Case Study 2 — The Template That Answered: A Composite

Constructed. The practice, the payer, and the figures are not real. The failure — a documentation template that supplied an answer instead of asking a question — is ordinary, and it is the first instance in this book where the mechanism moved out of the claim and into the medical record.

Background

Section 29.8 drew a line and said it was worth memorizing:

An edit may STOP a claim and ask a question. It may not ANSWER the question.

And then extended it:

The same applies to templates. A template that asks the physician a question is documentation improvement. A template that carries forward an answer is Chapter 4 §4.7's copy-forward problem, and it degrades every other assertion in the note.

This is a practice that crossed the line in the second way, for the best of reasons, and was praised for it for two years.

The composite

Constructed.

A seven-provider practice with a real and well-understood problem: modifier 25 denials. Office visits billed alongside minor procedures, denied under the bundling edits Chapter 21 covers, appealed one at a time.

They had done the analysis correctly. The log showed the category. The overturn rate on appeal was high, which meant the denials were wrong and the documentation supported the visits. §29.7's data, used properly.

And somebody drew the right conclusion: if we win these on appeal because the note supports them, the note should make that visible before the claim goes out.

That conclusion is correct. What happened next is the case study.

What was built

(Constructed.)

A template. On any encounter where a minor procedure was documented, the note's assessment section gained a line:

"A significant, separately identifiable evaluation and management service was performed in addition to the procedure."

It appeared automatically.

Nobody involved thought of this as improper. The sentence was true on the great majority of those encounters — the practice had the appeal outcomes to prove it. It was written to save physicians from typing a sentence they would have written anyway.

What happened

The denials dropped.

(Constructed.) Substantially, and quickly. The modifier 25 category shrank to a fraction of what it had been, and the appeals it had been generating largely stopped.

And this was reported as a success, correctly by every measure the practice had:

   MODIFIER 25 DENIALS ......... down sharply
   APPEALS FILED ............... down
   STAFF TIME ON APPEALS ....... down
   OVERTURN RATE ............... unchanged (still high)
   PREVENTABLE DENIAL RATE ..... improved

The person who built the template was praised. (Constructed.) They deserved to be praised for the analysis, which was excellent, and the recognition attached to the whole thing.

Two years passed.

The audit

(Constructed.)

A payer's review, sampling charts on the modifier 25 pairing — a documented review target, and the practice's volume had risen.

The reviewer read the notes.

On most of the sampled encounters the sentence was true and the note supported it. (Constructed.) That is the important detail and it is why the finding took a while to develop.

On a minority, it was not. Encounters where the record documented the procedure and essentially nothing else — no separate history, no separate assessment, no other problem addressed — carried a sentence asserting that a significant, separately identifiable evaluation and management service had been performed.

The note said something the physician had not said.

And the reviewer's question was the one nobody in the practice had anticipated: who wrote this sentence?

Why this is worse than the claim version

Chapter 21's Account 31-2245 was modifier 59 appended by a billing macro. It cost \$25,720.80 and eleven defensible claims that could not be defended.

This is the same mechanism and the consequences are not the same, for three reasons.

The assertion is in the medical record. Not on a claim — in the document that is also the clinical record, the legal record, and the thing every subsequent provider reads. Chapter 4 §4.7's argument about copy-forward applies at full strength: once a note contains an assertion that was generated rather than made, every other assertion in it becomes a question.

The physician's signature is on it. (Constructed, and it is the part that produced the most difficult conversations.) Each of the seven had attested to notes containing a sentence they had not written and, on a minority of encounters, would not have written. They had not read it, because it was always there, which is the entire failure mode of a pre-populated field.

And it is not fixable retroactively. A claim can be corrected. A note can be amended, with a visible amendment trail, and cannot be un-asserted — and an amendment to a two-year-old note made during a payer review is a document nobody wants to create.

What it cost

(Constructed.)

Repayment on the sampled and extrapolated encounters where the assertion was unsupported — Chapter 37 §37.6.

A corrective action plan requiring the template to be changed and the practice's notes to be reviewed prospectively.

And the thing that is hardest to price: the practice's documentation, across two years, now carried a known generated assertion. Any subsequent review of any note from that period began with a question about what else was automatic. Chapter 37's auditors call this a scope problem — a single defect in a template does not stay confined to the encounters it was wrong about.

What it shows

First, the analysis was right and the intervention was wrong, and those are separable. The practice correctly identified its largest winnable denial category and correctly concluded that the fix belonged in documentation. Everything up to the design of the template was exemplary. The failure was a single design decision: the template asserted instead of asking.

Second, a prompt and a default are not variations of the same thing. "Was a significant, separately identifiable service performed today?" and "A significant, separately identifiable service was performed" are one word apart in effort and completely different in kind. The first produces evidence. The second produces text.

Third, the metric improved and the improvement was real. This is the third instance of the book's favorable-trend thread and it is the most uncomfortable: the denials genuinely fell, the appeals genuinely stopped, the staff time genuinely dropped, and every one of those was reported honestly by somebody who had done good work. A number that improves for a bad reason is not always a number that was measured wrong.

Fourth, the mechanism moved from the claim to the chart, and that is a category change. Chapter 25 §25.2 named the pattern — a field that stands in for a document is an assertion about that document's existence. Every earlier instance in this book lived on a claim. This one lives in the record, and the record has readers who are not payers.

And fifth, the sentence was true most of the time. (Constructed, and it is the reason nobody caught it.) A defect that is usually right is much harder to see than one that is usually wrong — Chapter 25's Case Study 2 made the same observation about a workaround that succeeded two-thirds of the time. Being mostly correct is what prevented anyone from looking.

The lesson

A template may ask the physician anything. It may not answer for them — and the difference is one word of grammar and an entire category of exposure.

Four carry-forwards:

Write prompts as questions. Literally, with a question mark, requiring an entry. Every automatic sentence in a note is an assertion nobody made, and there is no volume of good intent that changes that.

Audit your own templates for generated assertions. Pull twenty notes and ask, of each clinical statement: could this text have appeared without anyone deciding it? It is a fast review and almost nobody has ever run it.

When a denial category collapses after a documentation change, look at the notes. A large, fast improvement in a category that used to require argument is a finding, not only a success. Ask what the notes look like now — which is a question the practice in this composite could have answered in an afternoon at any point in two years.

And separate the analysis from the intervention when you evaluate either. The work that identified this category was genuinely good and should still be praised. A review that treats the whole project as a failure teaches people not to do the analysis, which is the opposite of the lesson.

Discussion questions

1. The practice correctly identified a real, winnable category and correctly concluded the fix belonged in documentation. At exactly which decision did it go wrong? State it in one sentence.

2. Rewrite the template's line as a compliant prompt. What does the physician have to do that they did not have to do before, and how long does it take?

3. The sentence was true on most encounters. Why did that make the defect harder to find? Name another case study in this book with the same property.

4. Seven physicians signed notes containing a sentence they had not written. Where does responsibility sit? Consider the template's author, the physicians, and whoever approved the change.

5. A claim can be corrected; a note can only be amended. What follows for how documentation interventions should be tested before deployment?

6. The denial metrics improved honestly and the improvement was real. How should a practice distinguish a genuine prevention success from this? Design the check.