Skip to main content
The Practice
Clinical operationsAugust 27, 2026

The Template Is Not the Problem. The Copy-Forward Is.

Why notes that read alike get flagged even when the care was skilled, which fields must change in every session note, and how to structure templates so the difference is forced.

Callie Editorial 14 min read
The copy-forward issue
What must vary
Session note
S

Relevant report

Caregiver reports carryover at home

O

Observable change

78% accuracy · minimal verbal cue

A

Clinical meaning

Self-monitoring is emerging

P

Next decision

Progress to conversational retell

At a glance

What you’ll leave with

  • Reviewers flag content that reads the same across visits or across patients — not the use of a template. Structure may repeat; wording that repeats is what reads as cloned.
  • Five things should be different in every treatment note: the minutes, the objective data, the patient’s response, the skilled-care sentence, and the plan for the next visit.
  • Build templates that prompt instead of pre-fill, and treat any copied-forward text as unverified until it has been edited against today’s session.

Somewhere in the middle of a full documentation day, every therapist has the same thought: I have written this note before. Same patient, same goals, a session that looked a lot like Tuesday’s — and an EHR offering to start today’s note from yesterday’s. Accepting that offer is how most cloned documentation happens: not fabrication, just a full note that nobody made false, and nobody made true either.

When cloned notes surface in an audit, the instinct is to blame the template and go back to free-text. That is the wrong lesson. Templates are how a practice keeps forty notes a week consistent, complete, and fast — and no payer prohibits them. The problem is narrower and more fixable: templates that arrive pre-filled with content, and copy-forward habits that make last session’s note the default state of this one. This article separates what may legitimately repeat from what must vary, shows both versions of the same note side by side, and ends with template structures that make the honest version the easy one.

The problem

What reviewers actually flag — and what they don’t

Medicare review contractors have defined the term directly: documentation is considered cloned when each entry in the record is worded exactly like or similar to previous entries, and they apply the same reading when notes repeat from patient to patient. Contractor guidance is equally direct about the consequence — cloned documentation does not demonstrate that the service was medically necessary, and it exposes the claim to denial and overpayment recovery. The wording matters: the finding is about entries that cannot be told apart, not about the format that produced them.

Federal program-integrity guidance says the same thing from the other direction. CMS’s Documentation Matters fact sheets tell practitioners to watch for cloned notes — notes that appear identical across different visits — because the record has to show the differences and the needs of the patient at each encounter. And the HHS Office of Inspector General, reviewing how EHRs can be misused, singled out exactly two documentation practices as fraud vulnerabilities: copy-paste cloning, and overdocumentation — inserting false or irrelevant material to make a service look better supported than it was. A template is on neither list. Identical content is both.

The mechanism

How a good template produces a cloned note

Templates go wrong in two specific ways, and neither is the existence of the template. The first is pre-filled narrative: a default sentence like “Patient participated in therapeutic activities with cues as needed and tolerated treatment well” sitting in the response field before the session has happened. Default text does not describe your patient; it describes the template author’s imaginary one. The second is copy-forward as the starting state: when today’s note opens already full of yesterday’s content, editing becomes optional — and on the day with three evaluations and a late cancellation, optional means skipped.

Both failures share a mechanism worth naming, because it also applies to AI-drafted notes: anything that starts the note full converts the clinician’s job from writing to verifying, and verification is the step that collapses under time pressure. A note that starts empty in the fields that matter cannot be signed by momentum. That single design property — which fields start empty — is most of what separates a defensible template from a cloning machine.

The dividing line

What may repeat, and what must vary

Medicare’s treatment-note requirements already force a floor of variance. For each treatment day, the note must record the date, each specific intervention furnished and billed under timed and untimed codes, the total timed-code minutes and total treatment time, and the signature and professional identification of the qualified clinician. Those elements change — or should change — with every real session. Clinical defensibility asks for more than the billing floor, though, and the useful way to think about it is field by field: some parts of a note earn their consistency, and some parts are only true if they are new.

Field by field: where consistency is safe and where it reads as cloning

Note elementRepeat or vary?What a reviewer expects to see
Section structure, headers, defined scales and cue hierarchiesMay repeatIdentical structure across notes is fine — a defined assist scale used consistently makes the data more credible, not less.
Diagnosis, precautions, and the active goal listMay repeatStable facts may carry forward from the plan of care, updated at progress reports or when status genuinely changes.
Date, timed-code minutes, and total treatment timeMust varyRequired on every note. Identical minute patterns across many visits are one of the first things a reviewer cross-checks.
Objective performance dataMust varyToday’s accuracy, assist level, distance, or trial counts — measured this session, not inherited from the last one.
Patient and caregiver responseMust varyWhat this patient did, said, or struggled with today. This is the field where cloned notes are easiest to spot.
The skilled-care sentenceMust varyThe judgment call you made mid-session — what you changed, graded, or cued, and why it required your license.
The plan for next visitMust varyWhat today’s data changes about next session. “Continue plan of care” repeated for weeks reads as no clinical reasoning.

