Underpayments are hard to see because the claim looks paid. The shortfall sits inside a contractual adjustment (often CO-45) and never appears on a denial report.
How detection works
- Model the contract: fee schedules by code, plus payment rules for bilateral procedures, multiple procedures, anesthesia units, implants and carve-outs.
- Read every remittance: allowed amount, paid amount and adjustment codes for each line.
- Compare: expected allowed vs actual allowed.
- Flag: variances above a threshold, grouped by payer and pattern.
Patterns it finds in procedural billing
| Pattern | Example |
|---|---|
| Bilateral paid as one side | Modifier 50 line allowed at 100% instead of the contracted bilateral rate |
| Multiple procedure reduction on the wrong line | The highest-value procedure reduced instead of the second |
| Anesthesia time recalculated | Minutes rounded down or capped against the contract |
| Old fee schedule | Payer still paying last year's rates after an update |
| Carve-out ignored | Implant absorbed into a case rate instead of paid separately |
Why AI helps here
A practice billing thousands of lines a month can't compare each one with a contract by hand. Software can, every day, and can group variances so one appeal addresses a whole pattern.
What you need to start
- current payer contracts and fee schedules
- 835 remittance files for the review period
- the contract's reconsideration window, so recent claims are worked first
The human part
Flags are reviewed before anything is sent. Some variances are legitimate (patient eligibility changes, correct edits), and a credible reconsideration includes the contract language that applies.
Related: underpayment recovery and what is an underpayment in medical billing.
This article is general billing information, not legal or coding-compliance advice. Payer rules, CPT descriptors and federal guidance change; check the current versions for your claims.