Major depressive disorder ICD 10: choosing between F32 and F33

TL;DR: Major depressive disorder ICD 10 coding comes down to two decisions. F32 covers a single episode, and F33 covers recurrent episodes, and then a severity digit follows. Most miscoding happens on the first decision because prior episodes go unasked about, and most audit failures happen on the second because the severity in the code is not evidenced in the chart.

Key takeaways

• F32 is a single episode; F33 is recurrent. That choice depends on the history you have to ask for.
• The severity digit is a claim your documentation has to support.
• F33.1 is "Major depressive disorder, recurrent, moderate" and is valid for billing.
• Prevalence peaks sharply in young adults, which is worth checking your caseload against.
• Unspecified codes are for genuine uncertainty, not for saving time at intake.

Two clinicians assess the same client and code differently. One writes F32.1, single episode moderate. The other writes F33.1, recurrent moderate. Neither is careless. One asked whether this had happened before, and one did not.

That single question separates the two code families, and it is skipped more often than any other part of a depression assessment.

Which ICD-10 code do you use for major depressive disorder?

You pick from two families and then add severity. F32 covers a single episode. F33 covers recurrent episodes. Within each, a further digit records severity, so F33.1 carries the descriptor "Major depressive disorder, recurrent, moderate" and is valid for billing (AAPC).

This piece is about making those two choices well. For the broader picture of depression coding across the F32 and F33 range, our depression ICD-10 guide covers the full set, and we have a dedicated walkthrough of F33.2 for recurrent severe presentations without psychotic features.

The decision sequence is short:

1. Establish whether this is a first episode or a recurrence by asking directly about prior periods.
2. Assess current severity against symptom count, intensity, and functional impact.
3. Select the family, then the severity digit.
4. Use an unspecified code only where the history is genuinely unavailable.

Step one is the one that fails. Clients rarely volunteer a depressive episode from eight years ago that resolved without treatment, because they did not label it as one at the time. If you do not ask, the answer is silently no, and the chart records a single episode by default rather than by finding.

Why does the severity digit cause audit problems?

Because it is a specific claim, and specific claims get checked. When you code moderate, you are asserting a level of severity, and a reviewer can compare that assertion against what the note describes.

The mismatch runs in both directions, and both are problems.

Coded severe, documented mild. The note describes a client who is working, sleeping reasonably, and reporting a 4 out of 10 mood, while the code says severe. That reads as upcoding, whether or not it was intended.

Coded moderate, documented severe. The note describes daily suicidal ideation, five weeks off work, and significant weight loss, while the code says moderate. This one is more common than people expect, usually because the code was set at intake and never revisited, and it undermines any request for a higher level of care.

The fix is not complicated. Whatever severity the code claims, the note should contain the symptom count, the functional impact, and ideally a measure score that supports it. Our guides to depression progress note examples and MADRS scoring both cover what that looks like in practice.

One more discipline. Severity changes during treatment, which means the code should sometimes change too. A client who was severe at intake and is now mild should not still be carrying the intake code in month five.

What do the prevalence figures suggest about your caseload?

That your younger clients are considerably more likely to be presenting with this than your older ones, by a factor that surprises most clinicians.

In 2021, an estimated 21.0 million US adults had at least one major depressive episode, representing 8.3% of all US adults (NIMH). Prevalence was higher among females at 10.3% than males at 6.2%.

The age gradient is the striking part. Prevalence was highest among adults aged 18 to 25 at 18.6%, falling to 9.3% for ages 26 to 49 and 4.5% for those 50 and over (NIMH). Nearly one in five young adults.

There is also a treatment gap worth knowing. Among adults with a major depressive episode in 2021, an estimated 61.0% received treatment in the past year (NIMH). Roughly two in five did not.

What does the note need to contain?

Symptom coverage across domains, duration, severity evidence, risk status, and functional impact. Depression notes fail more often on domain coverage than on anything else.

Weak: Client reports low mood continues. Some improvement noted. Will continue weekly sessions.

Defensible: Client reports mood 3 out of 10, describing it as "flat rather than sad." Sleeping five hours with early-morning waking at 4 am on four nights this week. Appetite reduced, with an estimated 3kg weight loss over six weeks. Anhedonia evident in withdrawal from her walking group, now missed for five consecutive weeks. Concentration difficulty resulting in two missed work deadlines. Denies suicidal ideation on direct questioning, with no plan, intent, or means. PHQ-9 administered: 18, up from 15 three weeks ago. Presentation consistent with recurrent moderate episode, with current trajectory worsening.

That version supports a severity digit. It covers mood, sleep, appetite, anhedonia, and concentration, quantifies the functional cost, states the risk screen explicitly including its negative result, and gives a measure with a comparison point.

The risk line is not optional. In a depression chart, silence on risk is read as a gap in the assessment rather than as an absence of risk, and it is the single most consequential omission you can make.
When would you use F32 or F33, and when would you not?