The centerpiece

One template, one week, two versions of the record

Worked example

Daniel’s week, cloned and un-cloned

A fictional composite case, written to show the mechanics — not a clinical recommendation or a real chart. Daniel, 58, attends outpatient occupational therapy twice weekly after a distal radius fracture, working on the fine-motor and self-care demands of returning to his job. The clinic uses one standard treatment-note template for every visit.

The template both versions use

Five prompts, identical every visit: interventions furnished with timed-code minutes and total treatment time; objective performance data; patient response; skilled services provided this session; plan for next visit. Nothing about this structure will change between the two versions — only what goes into it.

The cloned week

Monday: “Patient participated in therapeutic exercise and ADL retraining with verbal cues as needed. Tolerated treatment well. Continue plan of care.” Thursday: the same three sentences, new date, same minutes. The following Monday: the same again. Each note is complete on its face — interventions, a response, a plan. Read in sequence, they are one note photocopied three times, and they leave no evidence that three distinct skilled sessions happened.

What actually happened that week

Monday, Daniel managed shirt buttons with setup and moderate verbal cueing, succeeding on roughly half his attempts, and reported his hand “gives out” by afternoon at work. Thursday he arrived fatigued after a full shift; the clinician downgraded the resistance work, switched to pacing strategies, and buttoning dropped noticeably. The next Monday, rested, he buttoned without setup on most attempts and the clinician progressed him to a smaller button size and timed repetitions. Three genuinely different sessions — none of it visible in the cloned version.

The same template, filled honestly (Thursday’s note)

Interventions and minutes as furnished that day. Objective: “Buttoning trials with 3/4-inch buttons: 3 of 8 successful with setup and moderate verbal cues, down from 5 of 8 on Monday. Grip endurance reduced; patient rated hand fatigue 7/10 after work shift.” Response: “Patient frustrated by regression; receptive to pacing discussion; identified end-of-shift tasks as hardest.” Skilled services: “Downgraded resistance program in response to observed fatigue; introduced pacing and joint-protection strategies specific to his assembly tasks; graded buttoning task to maintain success rate above frustration threshold.” Plan: “Reassess buttoning rested vs. post-shift next visit to separate endurance from motor recovery; progress button size if rested performance holds.”

What changed between the versions

Structure: identical. Content: nothing survived. The data is measured, the response is Daniel’s and dated to a work shift, the skilled sentence records two mid-session decisions, and the plan is a hypothesis today’s data created. A reviewer could place the three honest notes in order without looking at the dates. That is the test the cloned week fails.

The cost of the honest version

Perhaps ninety extra seconds per note, most of it typing numbers the clinician already knew. The cloned week does not save documentation time so much as defer it — to an additional documentation request, an appeal letter, or a repayment demand, each of which takes longer than every honest note that year combined.

A good template makes your notes rhyme. Cloning makes them repeat.

The fix

Structure the template so variance is the default

You do not fix cloning with a policy memo asking people to try harder. You fix it in the template itself, by deciding which fields start empty and what each empty field asks for. Five design moves do most of the work.

  1. 01

    Prompt, don’t pre-fill

    Replace every default narrative sentence with the question it was answering: “What did the patient do when the task got hard?” instead of “Tolerated treatment well.” A prompt is invisible in the signed note; a default sentence becomes the note. If a field ships with prose in it, that prose will be in half your charts by spring.

  2. 02

    Make today’s data structurally unavoidable

    Give objective measures their own labeled fields — trials, assist level, accuracy, duration — so a note without new numbers looks visibly unfinished rather than plausibly complete. The minutes fields (timed codes and total treatment time) belong here too: they are required on every Medicare treatment note, and they should come from today’s clock, not the template’s default.

  3. 03

    Add a delta line

    One required sentence: “Changed since last visit.” It is the cheapest anti-cloning device that exists, because it cannot be copied forward — a delta line that repeats is self-evidently false. It also forces exactly the comparison a reviewer will make: this note against the last one.

  4. 04

    Put the skilled-care prompt next to the intervention, not at the end

    ASHA’s Medicare guidance makes the point that skilled services that are not documented as skilled look unskilled — and unskilled care is not covered. The sentence that prevents that is a description of a decision: what you graded, modified, or cued mid-session and why. Prompt for it beside each intervention while the decision is fresh, not in a summary field at the bottom that defaults to boilerplate.

