Case Study 2 — The Heart Failure Nobody Specified: A Composite

A composite built from §11.3's axes, documented specificity patterns, and risk-adjustment mechanics. Tier 3; the organization and figures are constructed. The pattern is well documented.

Background

Section 11.3 listed four axes of heart failure classification and noted, without dwelling on it, that the unspecified heart failure code is used a great deal.

Chapter 7 §7.9 listed four costs of an unspecified code and said the largest one never appears on a claim. Chapter 10's Case Study 2 showed a coding error that produced no financial signal.

This case study puts those together and attaches a mechanism.

The composite

Constructed. Not a real organization.

A cardiology practice of fourteen physicians participates in several risk-bearing arrangements. Its coding is competent: audits are clean, the coders are certified, and nothing about the practice would raise a flag.

In an annual review of its own data, the practice notices that a large majority of its heart failure diagnoses are coded to the unspecified option.

Not the specified systolic, diastolic, or combined codes. Not the acuity-specified codes. The unspecified one.

For a cardiology practice, that is a striking number. These are the physicians who determine heart failure type. The echocardiograms are in the record. The ejection fractions are in the record.

What the investigation found

The practice did what Chapter 10's Case Study 2 says to do: it pulled charts and determined, for each unspecified code, whether the record supported more.

Three causes, in the proportions that matter.

Cause 1 — the record supported more and the coder did not look. A minority of cases. The echocardiogram report in the chart stated the ejection fraction and the type; the coder coded from the assessment, which said "CHF."

This is Chapter 7 §7.9's thirty-second search, not performed. It is the cheapest of the three to fix and the smallest of the three.

Cause 2 — the assessment said "CHF" and nothing in the encounter's own documentation specified. The type was determined at some prior encounter, is in the chart somewhere, and is not in this note.

This is the interesting one, because it is not obviously anyone's error. The coder is coding this encounter. The type exists in the record but not in this document. Whether the coder may reach across encounters for it is a genuine question, and the answer depends on organizational policy, on what the record for this encounter incorporates, and on whether the condition was actually addressed in a way the prior specificity describes.

The practice's honest conclusion was that its coders had no policy, no time, and no guidance, and had defaulted to the safest-feeling option.

Cause 3 — the documentation habit. The largest share. Physicians who know the type, determined it themselves, and write "CHF" in the assessment because that is how they have always written it and it communicates perfectly well to every clinician who reads it.

Nothing is wrong with the medicine. The note is clinically adequate. It is administratively incomplete, and the physicians had never been told that the distinction mattered.

What it cost

Constructed, and the mechanism is the point rather than the figure.

Nothing, on a fee-for-service claim. The unspecified heart failure code and a specified one are paid identically for an office visit. Every claim was correct, every claim paid, and no control the practice operated would have flagged any of it.

Materially, under the practice's risk-bearing arrangements. Heart failure specificity affects a risk score, and a population coded predominantly to the unspecified option is described as less ill than it is. The practice was being paid for a healthier panel than the one it was treating.

And in its own analytics, the practice could not answer basic questions about its own patients — how many had systolic versus diastolic failure — from its own coded data, despite having the information in every chart.

What it shows

First, it is the cleanest available demonstration that unspecified coding is not conservative. Chapter 5 §5.8 argued that downcoding is not the safe choice; this is the version with a mechanism. The practice was not underpaid on any claim. It was underpaid on a population, invisibly, for years, and nothing in its revenue cycle could see it.

Second, the three causes need three different fixes and the proportions determine the priority.

An organization that responds to this finding with coder training has addressed the smallest share. That is Chapter 7's Case Study 2 restated with numbers attached, and it is the most common misdiagnosis in this area.

Third, the middle cause is a genuinely hard question this book does not resolve. May a coder use specificity documented at a prior encounter? The honest answer is that it depends — on what the record for this encounter incorporates, on organizational policy, and on whether the condition was addressed in a way that prior specificity accurately describes. What is not acceptable is having no answer, which is what the practice had.

Fourth: the physicians were not doing anything wrong. They were writing notes that communicate perfectly to other clinicians. The gap is between clinical adequacy and administrative completeness, and nobody had ever explained to them that the second thing existed or mattered. That conversation is Chapter 38's subject, and it goes better when it starts from "your note is underselling what you determined" than from "you are documenting incorrectly."

The lesson

An unspecified code is a measurement of your documentation, not a decision about your coding.

Three carry-forwards:

Measure your own unspecified rate, by condition. It is available from your own data, it takes an afternoon, and a rate that is high for a condition your specialty determines is a finding.

Then diagnose the cause before you fix anything. Pull twenty charts. Determine, for each, whether the record supported more. The proportions determine whether this is a coder problem, a policy problem, or a documentation problem — and they are usually not the one you assumed.

And decide the prior-encounter question in advance. Whether a coder may use specificity from elsewhere in the record is a policy question with a defensible answer either way. Having no policy means every coder decides individually, inconsistently, under time pressure.

Discussion questions

1. The practice was paid correctly on every claim and underpaid on the population. Explain that apparent contradiction to a physician-owner in three sentences.

2. May a coder use heart failure specificity documented in an echocardiogram report in the chart but not referenced in the encounter note? Argue both sides, then say what a defensible policy would look like.

3. The largest cause was physician documentation habit, and the physicians were doing nothing clinically wrong. How would you open that conversation? Write the first two sentences.

4. §7.9 listed four costs of unspecified coding. Which of them appear in this case study, and which one is doing the most damage?

5. Compare this with Chapter 10's Case Study 2 (the history code that wasn't). One overstates a population's burden and one understates it. Both produced no financial signal on any claim. What does that pair suggest about what a revenue cycle's controls are actually built to detect?