Cannabis use disorder ICD-10: F12 coding for billers
Cannabis use disorder ICD-10 coding: how DSM-5 severity maps to F12, whether legal cannabis changes the code, and when no code should be assigned at all.
In this article
- What is the cannabis use disorder ICD-10 code?
- How does DSM-5 severity map to F12?
- Does legal cannabis change the code?
- When should you not assign a cannabis code at all?
- How do you code cannabis withdrawal?
- What makes cannabis claims deny?
- How do you code cannabis-induced psychosis?
- What does a defensible cannabis admission chart contain?
- How the use-to-disorder link can be automated
- FAQ
TL;DR: Cannabis use disorder ICD-10 coding runs through F12, and the hardest question is not which code to pick. It is whether a code belongs on the claim at all. Legalization changed how patients talk about cannabis and changed nothing about the coding rules, which is where most facilities get into trouble.
What this will cover
- What is the cannabis use disorder ICD-10 code?
- How DSM-5 severity maps to F12
- Does legal cannabis change the code?
- When you shouldn't assign a cannabis code at all
- How to code cannabis withdrawal
- What makes cannabis claims deny
A client tells you they smoke every evening. They're relaxed about it, they have a card, and in their state it's about as remarkable as saying they have a glass of wine with dinner.
You now have a decision that a client who hedged about alcohol would never force you to make. Is this a diagnosis? And if you code it, what happens to that record?
That question didn't exist in the same form fifteen years ago, and the code set hasn't adapted to it.
Cannabis has become the substance patients disclose most freely and clinicians document most loosely. A patient who would hedge about alcohol will tell you exactly how much they smoke, because in most of the country it is legal and unremarkable.
That candor produces a lot of information and not much diagnosis. Somewhere between "patient reports daily cannabis use" and a claim, someone has to decide whether that sentence describes a disorder, and the decision is often made by whoever is coding rather than whoever examined the patient.
What is the cannabis use disorder ICD-10 code?
Cannabis-related disorders sit in F12, following the block's standard pattern. F12.10 is cannabis abuse uncomplicated, F12.20 is cannabis dependence uncomplicated, F12.21 is dependence in remission, and F12.23 is dependence with withdrawal.
| Code | Reads as |
|---|---|
| F12.10 | Cannabis abuse, uncomplicated |
| F12.11 | Cannabis abuse, in remission |
| F12.20 | Cannabis dependence, uncomplicated |
| F12.21 | Cannabis dependence, in remission |
| F12.23 | Cannabis dependence with withdrawal |
| F12.90 | Cannabis use, unspecified, uncomplicated |
F12.23 is valid and billable for FY 2026 claims, which is worth knowing because plenty of clinicians still believe cannabis withdrawal is not a recognized condition.
How does DSM-5 severity map to F12?
More cleanly than in most categories, though the mapping is still a convention rather than a rule. F12.10 corresponds to DSM-5 mild cannabis use disorder, while F12.20 covers moderate or severe cannabis use disorder without withdrawal or other complications (Blueprint).
That gives your team a working translation. A clinician documenting "moderate cannabis use disorder" is describing something the coder can reasonably render as F12.20, and a mild presentation goes to F12.10.
Write the convention down anyway. It is your organization's interpretation of how two systems relate, not a published crosswalk, and an auditor will want to see that you applied it consistently rather than case by case.
Does legal cannabis change the code?
No. There is no ICD-10 pathway that separates therapeutic use from recreational use, and no CPT or HCPCS entry for medical marijuana. A patient using cannabis under a state program and a patient using it recreationally reach the same codes, judged by the same clinical criteria.
What legalization changed is the conversation. Patients disclose more, clinicians normalize it more, and the threshold for calling something a disorder gets fuzzier. None of that alters what a payer expects to see supporting a claim.
The practical risk sits in both directions. Coding a disorder because a patient uses cannabis regularly, without impairment documented, overstates the chart. Failing to code a genuine cannabis use disorder because it feels unremarkable in a legal state understates it, and undercuts the treatment you are billing for.
When should you not assign a cannabis code at all?
When the chart records use and nothing else. A code belongs on the claim when the provider has documented that the substance use relates to a disorder or a medical condition being addressed. Use reported in a social history, with no complaint, impairment, or treatment attached, does not meet that standard.
This applies regardless of the state's legal framework. Recreational legality does not make a code more appropriate, and it does not make one less appropriate either. The question is always whether the record establishes a clinical relationship.
The honest version of this is uncomfortable. Some facilities code cannabis on every admission because it appears on the drug screen, which turns a lab result into a diagnosis. A positive screen is evidence you can use. It is not, by itself, a disorder.
How do you code cannabis withdrawal?
With F12.23, when dependence criteria are met and the patient shows clinically significant withdrawal after stopping or cutting down. The symptoms are real, well described, and routinely missed in documentation because they present as irritability and sleep disruption rather than as anything dramatic.
This matters for detox and residential claims specifically. A cannabis-dependent patient in a monitored setting whose chart records only F12.20 has a weaker medical necessity argument than one whose chart documents withdrawal.
The documentation standard mirrors the rest of the block. A symptom list in a nursing note is not the same as a provider assessment naming cannabis withdrawal, and only the second one defends the code.
Common pitfalls to avoid during coding cannabis
Coding from the drug screen. The single most common error with this substance, and the one with the longest tail for your client.
