Case Study 2 — The Emergency Department Is Outpatient: A Composite

A composite built from the setting-determination rules in Sections II and IV and from documented audit patterns. Tier 3; the hospital and its figures are constructed. The rule is real and it is frequently misapplied.

Background

Section 9.5 gave the uncertain-diagnosis rule in both settings and flagged a trap: the emergency department is outpatient. A patient seen in an ED and discharged home is an outpatient encounter regardless of how sick they were, how long they were there, or how much was done.

That sentence is easy to say and surprisingly hard to apply, because an emergency department does not feel like an outpatient setting. It is inside a hospital. It admits people. It runs a full diagnostic workup. It produces documentation that reads like an inpatient record, complete with hedged differentials.

This case study is about what happens when a coding operation gets the setting determination wrong at scale.

The composite

Constructed. Not a real organization.

A regional hospital's coding department is organized by coder, not by setting. Inpatient coders code inpatient records. Outpatient coders code clinic and ancillary records.

Emergency department records go to whoever is available, because ED volume is unpredictable and the department's staffing model treats ED charts as a flexible overflow queue. In practice this means that roughly a third of ED professional records in any given month are coded by staff whose primary work is inpatient.

Those coders apply Section II. Not out of carelessness — out of habit, and out of an entirely reasonable reading of a record that looks like the records they code all day.

The specific consequence: an ED note documenting "possible pulmonary embolism, CT negative, discharged with follow-up" gets coded, by an inpatient-trained coder, as pulmonary embolism — because in their setting, a diagnosis documented as possible at the time of discharge is coded as if established.

In the emergency department, that is wrong. The correct codes are the documented signs and symptoms — the chest pain, the shortness of breath — and the suspected diagnosis is not coded at all.

What it produced

Three consequences, in ascending order of seriousness. (Figures constructed.)

1. A visible coding-accuracy problem. An internal audit sampling ED records found a pattern of uncertain diagnoses coded as established. The finding was clear and the remediation was obvious.

2. An invisible data problem. For the period involved, the hospital's own data showed a rate of pulmonary embolism, and of several other serious conditions, that was substantially higher than reality — because ruled-out conditions had been coded as present. That data fed quality reporting, internal analytics, and the hospital's understanding of its own case mix.

3. And a patient-record problem, which is the one worth sitting with. Patients who did not have a pulmonary embolism had one on their record. The consequences of that are not billing consequences. They are consequences for future care, for the next clinician who reads the problem list, and — depending on the condition — potentially for the patient's insurability. Section IV's rule exists substantially to prevent exactly this, and §9.5 said so.

What went wrong, structurally

The department was organized around people rather than around rules.

This is the finding, and it generalizes well beyond emergency departments. The Guidelines are partitioned by setting, and Sections II and IV genuinely disagree. An organization that assigns work by availability rather than by setting has created a situation where the applicable rule depends on who picked up the chart — which is not a rule at all.

Three specific failures:

No setting determination step. Nothing in the workflow required a coder to establish, before coding, which section governed. The setting was assumed from the document's appearance.

No cross-setting training. Inpatient coders were not taught Section IV, because their job did not require it — until the staffing model made it require it, and nobody revisited the training.

And the error was undetectable from the claim. This is the part that made it persist. A claim carrying a pulmonary embolism code for a patient who was worked up for one is not implausible. It does not reject. It does not edit. Nothing about the claim signals that the diagnosis was ruled out — the information that would reveal the error is in the record, not on the claim.

What it shows

First, "which section governs" is a step, not an assumption. §9.2 said to know which setting you are in before you look for a rule. This is what happens when that is left implicit.

Second, the emergency department is the hardest case and it is not the only one. Hospital outpatient departments, observation, ambulatory surgery, and provider-based clinics all sit inside hospitals and are all outpatient for coding purposes. Observation is the sharpest — Chapter 3's Case Study 1 showed that observation versus inpatient is itself a contested determination, and the coding rules turn on it.

Third, the harm was not primarily financial, which is unusual for this book. The claims paid. The money was approximately right. What was wrong was the record — the hospital's data and the patients' problem lists — and no financial control would have detected it.

Fourth, it illustrates why Section IV's rule exists. A reader can experience the outpatient uncertain-diagnosis rule as pedantic: the physician clearly suspected a pulmonary embolism, the workup was real, why not record it? This case study is the answer. Coding a suspicion as a fact puts a disease on a person's record, and the outpatient setting — a snapshot, with documented symptoms available to code instead — has no good reason to do that.

The lesson

Determine the setting before you determine anything else.

Three carry-forwards:

Make it an explicit workflow step. Not "the coder will know." A field, a queue assignment, a question at the top of the abstract — something that forces the determination and records it.

Train for the settings a coder will actually see, not the ones in their job title. If a staffing model routes ED charts to inpatient coders, those coders need Section IV. That is a training consequence of a staffing decision, and the two decisions are usually made by different people who do not talk to each other.

And remember that some errors have no financial signal. This entire failure was invisible to every claim-based control the organization had. Finding it required reading records against the rules — which is Chapter 37's internal audit, and it is the only instrument that detects this class of error at all.

Discussion questions

1. The claims paid and the money was approximately right. Does that make this a less serious problem than one that costs money? Answer with reference to who was harmed.

2. §9.5 explains the outpatient rule as protecting the patient and the record. Before this case study, did that explanation persuade you? Has it now, and what changed?

3. Design the workflow control that forces a setting determination. Be specific: where does it sit, who does it, what happens if it is skipped, and how would you know it was working?

4. The inpatient coders were applying a rule correctly — the wrong rule, correctly. Is that a training failure, a staffing failure, or a management failure? Does the answer change what you would fix first?

5. Name two other errors in this book that had no financial signal, and say what they have in common with this one.