Coding to the right digit: severity and specifiers that matter

How specific does an F code actually need to be?

As specific as your current assessment genuinely supports, carried to the fourth or fifth character where the code structure allows it — and no further than the evidence in front of you. A specifier is the part of the code that names severity, episode status (single or recurrent), remission status, or a subtype; leaving it at the least-specific option when your notes clearly describe a moderate, recurrent, or in-partial-remission presentation is the gap reviewers notice first.

The fourth and fifth characters typically carry:

• Severity — mild, moderate, or severe, where the code family supports the distinction.
• Episode pattern — single episode versus recurrent.
• Remission status — in partial or full remission, where applicable.
• Subtype or specifier — with anxious distress, with mixed features, and similar clinically meaningful modifiers.

None of this means guessing at a digit you can't yet support. It means not stopping one digit short of what your documentation already shows.

Where a defensible severity specifier actually comes from

A severity specifier is strongest when it traces back to something more concrete than clinical impression alone. A validated instrument score is the clearest example: the PHQ-9's own validation work sets its severity bands at 5, 10, 15, and 20, with a score of 10 or above showing 88% sensitivity and specificity for major depression 1Ref 1Kroenke K, Spitzer RL, Williams JB (2001).The PHQ-9: validity of a brief depression severity measure.Supports using the PHQ-9's validated severity bands (5/10/15/20) as an objective, defensible basis for choosing a severity specifier. — a score in that range gives you a defensible basis for choosing moderate over mild that holds up under review.

You don't need a screening tool for every specifier decision, but where one is already part of your workflow, let it do double duty: the same score that informs your clinical picture also documents the reasoning behind the digit you chose. A specifier backed by a number and a date is harder to challenge than a specifier that only appears in the code itself with no supporting narrative.

Where no instrument applies, describe the functional and symptom picture in enough detail that a reader could independently arrive at the same severity level you coded.

When unspecified is the honest answer — and when it stops being one

Unspecified is a legitimate, common code in the first session or two, when the differential is still open and you genuinely don't yet know whether the presentation is single-episode or recurrent, mild or moderate. Using it there reflects where your clinical picture actually stands, and no reviewer expects certainty in session one.

The convention that experienced clinicians follow is to revisit the specificity at the point your assessment settles — often by the third or fourth session, once symptom count, duration, and functional impact are established. An unspecified code that persists for months of ongoing treatment, with a chart that otherwise reads as a settled, well-understood case, is what draws a reviewer's attention: not because unspecified is wrong, but because it stopped matching the chart around it.

The fix is simple and low-effort: when you update your clinical impression, update the code in the same note, not in a separate cleanup pass later.

The accuracy obligation behind the digit you choose

Whichever license governs your practice, the underlying obligation is the same: the diagnosis and its specifiers reflect your actual clinical judgment, never a digit chosen because it reimburses better or avoids a utilization-review flag. The psychology ethics code addresses accurate diagnosis and documentation directly 2Ref 2American Psychological Association (2017).Ethical Principles of Psychologists and Code of Conduct.Supports the ethical obligation for psychologists to document diagnoses, and their specifiers, accurately rather than for reimbursement convenience., the social work code carries the parallel obligation for LCSWs and LMSWs 3Ref 3National Association of Social Workers (2021).NASW Code of Ethics.Supports the parallel accuracy-in-documentation obligation for social workers coding diagnosis specificity., and the counseling code does the same for LPCs and LMHCs 4Ref 4American Counseling Association (2014).ACA Code of Ethics.Supports the parallel accuracy-in-documentation obligation for counselors coding diagnosis specificity. — three different licensing bodies converging on one rule.

This cuts in both directions. Coding a presentation as more severe than your assessment supports, to strengthen a medical-necessity argument, is the same violation as coding it less severe to avoid a conversation with the client about their diagnosis. Neither is a documentation shortcut; both are the line an ethics complaint or a payer audit is built to catch.

When you're genuinely unsure between two adjacent specifiers, document the ambiguity itself rather than picking one and hiding the uncertainty — a note that says why you chose moderate over severe survives scrutiny better than a code with no reasoning behind it at all.

How the F code and the procedure code read together

A reviewer never reads the diagnosis code alone; it's read against the procedure code on the same claim line. The psychotherapy CPT family — the 90791 evaluation, the 90832/90834/90837 timed individual codes, 90846/90847 for family work, 90853 for group — is where the service type and duration sit, while the F code is where the clinical justification for choosing that service sits 5Ref 5APA Services, Inc. (2025).Psychotherapy Codes for Psychologists.Names the psychotherapy CPT family that reviewers read alongside the F code and its specificity on the same claim.. A vague, unspecified diagnosis paired with a maximum-length 90837 raises a different question than the same code paired with a well-specified, recurrent, moderate presentation.

CPT itself is maintained separately by the AMA's CPT Editorial Panel and updated annually as the HIPAA-named procedure code set 6Ref 6American Medical Association (2026).CPT® (Current Procedural Terminology).Establishes that CPT is governed and updated separately from ICD-10-CM, framing why the two codes on a claim are reviewed on their own terms. — a reminder that the two codes on your claim come from two different governing bodies, on two different update cycles, and a reviewer expects each to do its own job well rather than one propping up the other.

The practical habit: before submitting, read the diagnosis and procedure code together as a reviewer would, not as two fields you filled in separately.

Updating specificity as the clinical picture develops

Specificity isn't a one-time decision made at intake and left alone. As the case develops — a single episode becomes recurrent, a moderate presentation resolves toward partial remission, a comorbid pattern emerges — the code should move with it, the same diagnosis-coding discipline that governs the broader bh-documentation question of translating DSM-5-TR into a billable ICD-10-CM code.

When a psychosocial or environmental stressor is doing real clinical work in the case but isn't itself the principal diagnosis, z-codes are the other half of specific, honest coding: they let housing instability, a relationship conflict, or job loss show up in the record without distorting the primary F code to compensate.

Build the review into your existing rhythm — a quick check at each treatment-plan update, rather than a separate specificity audit — and the code stays current without becoming its own task.