Not coding a real disorder because it feels minor. The mirror image, and it's why some cannabis-primary clients never get the level of care their presentation warrants.
Recording frequency instead of impairment. "Smokes daily" isn't a diagnosis. "Two failed attempts to cut down, missing shifts, continued use despite job warning" is.
Missing withdrawal. Cannabis withdrawal presents as irritability and sleep disruption rather than anything dramatic, so it gets treated and never diagnosed.
Applying the medical cannabis exception that doesn't exist. There's no separate code path. If a client on a medical program meets disorder criteria, they meet them.
What makes cannabis claims deny?
Weak linkage between the code and the service, more than anything else. Insurers denied 19% of in-network marketplace claims in 2024, with rates as high as 36% at some insurers (KFF).
| Denial | What to check |
|---|---|
| CO-50 medical necessity | Does the chart document impairment, or only use? |
| CO-16 missing information | Is a documented withdrawal missing from the code? |
| CO-197 authorization | Does the billed diagnosis match the authorized one? |
Cannabis-only admissions draw more scrutiny than most, because payers know that daily cannabis use is common and impairment is what separates a disorder from a habit. If cannabis is the primary diagnosis on a residential claim, the chart needs to carry that weight explicitly.
Fewer than 1% of denied claims were appealed in 2024, and insurers upheld 66% of the ones that were (KFF). Getting the documentation right at admission is worth more than any appeal strategy.
How do you code cannabis-induced psychosis?
Through the F12.15 and F12.25 series, depending on whether the underlying pattern is abuse or dependence. F12.150 is cannabis abuse with psychotic disorder with delusions, F12.151 covers hallucinations, and F12.159 is the unspecified version. The dependence equivalents run F12.250, F12.251, and F12.259.
These codes matter more every year as high-potency products become the norm. A patient presenting with cannabis-induced psychosis is a clinically serious admission, and coding it as F12.20 describes a far calmer picture than the one your staff managed.
The causal claim has to come from the provider. Coding an induced psychotic disorder asserts the cannabis caused it, which is different from a patient who has both a primary psychotic disorder and a cannabis use disorder.
Getting that distinction wrong runs in both directions. Coding induced psychosis when a primary psychotic disorder exists understates a chronic condition, and coding a primary disorder for a substance-induced episode overstates one.
What does a defensible cannabis admission chart contain?
Impairment, not frequency. The chart that survives review documents what the cannabis use is doing to the patient rather than how much of it there is.
Four elements carry the weight. Failed attempts to cut down, functional consequences at work, school, or home, withdrawal on cessation, and continued use despite a documented problem the patient acknowledges.
A chart with all four supports a residential admission. A chart recording daily use, a positive screen, and nothing else supports an outpatient conversation and very little beyond it.
This is the honest limit on cannabis-primary admissions. Some patients using cannabis heavily do not have a disorder that justifies intensive treatment, and a facility that codes every heavy user into residential care will eventually be asked to explain the pattern.
How the use-to-disorder link can be automated
The gap here is judgment applied inconsistently across hundreds of charts. One clinician codes every positive screen, another codes almost none, and neither pattern is visible until a payer audits a sample.
Supa runs inside the systems your team already uses, reading the assessment alongside the diagnosis list. A cannabis code assigned where the chart documents use without impairment gets flagged before the claim goes out. Supabill also catches documented withdrawal that never reached the code, and checks the billed diagnosis against the authorized one.
The clinician decides whether a disorder is present. What the software removes is the drift between clinicians, which is the thing audits actually find.
If cannabis appears as a primary diagnosis on your residential claims, it is worth checking what those charts support. Book a demo and we will review a sample with you.
FAQ
Q: What is the ICD-10 code for cannabis use disorder?
A: F12.10 for cannabis abuse uncomplicated, F12.20 for cannabis dependence uncomplicated, and F12.90 for unspecified use. Complications such as withdrawal or remission take further digits, so cannabis dependence with withdrawal is F12.23.
Q: Does medical marijuana get a different ICD-10 code?
A: No. The code set has no pathway separating therapeutic from recreational cannabis use. Coding depends on whether a clinician has documented a substance use disorder, not on how the patient obtained the cannabis or whether their state permits it.
Q: Is cannabis withdrawal a billable diagnosis?
A: Yes. F12.23 covers cannabis dependence with withdrawal and is valid for current claims. It needs a provider assessment naming the withdrawal, since symptom notes recorded by nursing staff do not on their own establish the diagnosis.
Q: Should I code cannabis use found only on a drug screen?
A: Not on its own. A positive screen is a lab result, and a diagnosis requires a provider to document a relationship between the use and a disorder or condition being treated. Coding from the screen alone overstates what the chart establishes.
Q: How does DSM-5 mild cannabis use disorder translate to ICD-10?
A: Mild generally corresponds to F12.10, cannabis abuse, while moderate and severe map to the F12.20 dependence codes. Treat that as a documented internal convention and apply it consistently rather than deciding case by case.
Q: Why do cannabis-primary claims get scrutinized more?
A: Because regular cannabis use is common and does not by itself indicate a disorder. Payers look for documented impairment, failed attempts to cut down, or withdrawal, so a chart that records only frequency gives a reviewer an easy challenge.
Keep reading
All articlesRun this on your own practice.
Watch ambient agents handle your front desk, documentation, and billing — inside the tools you already use.


