Case Study 1 — Where the Patient Went: A Composite

Constructed. The hospital and the figures are not real. The failure — a two-digit field that decides payment, entered by staff who are not told it decides anything, with no feedback loop of any kind — is ordinary, and discharge status is a documented audit target for exactly this reason.

Background

Section 26.7 said two things that sit uncomfortably together:

Patient discharge status decides more money than any other small field on the form.

It is frequently entered by unit clerical staff at the end of a stay, from a discharge order, under time pressure, into a field with sixty-odd possible values. It is not a coding decision and it is not usually reviewed by anyone who codes.

This is a hospital where both of those were true for a long time.

The composite

Constructed.

A community hospital. Discharge status is entered on the unit, by the clerk who closes out the account, from the discharge order and whatever the nurse tells her.

She is good at her job. She has been doing it for eleven years. Nobody has ever told her that the field affects payment, and nothing in her training suggested it — she was taught that it records where the patient went, which is exactly what it does.

Her default, when the order says "discharge" and the destination is not obvious, is 01 — discharged home.

That is right most of the time, because most patients do go home.

Where it went wrong

Three situations, all common, all producing an incorrect 01.

A patient discharged to a rehabilitation facility, where the order said "discharge" and the arrangements were made by case management in a different system. The clerk saw a discharge order and a patient leaving.

A patient discharged home with home health services arranged. Status 06, not 01 — and the distinction is invisible from the unit, because the patient did go home.

And a patient transferred to another acute care hospital for a service this one does not provide, recorded as 01 because the transfer was arranged clinically and the paperwork said the patient was leaving.

Why it produced money

§26.7's transfer rule.

Under the inpatient prospective payment system, a hospital that transfers a patient rather than discharging them may be paid a per-diem amount rather than the full DRG — for defined DRGs, when the stay is shorter than the geometric mean length of stay.

A transfer coded as a discharge is paid the full DRG when a per diem was due.

(Constructed.) Not on every case — the rule reaches specific DRGs and specific length-of-stay conditions — but on a meaningful fraction of the transfers, over years.

And the home health cases ran the same direction under a related provision.

Why nothing caught it

The claims paid. Correctly formatted, internally consistent, adjudicated without incident.

Coding never saw the field. The coders coded diagnoses and procedures from the record. Discharge status arrived on the claim from the registration and unit systems, and a coder reviewing a chart for sequencing has no reason to look at FL 17.

The remark field said so, and nobody read it. (Constructed.) On a number of the transfer cases, FL 80 contained a remark naming the receiving facility — §26.10's specific pattern. Somebody wrote down what happened in the one field nobody adjudicates.

And the clerk received no feedback of any kind. Eleven years, thousands of accounts, and not one report ever came back to her about a discharge status. She had no way to learn she was wrong because nothing in the organization was structured to tell her.

How it surfaced

An external audit, targeting the transfer rule specifically — which is a known review area and which requires no chart to target: a facility's distribution of discharge statuses is computable from claims data, and this one had markedly fewer transfers than comparable facilities.

(Constructed.) The review pulled records, compared FL 17 against the discharge summaries, and found a consistent pattern.

The repayment was substantial, and — because the pattern was consistent and the population was large — it was extrapolated. Chapter 37 §37.6 covers the arithmetic that turns a sample into a number.

What it shows

First, the field that decided the money was owned by nobody who knew it decided anything. Not coding, not the business office, not compliance. A clerk on a unit, doing exactly what she was trained to do.

Second, the organization was detectable from outside before it was detectable from inside. The auditor found it by comparing distributions across facilities — a comparison the hospital could have run on itself at any time and never did. Chapter 21 §21.10's principle, in a field nobody thinks of as a compliance risk.

Third, FL 80 contained the answer. The remark field named the receiving facility on a number of the mis-coded cases. The information was on the claim, in the one field that is not adjudicated, and reading it would have been free.

Fourth, the clerk is the wrong place to put the blame and the right place to put a report. She was never told, never measured, and never corrected. Chapter 24 §24.10 made the general argument — the front end is measured by the back end's failures — and this is the same structure one department over.

And fifth, this is another in the book's long run of failures with no financial signal, and it has a feature the others did not: the person creating it could not have detected it even in principle, because the consequence occurs in a payment system she has no visibility into and no reason to know exists.

The lesson

A field that decides payment must be owned by someone who knows it decides payment.

Four carry-forwards:

Reconcile FL 17 against the discharge summary, on a sample, on a schedule. It is one of the few facility-side checks a coder can run and that almost nobody runs. Fifty charts a quarter.

Read FL 80. The remark field is free to read and frequently contains the fact that contradicts a coded field.

Run your own distribution. Discharge statuses by DRG, compared across time and — if you can obtain it — against peers. An auditor will run it; you may as well run it first.

And tell the person entering the field what it does. Not as a compliance lecture. As information they were never given, and which several people in this composite would have wanted.

Discussion questions

1. The clerk was trained correctly and did the job as trained. Where should the correction have come from, and what would it have cost?

2. FL 80 contained the answer on a number of cases. Design the check. Is it automatable, and what is its false-positive rate?

3. The auditor detected the pattern from claims data without a chart. Should the hospital have run that comparison on itself? What would have prompted it to?

4. The repayment was extrapolated. (Chapter 37) What does extrapolation do to a consistent error, and why does consistency make it worse rather than better?

5. This is the first failure in this book where the person creating it could not have detected it even in principle. Does that change who is responsible? Does it change what a compliance program owes them?