Documentation Integrity
How a chronic condition disappears from the record
Mike Garrett, CPC, CRC · September 4, 2026
Every January, thousands of patients are cured of conditions that have no cure. Amputations grow back. Transplanted organs return to their donors. Chronic kidney disease resolves overnight.
None of that happened, of course. What happened is simpler and duller: the condition was documented last year and not this year. Risk adjustment resets annually — a diagnosis only counts for the payment year in which it is documented in a face-to-face encounter and supported by the record. A condition that isn’t re-documented doesn’t just go uncounted. As far as the data is concerned, the patient no longer has it.
The three places conditions go missing
The problem list that nobody reads into the note. The condition sits in the EHR problem list, carried forward visit after visit, but the provider never addresses it in the assessment. A problem list entry is not documentation of an encounter. If the note doesn’t show the condition was monitored, evaluated, assessed, or treated, a coder can’t code it — and an auditor who finds it coded anyway will ask what supported it.
The specialist’s chart that never made it home. The nephrologist is managing the CKD, the endocrinologist is managing the diabetes complications, and the primary care note says “stable, followed by specialty.” The specialist documentation may be excellent — but if those encounters aren’t in the record being coded, the conditions they support aren’t either.
The visit that never happened. Patients with the most significant chronic conditions are often the ones least likely to come in. A patient with four chronic conditions who skips their annual visit generates no encounters, no documentation, and no record that the conditions persist.
Why this is a documentation problem, not a coding problem
It’s tempting to treat this as something coders should fix at the back end. They can’t, and they shouldn’t try. A coder’s job is to code what the documentation supports — not to infer that a condition probably still exists because it existed last year. Coding from memory, from the problem list, or from prior-year claims is exactly the pattern that fails a Risk Adjustment Data Validation audit, because the sampled record has to stand on its own.
The durable fix is upstream. Providers need to know, before the encounter, which chronic conditions need to be addressed and documented this year — and their notes need to show the work: the status of the condition, what was evaluated, and what the plan is. That’s what prospective review is for, and it’s why we treat provider education as part of every engagement rather than an add-on.
What a defensible record looks like
For each chronic condition, at least once in the payment year, the record should show a face-to-face encounter where the note demonstrates the condition was actually addressed — monitored, evaluated, assessed, or treated — with enough specificity to support the code assigned. Not a problem-list mention. Not “history of” language for a condition that is current. Not a diagnosis that appears in the billing data with nothing behind it in the chart.
That standard protects everyone. The patient’s record accurately reflects their health. The organization’s data reflects its population. And when an auditor pulls the chart, the documentation is simply there.
Conditions shouldn’t disappear from the record any more than they should appear in it unsupported. Both directions are the same discipline: the record says what is true, and shows how it knows.