A New Jersey internal medicine practice adds 8.6% to E/M payment and 23% more diagnosis codes
A multi-provider internal medicine and weight-management practice in New Jersey ran every visit from nine weeks through Medical Copilot. The system coded each visit from the chart and compared the result with what the practice had billed.
The starting point
Providers picked their own E/M level and diagnosis codes from the EHR at the end of each visit. Charts were thorough, but the claim did not always reflect them. Chronic conditions that were managed were left off, levels were set by habit rather than medical decision making, and some charged services had no matching note.
Outside-lab orders went to the reference lab with no diagnosis attached. Every order still needs one to justify it.
What the system did
- Read the full encounter: assessment and plan, history, orders, results and medication changes.
- Set the E/M level from documented medical decision making: problems, data reviewed and risk.
- Coded every condition the visit addressed, at the specificity the chart supports, and removed codes the visit did not support.
- Sent the physician a short query only where the note could not support a code or a charge as written.
Results
| Measure | Billed | Coded | Change |
|---|---|---|---|
| E/M (672 visits with a billed level) | |||
| E/M payment, 2026 Medicare rates | $71,953 | $78,177 | +8.6% |
| Visits raised to a supported level | – | 212 | – |
| Visits lowered | – | 58 | – |
| Visits confirmed at the billed level | – | 358 | – |
| Diagnoses (1,123 visits) | |||
| Diagnosis codes on the claim | 4,926 | 6,082 | +23% |
| New conditions captured | – | 1,313 | – |
| Codes made more specific or corrected | – | 299 | – |
| Unsupported codes removed | – | 153 | – |
| Visits with at least one diagnosis added | – | 620 | 55% |
At this practice's volume, the E/M gain alone is about $36,000 a year at Medicare rates. That is a floor: it counts only visits whose E/M charge was already in the chart when they were coded.
Better documentation, not only better codes
Where the note could not support a code or a charge, the physician got one plain question inside their workflow. 14% of visits needed a diagnosis query and 6% a service query.
| What the query asked | Queries |
|---|---|
| Diagnosis · 239 queries on 185 visits | |
| Reason for an ordered test, so each order carries its diagnosis | 156 |
| Condition a medication treats | 39 |
| Abnormal or critical result the note did not address | 19 |
| Mental-health safety or an unassessed symptom | 4 |
| Other | 21 |
| Services and E/M · 79 queries on 76 visits | |
| Charged service the note does not document | 27 |
| Counseling type, or Medicare ABN for a counseling line | 22 |
| Medication decision that sets the E/M level | 12 |
| Service detail such as CGM supply, test method or route | 12 |
| Problem work beyond the preventive visit | 6 |
27 charges had no matching note. Catching them before the claim goes out protects the practice in an audit as much as the added codes raise payment.
Method: every visit from September 1 to November 2, 2026 was coded again on the current release, and the coded result is compared with what the practice billed. Payment uses 2026 Medicare national non-facility RVUs and conversion factor. Actual amounts depend on payer mix. Visits waiting on a query keep their billed level. The practice is not named.
See what your claims are missing
We run your recent visits through Medical Copilot and show you the same billed-versus-coded comparison.
Book a demo