Case Study 1 — When the Record Started Writing Itself

Real, documented federal policy and oversight activity. Tier 1 for the statutory and regulatory framework and for the existence of the oversight findings; qualitative for magnitudes.

Background

Section 4.6 described cloned documentation as a problem with three faces. Here is where the problem came from, because it was not an accident and it was not primarily the fault of the clinicians who produce it.

Before roughly 2009, most American clinical documentation was dictated, transcribed, or handwritten. It was frequently illegible, often thin, and almost always individual — a physician describing a specific patient in their own words, at whatever length they chose. Its characteristic failure was absence: the note did not say enough.

The Health Information Technology for Economic and Clinical Health (HITECH) Act, enacted as part of the American Recovery and Reinvestment Act of 2009, committed tens of billions of dollars to incentivizing adoption of electronic health records, through what became the Medicare and Medicaid EHR Incentive Programs — later restructured as Promoting Interoperability. Adoption, which had been slow for a decade, became near-universal within a few years.

The electronic record solved the legibility problem completely and the thinness problem almost too well.

Because an electronic record can generate text. It can carry forward yesterday's note. It can populate a complete review of systems from a single click. It can pull the problem list, the medication list, the last six laboratory results, and the vital signs into the body of a note automatically. It can offer a template that produces a paragraph of examination findings from a checkbox.

Every one of those features has a legitimate use. Together, they changed the characteristic failure of American clinical documentation from absence to assertion: the note now says a great deal, and some of what it says is not true.

The issue

Federal oversight identified the problem early and specifically.

The HHS Office of Inspector General examined electronic health record documentation practices and their fraud vulnerabilities, and reported findings that have shaped the field since: that copy-paste and over-documentation features created vulnerabilities to fraudulent billing, that many hospitals had policies that did not adequately address copy-paste, and that CMS and its contractors had limited ability to detect the practice in review. The OIG also reported on the concentration of higher-level evaluation and management billing over time and recommended review of physicians who consistently billed at the highest levels.

CMS and its contractors responded with guidance and with review activity. Medicare Administrative Contractors published articles addressing cloned documentation, with a formulation that has been repeated in many jurisdictions since: documentation is considered cloned when each entry is identical to previous entries, and cloned documentation does not meet medical necessity requirements for coverage because it fails to demonstrate the specific medical necessity for the encounter billed.

The mechanism of harm is worth stating precisely, because it is more subtle than "people cheated."

The note asserts more than the encounter contained. A template producing a twelve-system review of systems and a complete multi-system examination generates documentation that, under the pre-2021 evaluation and management guidelines, could support a high-level visit — for an encounter that was genuinely a five-minute medication refill. The provider did not intend to overbill. The provider clicked one button and the record made a claim on their behalf. The provider then signed it, which is the part that matters legally.

The record stops being evidence of anything. If every note says the same thing, then no note tells a reviewer — or the next clinician — what happened on any particular day. Documentation that cannot distinguish between encounters cannot justify any of them.

And the pattern is trivially detectable. This is the practical point for anyone in the field. Identical text across a panel is found by a script, not by a reviewer's judgment. A provider whose documentation is uniform is visible in data long before anyone reads a chart.

What it shows

First, it demonstrates that documentation problems are usually system problems. Section 4.6's 📞 On the Phone callout is built on this: the productive conversation with a physician about cloned documentation is about the template configuration, not about the physician's integrity. In the overwhelming majority of cases, the physician did not decide to auto-populate a review of systems. Someone configured the system that way, often years earlier, often to satisfy a different requirement that no longer exists.

Second, it explains the 2021 evaluation and management revision. Chapter 15 covers the change in detail; here is the causal link. The pre-2021 guidelines rewarded documented volume — bullet points of history and examination — and the electronic record made volume free. A payment rule that measures something a machine can generate for nothing will measure it wrongly. The 2021 revision removed history and examination as level drivers precisely because they had stopped carrying information, and replaced them with medical decision making or time, both of which are harder to auto-generate. It is the clearest example in this book of a documentation rule being rewritten because technology made the old rule meaningless.

Third, it shows why the audit trail matters. Section 4.5's warning about amending records in response to a request is enforceable because the electronic record keeps the history. The same technology that made cloning easy made concealment nearly impossible.

Fourth, and most usefully for a coder: it reframes what "supported by the documentation" means. If a note's examination section was generated rather than written, then it is documentation of a template, not documentation of an examination. A coder cannot always tell the difference from a single note — but a coder reading four notes for the same patient, or the same note structure across a panel, frequently can. What to do with that observation is §4.6's conversation, and it is a professional obligation rather than an option.

Outcome

Electronic health record adoption is essentially universal in American hospitals and physician practices. The documentation problems it created have been partially addressed through several distinct channels:

• Guidance, from contractors and professional organizations, on acceptable copy-forward practice.
• Organizational policy, with many health systems adopting explicit copy-paste policies, attribution requirements, and visual indicators showing which text was carried forward.
• The 2021 evaluation and management revision, which removed the incentive to over-document history and examination for office visits, and its subsequent extension to other E/M categories.
• Vendor changes, including provenance tagging that shows the origin and age of copied text.
• Detection, both by payers and by internal audit, using text-similarity analysis.

None of it is finished. Chapter 38 takes up the next chapter of the same story, in which the record does not merely copy text but generates it — and the questions raised there are recognizably the same questions, at a larger scale and with less transparency about provenance.

OIG reports, contractor guidance, and organizational policy expectations all continue to develop. Verify current guidance with your compliance officer and your MAC.

The lesson

Documentation that a machine produced is documentation of the machine.

Three carry-forwards:

When you see identical text, do not assume dishonesty and do not ignore it. The overwhelmingly likely explanation is a template configuration nobody has revisited. The professional response is to raise it, in the terms §4.6 gives, and to raise it with whoever owns the configuration.

Volume is not support. A long note is not a well-documented note, and a note that asserts a complete examination is worth less as evidence than a short one that describes what was actually found. This is the single most useful thing a new coder can internalize about electronic documentation, and it runs directly against the intuition that more text is safer.

And the record's metadata is part of the record. When it was written, when it was changed, what was carried forward and from where. You will rarely look at it. Someone reviewing your organization will.

Discussion questions

1. The case study argues that cloned documentation is usually a system problem rather than an individual one. Does that change who is responsible for it? Distinguish responsibility for the cause from responsibility for the signed note.

2. §4.6 says templates improve completeness and legibility. Construct the strongest defense of templated documentation, then say precisely where the defense stops working.

3. The 2021 evaluation and management revision removed history and examination as level drivers. Using this case study, explain why. Then predict one way providers or systems might adapt to the new rule in a way its authors did not intend. (Chapter 15 will tell you whether you were right.)

4. A coder notices that a provider's examination section is identical across every patient they see. Map out the escalation: who do they tell, in what order, and what is the coder's obligation if nothing changes?

5. Compare this case study's failure mode with Chapter 1's Case Study 1 (price transparency). One is about a record asserting too much; the other is about a price disclosing too little. Is there a common structural cause, or are they unrelated?