A Monthly Chart Audit That Fixes Templates, Not Therapists

Every group practice owner knows the charts are not all perfect. Somewhere in the EHR there is a note missing a co-signature, a plan of care that expired mid-episode, a treatment note whose minutes do not quite support the units billed. The uncomfortable part is not knowing which charts, or how many — and the usual way an owner finds out is when a payer’s reviewer reads them first.

The fix is not a heroic annual audit, and it is definitely not waiting for a records request. It is a small, boring, monthly rhythm: pull a handful of charts, score them against a written rubric, and treat every finding as evidence about the system — the templates, the training, the workflow — rather than a verdict on the clinician whose name is on the note. This article lays out that rhythm for a group therapy practice: how to sample, what to score, who reviews whom, and how findings turn into changes that stick.

External chart review is not hypothetical for a therapy practice; it is infrastructure. Under Medicare’s Targeted Probe and Educate program, a Medicare Administrative Contractor that flags a practice reviews 20–40 claims per round, for up to three rounds, with education between rounds. The CERT program, which measures Medicare’s improper payment rate, consistently attributes most improper payments not to fraud but to insufficient documentation — notes that fail to support what was billed, missing plans of care, absent signatures. Commercial payers and Medicaid programs run their own versions. None of this requires the practice to have done anything wrong to be selected.

Federal compliance guidance has treated self-auditing as a baseline expectation for small practices for a quarter century. The HHS Office of Inspector General’s compliance program guidance for individual and small group practices — written explicitly for practices without a compliance department — lists internal monitoring and auditing as the first of seven components, and suggests a concrete starting point: a baseline audit, then periodic self-audits at least annually, with a basic sample of five or more records per federal payer, or five to ten records per clinician. The OIG’s 2023 General Compliance Program Guidance carries the same element forward for every provider type. An internal review rhythm is not gold-plating; it is the floor the government already describes.

The practical translation for a group practice: the OIG’s annual five-to-ten-per-clinician sample, split into a monthly pull, is two or three charts per clinician per quarter — a volume a working practice can actually sustain. The failure mode to avoid is the opposite design: the ambitious annual audit that reads thirty charts in one exhausting week, produces a report nobody acts on, and is quietly skipped the following year.

Sampling answers three questions: which charts, chosen how, by whom. The unit worth reviewing is the episode, not the single note — a treatment note can only be judged against the evaluation, the plan of care, and the notes around it. Internal consistency across those documents is precisely what external reviewers read for, and what a single-note review can never catch.

1. 01

   Fix the monthly count in advance

   One to three episodes per clinician per month, set by team size and written into the review calendar. The number matters less than its constancy: a fixed count survives busy months, while “as many as we have time for” becomes zero by March.

2. 02

   Pull most of the sample at random

   Random selection is what makes the review honest — pulling charts by clinician roster and a random visit date takes minutes in any EHR. Hand-picked charts skew toward episodes someone is already worried about, which tells you about your worries, not your documentation.

3. 03

   Add targeted pulls where risk concentrates

   Reserve a slice of the sample for known risk: a clinician’s first ninety days, a newly joined payer, a new service line like teletherapy, and any visit type that has drawn denials. Targeted pulls answer specific questions; random pulls find the questions you did not know to ask.

4. 04

   Rotate the document types

   A review that only reads daily notes never catches the expired plan of care. Rotate emphasis across the episode: evaluations and their medical-necessity story one month, progress reporting and goal movement the next, discharge summaries and episode closure after that.

5. 05

   Log the pull before anyone reads

   Record which charts were selected, how, and by whom — before review starts. The pull list is what makes the sample defensible later, and it prevents the quiet swap of an awkward chart for an easier one.

A chart review without a written rubric drifts into style critique — whether the reviewer would have phrased the assessment differently — and style critique is exactly what makes peer review feel personal. The rubric below sticks to the questions external reviewers ask, compiled from CMS documentation requirements for outpatient therapy and the documentation guidance published by ASHA, AOTA, and APTA. Every item is a yes-or-no reading of the record itself, so two reviewers looking at the same episode should land on the same answers.

Score each item pass, fail, or not applicable, and require a one-line note for every fail pointing at the specific document — “POC certification expired March 4, visits continued through March 19” — because a finding that cannot be located cannot be fixed. Resist the urge to extend the rubric past ten items. Every addition makes the review slower, and the goal is a rubric the practice still uses in month eleven.

In most small group practices the default is that the owner reviews everything, forever. It fails twice: the owner becomes the bottleneck that makes the rhythm skippable, and review hardens into something done to the team rather than by it. The healthier structure treats reviewing as a clinical skill the whole team practices — reading a chart against a rubric is close kin to the self-editing that makes documentation better in the first place.

Most practices land on a blend: rotating peer pairs as the backbone, cross-discipline reads for the structural items when a discipline has only one clinician, and the owner reviewing a small slice of everyone — including, pointedly, having someone review the owner’s own charts. Nothing establishes that the rubric measures the system rather than rank faster than the owner’s charts taking their turn and their fails going into the same log as everyone else’s. Add one rule and say it out loud: rubric results never feed performance reviews or compensation. The moment scores carry consequences for individuals, clinicians start managing the sample instead of the documentation, and the data quietly rots.

This is the step that decides whether the rhythm survives. A review that ends in a spreadsheet of individual scores produces defensiveness and nothing else. A review that ends in a system change produces better charts next month. The pivot is aggregation: before anyone discusses a single finding, tally the month’s fails by rubric item across the whole team — with the clinician names set aside.

Read that tally with one question: is this pattern a person or a system? A miss that appears across several clinicians’ charts is a system finding by definition — the template does not prompt for total treatment time, the EHR does not surface the certification date anywhere a clinician looks, the intake workflow files the physician order where nobody can find it. Notably, this is the same posture Medicare’s own TPE program takes toward practices: review, then educate, then re-review — not review, then punish. An internal program has even less reason to skip straight to blame.

1. 01

   Aggregate before attributing

   Tally fails by rubric item for the month, names withheld. The distribution across items — not any individual chart — is the month’s actual finding.

2. 02

   Trace each pattern to its system cause

   For every item with multiple fails, name the template field, EHR behavior, or workflow step that permits the miss. If the honest answer is that the system already prevents it and one clinician skipped it anyway, that becomes a private coaching conversation — supportive, specific, and separate from the review meeting.

3. 03

   Change the artifact, not the exhortation

   Reminders decay in a week; templates do not. Add the prompt to the note template, surface the certification date on the schedule, move the co-signature into the signing workflow. One system change per month, done completely, beats five announced at once.

4. 04

   Log the change with an owner and a re-check date

   One line in a running log: the finding, the change, who made it, and which future month’s sample will confirm it worked. The log is also the practice’s answer if anyone ever asks whether it monitors its own documentation.

5. 05

   Close the loop with the team

   Share the aggregate pattern and the fix in the same short meeting — “progress notes were missing total time; the template now asks for it.” Findings that visibly turn into fixes are what make clinicians volunteer problems instead of hiding them.

“

“A chart audit that ends in a score changes nothing. A chart audit that ends in a template change fixes every future note at once.”

The version of this system that fails is always the ambitious one: a fifteen-page audit tool, every chart reviewed, a half-day meeting. The version that survives is small — a fixed monthly pull, ten yes-or-no questions, one system change at a time, one short close-out. If the practice ever faces a real external review, the artifacts this rhythm leaves behind — the pull lists, the rubric, the change log — are also precisely the evidence that the practice takes its documentation seriously, which is the posture federal compliance guidance has been asking small practices to adopt all along.