  5. 05

    End with a decision, not a status

    Replace the “Continue plan of care” free pass with a forced choice: progress, modify, or hold — and one sentence on what today’s data did to that decision. Weeks of “hold” with no rationale is its own audit finding; the field should make the reasoning, or its absence, visible to the clinician before it is visible to a reviewer.

The guardrails

If your EHR copies forward, treat the copied text as unverified

Copy-forward is not going away, and used honestly it has a legitimate job: carrying stable facts — precautions, diagnosis, the goal list — so the clinician spends their minutes on what changed. CMS’s documentation-integrity guidance points at the same guardrails from the fact-sheet side: turn off auto-fill features that write content on their own, make sure entries carry accurate date and time stamps, and make sure edits are attributable to the person who made them. The practical rule that covers all of it: anything that arrived in the note by software is unverified until you have read it against today’s session.

Field checklist

07 items

Before signing a note that started as last session’s

  • Every number in the note was measured today, not inherited from the previous visit.
  • The timed-code minutes and total treatment time match what was actually furnished today.
  • The response line describes something this patient did or said this session.
  • The skilled-care sentence names a decision made during this session.
  • The plan says what changes next visit — or why holding course is the clinical choice.
  • Read side by side with the previous note, a colleague could tell which session came first.
  • Anything copied forward that you did not verify has been deleted, not left in place.
What is cloned documentation in therapy?

Medicare review contractors describe documentation as cloned when entries are worded exactly like or similar to previous entries — whether repeated across a patient’s visits or from one patient to another. The finding is about content that cannot be distinguished between encounters, regardless of whether it was produced by copy-paste, a pre-filled template, or retyping the same sentences from memory.

Is it against Medicare rules to use documentation templates?

No payer rule prohibits templates or structured note formats. Medicare requires each treatment note to record the date, the specific interventions furnished and billed, the timed-code minutes and total treatment time, and the clinician’s signature and credentials — requirements a well-built template helps you meet. What draws scrutiny is note content that repeats across visits, which is a property of how the template is filled, not of the template.

Can two therapy notes legitimately look similar?

Yes. Recurring caseloads produce genuinely similar sessions, and consistent structure, scales, and phrasing conventions are good practice. The data, the patient’s response, and the plan should still differ — and when a session truly mirrors the last one, that stall is itself clinical information worth documenting, because a sustained plateau raises the continue-modify-discharge question a reviewer will also ask.

Is copying and pasting in an EHR illegal?

Copy-paste is a tool, not a violation. The risk is what it produces: the HHS Office of Inspector General has flagged copy-paste cloning and overdocumentation as the EHR documentation practices that can be used to misrepresent services, and a claim supported by documentation that misrepresents the encounter is where legal exposure begins. Copied text you have verified and updated against today’s session is documentation; copied text you have not is a liability.

Do AI-generated notes have the same cloning risk?

The mechanism is the same: any tool that starts the note full turns writing into verifying, and unverified output tends toward uniformity. An AI draft needs the same treatment as copied-forward text — check the numbers, the response, and the skilled-care sentence against what actually happened before signing. The clinician who signs owns the content, whatever produced the first draft.

What should change in every therapy treatment note?

At minimum: the date and the minutes (timed-code and total treatment time), today’s objective performance data, the patient’s response to this session, the skilled decision you made during it, and what happens next visit as a result. Structure, diagnosis, precautions, and the active goal list may reasonably carry forward between progress reports.

Primary sources

Bibliography / 8
  1. 01Medicare Benefit Policy Manual, Chapter 15, Section 220.3 — Documentation Requirements for Therapy Services (Pub 100-02)Centers for Medicare & Medicaid Services
  2. 02Complying With Outpatient Rehabilitation Therapy Documentation Requirements (MLN905365)Centers for Medicare & Medicaid Services
  3. 03Documentation Matters: Medicaid Documentation for Medical Professionals (Fact Sheet)Centers for Medicare & Medicaid Services
  4. 04CMS and Its Contractors Have Adopted Few Program Integrity Practices to Address Vulnerabilities in EHRs (OEI-01-11-00571)U.S. Department of Health and Human Services, Office of Inspector General
  5. 05Cloned Documentation Could Result in Medicare Denials for PaymentNational Government Services, Medicare Administrative Contractor
  6. 06Documentation of Skilled Versus Unskilled Care for Medicare Beneficiaries: Speech-Language Pathology ServicesAmerican Speech-Language-Hearing Association
  7. 07Documentation: Risk Management (Defensible Documentation)American Physical Therapy Association
  8. 08AOTA Official Documents — Guidelines for Documentation of Occupational TherapyAmerican Occupational Therapy Association

Written by Callie Editorial

Published August 27, 2026

Educational content, not legal, billing, or patient-specific clinical advice.