Author: Knicole C. Emanuel Esq. | October 21, 2026
Every extrapolated overpayment demand rests on a quiet assumption: that the right payment for a claim is a single, knowable number. Take that assumption away, and the math behind many audit demands falls apart.
The Variable-Rate Problem
Reimbursement for a single CPT® code is not fixed. Fee schedules are updated every year. Contract rates differ from plan to plan. Even within one payer, the same service can be paid at very different amounts, depending on the product, the provider type, the place of service, or the date.
Consider a recent commercial payer audit in which the payer reviewed 60 sampled claims for CPT 99214, a routine established-patient office visit. Those 60 claims were paid at seven different allowable amounts: $155.80, $199.62, $191.89, $187.49, $122.93, $117.57, and $114.91. That is one code, one provider, and one audit, with a spread of nearly 74 percent between the highest and lowest payment.
This example comes from a commercial plan, but the same dynamic exists in Medicare and Medicaid. Physician fee schedules are revised annually, geographic adjustments shift, and state Medicaid rates move with budgets and policy.
The Extrapolation Flaw
Auditors can work the same way. They review a sample, find claims they consider unsupported, reprice those claims, and then project the error rate across a universe of thousands of claims spanning several years. The problem is in the repricing step. Many methodologies apply one adjustment, such as a uniform percentage reduction or a single “correct” rate, across the entire universe.
When the underlying payments vary as widely as the example above, a uniform approach does not measure an overpayment. It produces a number that no claim in the universe actually reflects. Errors in the sample are multiplied across every extrapolated claim, and a small procedural shortcut becomes a six-figure demand.
The Fairness Argument
Recovery Audit Contractor (RAC), Uniform Program Integrity Contractor (UPIC), and Medicare Administrative Contractor (MAC) audits routinely reach back across multiple years of dates of service. Over that span, fee schedules, contract terms, and reimbursement rates almost certainly changed. A claim from January of one year and a claim from December of the next may have had very different correct payment amounts.
An extrapolation that ignores those changes treats the audit period as a frozen moment that never existed. The result is presented with statistical precision, including confidence levels and sampling frames, but if the rates feeding the model are wrong, that precision is false. It is arithmetic error dressed up as science.
What Providers Should Demand
Extrapolation is not automatically illegitimate. It is, however, only as reliable as its inputs, and providers are entitled to see those inputs. When facing an extrapolated demand, providers and their counsel should insist on:
- The specific rates applied to each sampled claim, both the original allowed amount and the repriced amount;
- Whether rate changes across the audit period were accounted for, and how;
- Which plans, products, or programs were included in the claims universe; and
- The full statistical methodology, including sampling design, confidence intervals, and precision.
If the auditor cannot show that its model reflects the rates that actually applied to each claim, the demand deserves serious challenge.
The Takeaway
An extrapolated overpayment is a projection, not a finding. When reimbursement for the same code varies by year, plan, and payer, and the methodology ignores that variation, the projection says more about the auditor’s shortcuts than about the provider’s billing. Extrapolation without a rate-adjusted methodology is not a reliable measure of overpayment, and providers should not accept it as one.
This article was originally published on RACmonitor.
