Case Study 2 — The Twenty-Six-Hour Day: A Composite

Constructed. The practice and the figures are not real. The detection method is entirely real, it requires no medical record whatsoever, and it is one of the few audit techniques a coder can run against their own organization in an afternoon.

Background

Section 15.8 said that time is now available on every office encounter, that it is frequently the more favorable road, and that the practical guidance most compliance programs give is to document time on every encounter and select by whichever method supports the higher level.

That guidance is correct. It is also, in the wrong hands, an instruction to write a number.

Section 15.9 said something else: when a code's appropriateness is measurable from claims data alone, the audit does not need your records to find you. Time-based E/M selection is the clearest example of that principle in the entire book, because time has a property no other coding element has.

Time is additive, and the day has a fixed length.

The composite

Constructed.

A specialty practice adopted time-based selection deliberately and enthusiastically after 2021. The reasoning was sound: their visits genuinely ran long, their medical decision making was frequently moderate when the encounter felt like more, and time was the road that better reflected the work.

They configured their templates to insert a total-time statement on every note. The default value was prefilled — a reasonable default, from a reasonable estimate of a typical visit — and the clinician could change it.

Almost nobody changed it.

Within a year the practice's distribution of established patient office visits had shifted upward by roughly a level, and prolonged services, which had previously been reported almost never, were being reported regularly.

The arithmetic

Constructed figures; the method is not.

One physician's claims for a single Tuesday:

958 minutes is 15 hours and 58 minutes — of documented, personally-performed, patient-specific time, on one calendar date, by one physician.

And that is only the E/M time. It excludes every separately reported procedure performed that day, which by rule is not counted in E/M time and therefore has to be added to the total to describe the physician's day. It excludes charting on patients not seen. It excludes every minute of the day that was not a billed service.

The practice's schedule that day ran from 8:00 a.m. to 5:00 p.m.

What happened

The finding did not come from an audit of charts. It came from a spreadsheet.

A billing supervisor, preparing a routine productivity report, sorted claims by rendering provider and date of service and summed the documented times. She was not looking for a problem; she was building a report on visit duration for a scheduling discussion.

The column did not balance against the day.

She ran it for every physician and every date in the quarter. The pattern was consistent — not because anyone was inflating time deliberately, but because a prefilled default that nobody edits becomes the documented time on every encounter, and a default chosen to describe a typical visit describes every visit once it is never changed.

Why nothing caught it

Every claim paid. There was no denial, no edit, no request for records, and no correspondence.

Each individual note looked fine. A note documenting 33 minutes on an established patient with moderate complexity is unremarkable. The problem is not visible in any single record — it is visible only in the sum, and no clinical or coding review ever looks at the sum.

The internal audit sampled charts. It pulled twenty encounters, scored each against its documentation, and found them supported. A chart audit cannot detect this failure, because each chart is internally consistent. The error is a property of the population, not of the record.

And the shift looked like success. Revenue per visit rose. In the absence of anyone asking why, a rising average is a good number.

What it cost

Constructed.

An overpayment across the affected period, on a large volume of claims, in a category — time-based E/M level selection and prolonged services — that is a documented audit priority.

Chapter 5 §5.1's sixty-day rule attached the moment the supervisor's spreadsheet balanced wrong, and the practice's obligation was no longer optional. Chapter 37 §37.9 covers what a self-disclosure looks like.

And the compliance characterization is the part worth sitting with. Nobody at this practice falsified a time. A system was configured to state a number, and the clinicians did not correct it. Chapter 5 §5.3 defines "knowingly" to include reckless disregard for the truth or falsity of the information — and a prefilled time statement, signed on every note, without review, is squarely the kind of thing that definition exists to reach. Chapter 14 §14.1 made the identical point about automatically appended modifiers. This is the same failure with a different field.

What it shows

First, the detection method requires no medical records at all. Sum documented time by provider by date and compare it to the length of a day. You can run this against your own organization this week. If you take one operational technique from Part III, take this one — it is cheap, it is fast, and it finds a problem that no chart review will ever surface.

Second, a default value in a template is a statement. It is signed, it enters the legal record, and it is offered to a payer in support of payment. A default that is never edited is not a shortcut. It is an assertion nobody made.

Third, the failure is invisible at the level of the individual record and obvious at the level of the population. That is a general property worth naming, because it defines an entire class of error: anything that is only wrong in aggregate cannot be found by sampling. Chart audits are essential and they are blind to this.

Fourth, revenue moving in the right direction is not evidence of anything. The practice's average revenue per visit improved and everyone was pleased. A number improving is a question, not an answer.

And fifth, the direction here is the opposite of Chapter 14's Case Study 2 and Case Study 1 of this chapter. Those two were underpayments — the organization lost money and nobody noticed. This one is an overpayment, and it will be discovered eventually by someone whether or not the practice finds it first. Chapter 37 §37.2 collects all of these, and this pairing is why: the same absence of a financial signal hides errors in both directions, and only one of them comes with a deadline.

The lesson

Time is the one coding element that can be checked against physics, and almost nobody checks it.

Three carry-forwards:

Sum documented time by provider by date, every month. Compare it to the hours in a working day. It is one query, it needs no charts, and it is the highest-yield compliance control in this chapter.

Never prefill a factual statement in a template. Prefill structure, prompts, and headings. Do not prefill numbers, findings, or times. A field that requires an entry gets one; a field that arrives already filled gets signed.

And treat a favorable trend as a question. When the average level or the average revenue per visit moves, find out why before you report it as a result. Chapter 37 §37.2 builds this into the monitoring program, and it is the single habit that most reliably separates a compliance function that works from one that files reports.

Discussion questions

1. The detection method here needs no charts. Write the query in plain language, then say what number you would consider a red flag and why. What is the honest false-positive rate of your threshold?

2. Nobody falsified a time. Does §5.3's "reckless disregard" reach this conduct? Argue both sides, then state where you land and what the practice's obligation is.

3. The practice's chart audit sampled twenty encounters and found them all supported. Was the audit done badly? If not, what does that say about what sampling can and cannot do?

4. §15.8 gives the standard compliance guidance — document time on every encounter and select by the better method. This composite is what that guidance looks like when implemented carelessly. Rewrite the guidance so it survives implementation.

5. This is an overpayment; Chapter 14's Case Study 2 was an underpayment. Both were invisible. Which would your organization detect faster, honestly, and what does the answer tell you about where its controls actually point?