Case Study 1 — The Practice That Never Changed Its Templates: A Composite

Constructed. The organization and figures are not real. The pattern — an organization whose knowledge of a rule went stale while the rule changed underneath it — is entirely ordinary, and it is the reason §15.3 spends as long as it does on what the old framework was.

Background

On January 1, 2021, the office visit rules changed. History and examination stopped counting toward level selection. Medical decision making was redefined. Time was redefined and made available on every encounter.

Nothing about that change forced an organization to do anything. No claim rejected. No edit fired. The old codes still existed, the same claims still went out, and the same payments came back.

An organization that did nothing simply kept operating under a framework that no longer existed.

The composite

Constructed.

A twelve-physician primary care group. Competent, stable, well-regarded. Their electronic record's note templates were built in the mid-2010s to satisfy the 1997 documentation guidelines, and they were built well — each template generated a complete review of systems, a full multisystem examination, and a detailed history from a manageable number of clicks.

Their coding staff had been trained on the same framework and had gotten genuinely good at it. Their internal audits — real audits, done conscientiously — scored history, examination, and medical decision making, and their error rates were low.

By the time anyone looked closely, it was well into the second year after the change.

What was actually happening

Three things at once, and only one of them involved money.

First, the coders were scoring a framework that no longer governed. They were counting review-of- systems items and examination bullets, and they were treating medical decision making as the third of three components rather than as the whole determination.

The direction of the error was mostly downward. A visit with three chronic conditions individually assessed and prescription drug management is moderate MDM — a 99214 — regardless of how much history is documented. Under the coders' framework, a note whose history was "expanded problem focused" got capped, and the visit was coded 99213.

Second, the physicians were still generating documentation that bought nothing. Every note carried a full review of systems and a full multisystem examination, largely template-generated, on visits where neither was medically appropriate. That documentation had cost them time, it had cost them attention, and as of January 2021 it had stopped affecting the code entirely.

Third — and this is the part that should worry a compliance officer more than the money — the template-generated content was an assertion. A complete review of systems appearing in a record asserts that a complete review of systems was performed. Chapter 5 §5.4 established what an assertion in the medical record is. The exposure did not shrink when the documentation stopped mattering for coding. It stayed exactly where it was and lost its only justification.

How it surfaced

Not through an audit. Through a new hire.

A coder who had trained after 2021 was handed the practice's internal audit worksheet during her first week. The worksheet had columns for history elements and examination bullets.

She asked what they were for.

The answer she got — "that's how we've always scored it" — is a complete description of the failure mode, and it is worth noticing that nobody in that conversation was doing anything wrong on purpose. The senior coder had learned a framework, applied it well for years, and had not been told it was gone. Nobody had told her because everyone assumed someone had.

What it cost

Constructed.

Revenue. A systematic downward shift of roughly one level on a meaningful share of established patient visits, across twelve physicians, for approximately two years. The recoverable portion was small — corrected claims fall inside each payer's timely filing window, which Chapter 1 §1.3 measured in months, not years. Most of it was simply gone.

Physician time. Two years of template-generated history and examination content that affected nothing. Not catastrophic per note. Enormous in aggregate, and it is the cost the physicians felt.

Risk that never went away. Two years of records asserting examinations at a level of detail nobody performed.

What it shows

First, a rule change with no enforcement event is the most dangerous kind. Chapter 6 §6.4 said that the code sets update annually and that most organizations handle updates as a data problem — loading new codes, retiring old ones. This was not a data problem. No file changed. The framework in people's heads changed, and there is no update file for that.

Second, undercoding is an error. This deserves saying plainly because organizations rarely treat it as one. Compliance programs are built around the risk of overcoding, for good reason, and the unstated corollary — that coding low is safe — is false. A code is an assertion about what happened. A 99213 on a moderate-complexity visit is an inaccurate assertion, and it happens to cost money rather than create liability. Accuracy is the standard in both directions.

Third, documentation that no longer buys anything is not neutral. It costs the clinician time and it continues to assert facts. An organization that removed the requirement and kept the template kept all of the cost and none of the benefit.

Fourth, this is a training-debt failure, and training debt is invisible on every report an organization runs. There is no dashboard for "what our staff believes." The practice's error rate looked excellent, because it was measured against the wrong standard — an internal audit can only find errors its worksheet has a column for.

The lesson

When the rule changes and nothing breaks, the organization will not notice unless someone makes it notice.

Four carry-forwards:

Treat guideline changes as a training event, not a data event. The annual update process at most organizations handles code additions and deletions competently and handles rule changes not at all, because rule changes do not arrive as a file.

Audit against the current standard, and check the worksheet before you check the charts. A tool built for the old framework will return a clean error rate on a book of business that is systematically wrong.

Ask the new person what looks strange. This is the second time in two chapters that a new employee's outside comparison found something an internal control could not — Chapter 14's Case Study 2 was the first. That is not a coincidence and it is worth building on deliberately.

And treat undercoding as an error. Not as caution, not as conservatism. As an error, with a root cause, in the same log as everything else.

Discussion questions

1. Nobody in this composite did anything wrong on purpose, and the internal audit error rate was excellent. Design the control that would have caught this. Where would it live, and who would run it?

2. The chapter argues that undercoding is an error. Argue the other side — that in a world of audit risk, coding conservatively is rational. Then say what is wrong with the argument.

3. The templates continued to assert examinations that were not performed. Was that exposure created by the 2021 change, or merely revealed by it? Does the answer change what the practice should do about the two years of prior records?

4. Compare this with Chapter 14's Case Study 2. Both were found by a new employee. What does that suggest about onboarding, and what is the practical limit of the technique?

5. (Chapter 33, forward) The physicians spent two years generating documentation that bought nothing. Draft the message you would send them about it. What do you say about the two years already spent?