of time on a lower-priority task. Reports re-reading emails four or five times without retention. Partner, contacted by phone with written consent on file, corroborates missed deadlines and reports household bills moved to his management after two late-payment penalties. In session, client required redirection twice during a 12-minute planning task and sustained focused attention for approximately 8 minutes. Impairment evident in occupational and domestic settings.

The second version establishes both settings, names the collateral source and its consent status, and includes an in-session observation. That last element matters because it is the only data point you personally witnessed, which gives it a different evidentiary quality from everything else in the note.

When would you use an ADHD code, and when would you not?

Three conditions gate the diagnosis: symptoms present before age 12, impairment in two or more settings, and symptoms not better explained by another condition. Presentation then determines which code. Most miscoding happens because the second and third conditions are assumed rather than established.

Use an ADHD code here

An adult client reports lifelong difficulty finishing tasks, losing items, and missing deadlines. School reports from childhood describe her as "away with the fairies." Her partner corroborates missed bills and unfinished projects at home, and her manager has raised deadlines at work. Hyperactivity is absent.

Code the predominantly inattentive presentation. Childhood onset is documented, impairment is established in both domestic and occupational settings by a source other than the client, and the symptom domain is clear.

A nine-year-old is out of his seat repeatedly at school and at home, interrupts constantly, and also loses homework and cannot sustain attention on non-preferred tasks. Parent and teacher rating scales both return clinically significant scores across domains.

Code the combined presentation. Both symptom domains are impairing and two independent informants confirm it across settings.

Do not use an ADHD code here

An adult reports serious concentration problems at work over the past eight months. At home, he manages fine. He describes no childhood difficulties and did well academically.

Do not code ADHD yet, and probably not at all. One setting, adult onset, and no developmental history. Screen for depression, anxiety, sleep disorder, and substance use before considering a neurodevelopmental explanation for a problem that started in adulthood.

A client reports classic inattentive symptoms. She also has a PHQ-9 of 22, has not slept more than five hours in months, and describes the concentration problems as beginning alongside the low mood.

Treat the depression and reassess. Concentration impairment is a criterion of a depressive episode, so coding both without separating them double-counts the same symptom. If the inattention persists once the mood lifts, and a developmental history emerges, code it then.

A client's symptoms are clear, but you have only their own account, and they cannot reliably recall the week.

Do not code on self-report alone in this diagnosis. The recall difficulty is itself part of the presentation, which is exactly why collateral matters more here than elsewhere. Obtain consent, gather a second source, and code once the cross-setting requirement is evidenced.

A client coded F90.1 at age eight is now sixteen. The hyperactivity has largely resolved, but inattention continues to impair schoolwork.

Update the code rather than leave it. The presentation is not fixed, and a chart still carrying the childhood code describes a client who no longer exists.

How can cross-setting evidence be automated?

The two-setting requirement is a collection problem, and collection problems are where software genuinely helps. The mechanism runs in three parts, and none of them involves the software making a diagnostic call.

Capturing the concrete incident. ADHD notes fail on specificity. "Struggles with focus" is what survives when a note is written at seven in the evening, while "missed a project deadline on Tuesday, third this quarter" is what survives when the note is drafted from the session itself. The second one evidences a diagnosis. The first does not.

Holding the collateral trail. Who said what, when, and with whose consent is exactly the kind of structured detail that gets lost in prose and matters enormously under review. When it is captured as it arrives, the second setting stays documented rather than remembered.

Keeping the in-session observation. Response latency, redirections needed, how long a non-preferred task was sustained: these are observations you make and rarely write down, because writing them down mid-session competes with doing the session. A draft generated from the session can carry them without you stopping to record them.

Supanote drafts the note from the session, so the incidents, the collateral, and your own observations land in the record together rather than being reassembled later. You review, edit, and sign every draft, and the clinical framing stays yours.

The limit is firm and worth stating. It cannot tell you whether the presentation is combined or predominantly inattentive, and it cannot judge whether a client's reported difficulty rises to clinical impairment. Those decisions require developmental history, the differential, and a clinician. What it removes is the reason good clinicians end up with thin charts, which is that the specifics were gone by the time anyone wrote them down.

Want the detail that supports the diagnosis to survive the day? Try for Free.

FAQ

Q: What is the ICD-10 code for ADHD combined type?

A: F90.2, officially "Attention-deficit hyperactivity disorder, combined type." It is billable and applies when both inattentive and hyperactive-impulsive symptom domains carry clinically significant impairment rather than one predominating.

Q: Can the ADHD code change over time?

A: Yes, and it often should. Presentation shifts with development, particularly as hyperactivity attenuates while inattention persists. Update the code when the picture changes, and document what you observed that prompted the change.

Q: Do I need collateral information to diagnose ADHD?

A: In practice, usually. The diagnosis requires impairment across more than one setting, and you only observe one. Collateral from a partner, parent, or teacher is normally how the second setting enters the record, with consent documented beforehand.

Q: Which code do I use if I am not sure of the presentation?

A: F90.9 covers unspecified presentations and is appropriate during assessment. Treat it as temporary. Once you have enough history to identify the predominant domain, update to the specific code rather than leaving the unspecified one in place.

Q: Can I diagnose ADHD if I am not the prescriber?

A: Scope depends on your license and jurisdiction, so check both. Where you are working alongside a prescriber, document your findings, name the diagnosing clinician and date, and record what you communicated to the prescriber and when.

Q: How do I document ADHD when the client cannot recall the week?

A: Record that as clinical data rather than working around it. Then change the method, using logs, phone capture, or collateral report, and document the change and its rationale. Recall difficulty is part of the presentation, not an obstacle to documenting it.