Wynda Clayton audited Medicare Advantage records for the federal government before she became a compliance executive. A pattern she describes from that work, and still sees constantly from the other side of the table, has nothing to do with fraud. It is about language that looks connected on the page and is not.
Her clearest example is diabetes. A chart that reads “diabetes with neuropathy” or “diabetic CKD” does not, on its own, establish that the complication is actually caused by the diabetes. Without a word like “due to” or “secondary to” tying the two together, a coder reviewing that chart later has no documented basis to treat them as linked, and defaults to coding the conditions separately, whatever the treating clinician actually meant.
The same structural problem, by Clayton’s own account, runs straight through psychiatric documentation. Arguably worse.
A distinction most non-clinicians never encounter
Major depressive disorder has a documented, clinically meaningful structure. Recurrent. In partial remission. In full remission. These are not interchangeable descriptions of the same patient at different points of a conversation. They are distinct clinical states, each implying something different about symptom burden, relapse risk, and what treatment adjustment might be needed next.
In an actual chart, under actual time pressure, that structure regularly collapses into two words: depression, stable.
Most medical diagnoses have some physical correlate a clinician can point to on the page: an imaging result, a lab value. Psychiatric severity mostly does not work that way. It lives in a clinical interview, a mental status exam, a clinician’s judgement about symptom trajectory over time. That judgement is frequently accurate. It is also, by its nature, harder to compress into the kind of explicit, linked language a later chart review needs in order to treat a diagnosis as current, active, and supported: the same gap Clayton flags in the diabetes example, just without a convenient linking word to look for.
Where the paper trail becomes consequential
Diagnosis codes drawn from clinical notes feed directly into how Medicare Advantage plans get paid for the patients they cover, and federal auditors periodically review samples of those diagnoses against the documentation behind them. The Department of Health and Human Services Office of Inspector General has published guidance stating that in its review of Medicare Advantage records, “a high percentage of the associated HCCs could not be validated,” giving the agency “indications that MAOs are submitting diagnosis codes for payment that are not verifiable.” Among the specific conduct it names as revealed by federal investigations is leaving a previously submitted diagnosis in place even after a later chart review shows the documentation no longer supports it.
Psychiatric conditions sit close to the centre of this problem, not through carelessness, but through the structural mismatch above. A condition genuinely being actively managed can still fail review if the note never shows current evidence: no mention of symptom trajectory, no treatment adjustment, no reassessment of severity, just the same two words repeated across a year of visits.
Dementia carries the same pattern, differently shaped
Severity staging, mild, moderate, or severe cognitive impairment, and the functional differences between them, carries real weight for care planning. A chart that simply says “dementia” without a severity marker is not wrong, exactly. It is incomplete in a way that eventually shows up as a gap, usually discovered by a reviewer working from a note that was never written with that later scrutiny in mind.
This is the specific work retrospective HCC coding review does: checking a chart after the fact against the same documentation standard an auditor would apply, before an external reviewer gets there first. Done well, it is closer to translation than to auditing, taking clinical judgement a psychiatrist or geriatrician already exercised and making sure the record actually shows the connections and the current evidence a reviewer will look for, the “due to,” the trajectory, the reassessment, whatever Clayton’s pattern says was missing.
What this actually asks of the system, not the clinician
This is not an argument that psychiatric clinicians should write more, under time pressure that is already, by most accounts, unsustainable. Depression is often an accurate seven-word summary of forty minutes of clinical reasoning, and asking a clinician to type a paragraph instead is not a realistic fix.
The more useful question, the one Clayton’s background as a former auditor keeps her focused on, is what the systems downstream of that note are built to catch. A tool that flags when a chart lacks the specificity a reviewer will eventually demand, before that gap becomes someone else’s problem months later, protects the clinician’s time rather than adding to it.
Mental health documentation was never going to be simple to code. The question worth asking is whether the systems built to review it actually understand that complexity, in the same specific, structural way a former federal auditor learned to see it, or just flatten it the same way the two-word note already has.
Amelia Hart, a psychology graduate from the University of Hertfordshire, has a keen interest in the fields of mental health, wellness, and lifestyle.
