Case Study 2 — The Criteria Nobody Could Produce: A Composite

Constructed. The hospital and figures are not real. The situation — an organization asked to produce the written standard it has been applying for years, and unable to — is common enough that the request itself is a recognized audit technique.

Background

Section 16.9 said that there is no national leveling system for facility emergency department visits. Each facility develops its own written criteria, applies them consistently, and must be able to demonstrate that it did.

That is an unusual arrangement. Almost everything else in this book is governed by a rule somebody else wrote. Here, the facility writes the rule and then has to live up to it — which is more freedom and substantially more exposure.

The composite

Constructed.

A community hospital's emergency department. Its facility leveling criteria were written when the department implemented its current electronic record — a genuinely good document, developed by a committee that included nursing, coding, and finance, tied to specific nursing interventions and monitoring, and mapped to the five levels.

That was nine years ago.

In the years since: the department was renovated and expanded. Triage was restructured. Two new point-of-care tests were introduced. A behavioral health hold protocol was added. The electronic record was upgraded twice, and on the second upgrade the leveling logic was reimplemented by an analyst working from how the old system behaved, because the criteria document could not be found.

Nobody set out to change the criteria. The criteria changed anyway, one reasonable operational decision at a time, and the document that was supposed to describe them stopped describing them somewhere around year four.

What the distribution did

Constructed figures.

The distribution shifted upward by most of a level over nine years.

And here is the difficulty: that may be entirely correct. Emergency department acuity did rise nationally over that period. The department expanded its capability. It began holding behavioral health patients who consume enormous nursing resources. Every one of those is a real reason for a real shift.

It is also exactly what an upward drift in undocumented criteria looks like. From the outside, and from the inside, the two are indistinguishable — which is the entire problem.

The request

A payer's audit did not begin with charts. It began with a letter.

"Please provide the written criteria your facility uses to assign emergency department visit levels, together with documentation of the date of last revision and the process by which the criteria are reviewed."

Nobody could produce a current document.

What existed: the nine-year-old original, in a shared folder, unrevised. A one-page summary produced for a training session four years earlier that did not match it. And the configuration in the electronic record, which matched neither and which nobody had documented.

Why this was worse than a coding error

A coding error is a defect in an instance. You find it, you fix it, you quantify it, you correct the claims.

This was a defect in the standard, and it had three consequences a coding error does not have.

Every claim in the period was affected. Not a sample. Not a subset. The entire population of emergency department facility claims for nine years was assigned by a process the hospital could not describe.

The hospital could not defend the levels even where they were right. This is the part that surprises people. A great many of those claims were almost certainly assigned correctly — the levels reflected real resource consumption. But "correct" here means "consistent with our written criteria," and there were no current written criteria to be consistent with. The hospital lost the ability to prove accuracy it actually had.

And it could not tell an honest story about the shift. Asked why 99285 doubled, the right answer was probably "our case mix changed and here is how." Without a documented criteria history, that answer is indistinguishable from an excuse, and the hospital knew it.

What it cost

Constructed.

An extended audit, a full chart review at the hospital's expense, a settlement, and a corrective action plan requiring documented criteria, annual review, and periodic validation.

The largest cost was not the settlement. It was eleven months of senior staff time, the reconstruction of a criteria document from scratch under audit conditions, and a compliance posture in which every subsequent question started from a deficit.

What it shows

First, the absence of an external standard is not the absence of a standard. Where a rule lets you write your own, the writing is the compliance obligation — and organizations consistently read "develop your own criteria" as latitude rather than as a requirement with a deliverable.

Second, a document that is not maintained is worse than one that never existed. The nine-year-old criteria did active harm: it demonstrated that the hospital knew it needed criteria, had written them, and then stopped. An organization with no document can say it did not know. This one could not.

Third, "we've always done it this way" is not a criterion, and it is exactly what fills the vacuum. The electronic record analyst who reimplemented the logic from the old system's behavior was doing competent, conscientious work. The system's behavior became the standard because the standard was missing, and once that happens the standard is whatever the software does, which nobody can read.

Fourth, this is a third form of the invisible failure this book keeps returning to. Chapter 14's bilateral convention was invisible because nothing denied. Chapter 15's time defaults were invisible because each note looked fine. This one was invisible because there was nothing to compare against. No control can detect drift from a baseline that does not exist.

And fifth: the audit began with a document request, not a chart request. That is worth internalizing. Some audits test your claims. Some test whether you can describe how you produce them, and the second kind is answered in an afternoon or not at all.

The lesson

Where you write the rule, the rule is a deliverable — with a version, a date, an owner, and a review cycle.

Four carry-forwards:

Find out today whether your organization's facility leveling criteria exist and when they were last revised. If you are on the professional side, ask the equivalent question about whatever standard your organization writes for itself — charge capture rules, scrubber edits, query templates. Chapter 6 §6.5's scrubber rules are the same problem in a different department.

Attach a review cycle to anything you write. Annually, with a named owner. A criteria document without a review date is a document that will be nine years old.

Validate the software against the document, not the other way around. When a system is reimplemented, the source of truth is the written standard. If the standard cannot be found, stop and find it — that is the moment this failure is created, and it is the only cheap moment to prevent it.

And watch your own distribution. The shift here was gradual, defensible, and possibly correct. A hospital that had been tracking its own distribution annually would have had a documented, contemporaneous explanation for each year's movement — which is not a defense against being audited, but is very close to a complete defense once you are.

Discussion questions

1. The distribution shift may have been entirely legitimate. Does that change the hospital's position? What, specifically, did it lose by being unable to document its criteria?

2. The chapter says a stale document is worse than no document. Argue against that, then say where you land.

3. Design the annual review. Who is in the room, what do they look at, and what is the output? How long does it take, honestly?

4. The electronic record analyst reimplemented logic from system behavior because no document could be found. What should they have done, and what would it have cost the project?

5. Compare the three invisible failures: Chapter 14's bilateral convention, Chapter 15's time defaults, and this one. Each was undetectable for a different reason. Name the three reasons, and say which of the three your own organization is least equipped to catch.