E/M Distribution Shift After a Template Rollout

Start with a defensible measurement, not with the anecdote of ten charts:

• Draw a statistically valid random sample per provider (typically 30–50 records per provider for a 95%/±10% confidence read), spread across the two quarters, and score each against the current E/M rules (problem complexity, data, risk - or total time if the provider documented time).
• Blind the review: the auditor should not know which provider or which template version produced the note.
• Test three separate hypotheses the data can distinguish:

1. Template artifact - the template pre-populates ROS/exam text, producing cloned documentation with no clinical content behind the level.
2. Education gap - providers genuinely believe the new template's checkboxes equal MDM, and are leveling from data review that does not meet the 'data' element.
3. Deliberate upcoding - distribution shift concentrated in one or two providers rather than across all five.

• Compare the practice's distribution against a peer benchmark and against each provider's own pre-January baseline.

The key judgment: identical ROS text across patients is a documentation integrity finding that nearly always strips the level, regardless of whether the provider actually did the work.

Required response:

1. Quantify the overpayment by claim, not by extrapolation - extrapolated figures may inform a settlement but the refund is calculated on identified claims.
2. Report and return within 60 days of identifying the overpayment, with an explanation of the error.
3. Evaluate self-disclosure if the pattern is systemic across payers and providers rather than a handful of claims.
4. Root-cause the template: remove pre-populated ROS/exam defaults, require the MDM elements to be affirmatively documented, and add a template change-control review before any future rollout.

The CO-97 denials are a different problem that looks like the same one. CO-97 means a code was bundled or not separately payable - a code-edit/modifier issue, often an E/M billed same-day as a procedure without supportable modifier 25, or a procedure bundle. Fixing the E/M level documentation will not fix a modifier edit, and appealing CO-97 denials without amending the record just recycles the same claim. Triage them separately: pull the denied claims, determine whether a modifier was missing and supportable or the service genuinely was bundled, and only appeal where the record supports it.

The rising days in A/R is the downstream symptom of both - denials sitting in appeal queues while staff rework the same accounts.

Monitor leading indicators, not just the lagging revenue number:

• E/M distribution by provider, monthly against each provider's own baseline and the peer benchmark - a return to the prior distribution is the signal the template fix worked.
• Internal audit accuracy rate on a rolling random sample (target ≥95% for E/M level accuracy), reported to the compliance committee quarterly.
• First-pass clean-claim rate and CO-97 denial rate, tracked separately from medical-necessity denials so the two root causes stay visible.
• Days in A/R and percent of A/R over 90 days, to confirm the backlog is clearing rather than being written off.
• Documentation integrity flags - percentage of notes with identical ROS/exam blocks on same-day visits, and percentage of notes signed more than 72 hours after the encounter.
• Template change control: no template modification goes live without a coding/compliance review and a sample of test claims.

Process control: feed audit findings into targeted provider education with a re-audit at 60–90 days, and only escalate to corrective action if a provider's accuracy does not improve after education. Document every step - the paper trail is what turns a payer investigation into a resolved audit.