Two decisions in sequence. First, has there been a prior episode, which sets F32 against F33. Then, what is the current severity, which sets the final digit. Both need evidence in the record, and the first one needs a question most assessments never ask.

Use these codes here

A client presents with a first-ever depressive episode: low mood, anhedonia, poor sleep, and concentration difficulty over ten weeks, with moderate functional impact. He is working but underperforming. On direct questioning, he has never experienced anything similar.

Code F32.1, single episode, moderate. The history was asked for, and the answer was documented, which is what makes the single episode designation a finding rather than an assumption.

A client describes a three-month depressive episode now, and on questioning recalls a similar six-month period at twenty-four that resolved without treatment. Current presentation is moderate.

Code F33.1, recurrent moderate. Note what produced the change. Nothing about her current presentation differs from the first example. The direct question about prior episodes is the entire difference between the two codes.

Do not use these codes here

A client presents with a depressive episode. Nobody asked about prior episodes, so the intake defaults to a single episode.

Do not code F32 by default. An unasked question is not a negative finding. Either ask it, or use a code reflecting the uncertainty and update once the history is available.

A client was coded F33.2 at intake five months ago during a severe episode. She is now working, sleeping normally, and reports mood 7 out of 10, but the chart still carries the intake code.

Update the code. The severity digit is a claim about the current picture, and one describing a client who no longer exists undermines both the record and any authorization request built on it.

A client reports low mood, low energy, and poor self-esteem persisting most days for the past three years, never severe enough to meet full episode criteria.

Consider persistent depressive disorder rather than a major depressive code. Chronic low-grade symptoms and discrete episodes are different presentations, and the duration here points away from F32 and F33.

A client's depressive symptoms began after starting a new medication; track its dose changes closely.

Do not code a primary depressive disorder without considering the substance or medication-induced alternative. Coordinate with the prescriber, document the temporal relationship, and code what the evidence supports rather than what is most familiar.

How the severity-to-code match can be automated

The recurring failure in depression charts is drift between the coded severity and the documented picture. The code is set once, and the clinical reality moves. That is a consistency problem across time, which is exactly what software handles well and clinicians handle poorly at volume.

Here is how it works in practice.

Domain coverage stops depending on recall. Depression is a syndrome across sleep, appetite, energy, concentration, anhedonia, guilt, psychomotor change, and suicidality. A note written from memory at the end of a day covers three or four of those. A note drafted from the session covers what was actually discussed, which is usually more.

The measure gains its comparison point. A PHQ-9 of 18 means little alone and a great deal against 15 three weeks earlier. When the score lands in the record consistently, the trajectory is available rather than needing reconstruction.

The risk line is always present. Including when negative, which is the case people forget. An explicit "denied, no plan or intent" is evidence of assessment, and a blank is evidence of nothing.

Supanote drafts the note from the session with the working diagnosis in view, so the symptom detail and the risk status accumulate in a form that either supports the coded severity or visibly does not. You review, edit, and sign every draft.

The limit, stated plainly: it will not choose F32 over F33 for you, because that depends on a history only you can elicit, and it will not set the severity digit, because that is a clinical judgment with billing consequences. It also cannot tell you a client is at risk. What it can do is make sure that when you decide, the evidence for your decision is in the chart rather than in your memory of a session six weeks ago.

If your caseload is small enough that you write notes between sessions and review diagnoses monthly, you will get less from this than the pitch suggests. The value is in volume, where drift accumulates unnoticed.

Want the chart to support the severity you coded? Try for Free.

FAQ

Q: What is the difference between F32 and F33?

A: F32 covers a single depressive episode, and F33 covers recurrent episodes. The distinction rests on whether the client has experienced a prior episode, which is history you have to ask about directly rather than infer from what they volunteer.

Q: What is the ICD-10 code for recurrent moderate depression?

A: F33.1, officially "Major depressive disorder, recurrent, moderate." It is valid for billing and requires both a documented history of at least one prior episode and current severity evidence supporting the moderate specifier.

Q: How do I decide between moderate and severe?

A: Assess symptom count, symptom intensity, and functional impairment together rather than any one alone. Whichever you choose, the note has to describe a picture matching it, since the severity digit is a claim a reviewer can check against your documentation.

Q: Should the code change as the client improves?

A: Yes. A client who was severe at intake and is now mild should not still carry the intake code months later. Update it and document what prompted the change, which turns a code edit into a defensible clinical decision.

Q: When should I use an unspecified depression code?

A: When the history genuinely is not available, such as an intake where the client cannot recall prior episodes and no records exist. Treat it as temporary and revisit it, since an unspecified code persisting for months suggests the assessment was never completed.

Q: Do I need a measure like the PHQ-9 to support the severity?

A: Not strictly, but it is the clearest evidence available, and it gives you a comparison point across time. A score recorded at a defined interval does more for a severity claim than several paragraphs of description.