Extrapolation

Extrapolation is commonly referred to as a statistical sampling for overpayment estimation / extrapolation. A contractor will use a supplier/provider claim history (paid and/or partially paid claims) as part of the post payment review. A sample of claims, a subset of the claim history, is pulled. The sample is considered a statistical sample which is different from a Probe sample. Once reviewed, the sample is used to calculate and extrapolate the sample error rate to the claims universe. The extrapolation calculations determine the overpayment amount.

The methodology used in the statistical sampling must be done by a statistician as defined by CMS Internet Only Manual (IOM), Publication 100-08, Medicare Program Integrity Program, Chapter 8.

On this page, view the process overview below.

Identify Extrapolation

The CMS contractor will send the provider/supplier a letter requesting money due based on a statistical sampling of an overpayment estimation or extrapolation. The initial post payment review letter will inform the provider/supplier that this is an extrapolation. The demand letters are tied to multiple claims and will typically include large dollar amounts.

Two Types of Overpayments

  1. Identified - One claim (or single claim) for one supplier/provider
  2. Extrapolation - Sample of claims used to estimate dollars paid in error for a universe of similar claims for a specific period of time

Post payment review letter will identify the type of overpayment.

Extrapolation Terms and Definitions

  • Universe (population) - Universe consists of all claim detail that is of interest and is defined prior to the selection of a sample. It can include such information as provider number, date range, and procedure codes to be selected, or any relevant criteria depending on situation
  • Sample Unit - Sample unit is level of detail at which randomization and selection of sample take place. It can be claim line, claim, beneficiary, or any other sampling unit appropriate for issue under review
  • Sample Frame - Complete listing of sample units (items). If sample unit is not claim line, sampling frame will be at level at which information is summarized. For example, if sample unit is claim, it will contain totals for claims. If sample unit is beneficiary, it will contain totals for beneficiaries
  • Random Sample (probability sample) - Random sample will consist of randomly selected sample units. Sample size is determined by post payment review contractor. For purposes of random sampling in this audit context, it must have potential to be replicated and meet requirements of probability outlined in CMS IOM, Publication 100-08, Medicare Program Integrity Program, Chapter 8.
  • Extrapolation - After audit review, sample data will be used to estimate overpayment or underpayment for entire sampling frame (and by extension, universe). It is calculation of overpayment estimate using statistical formulas. There are multiple ways to extrapolate error rates from sample to universe. Most of these methods account for error associated with sampling, which is calculated to provider's favor. CMS IOM, Publication 100-08, Medicare Program Integrity Program, Chapter 8 recommends lower bound of a 90% confidence level

Contact

The contractor initiating the post-payment review will contact the provider/supplier. Any review related inquiries must go to that specific contractor. If there are questions on the methodology used to calculate the dollar amounts included, contact the original contractor.

If there are general extrapolation questions, contact the Noridian Contact Center.

Appeal an Extrapolation

A provider/supplier who doesn't agree with the extrapolation may submit a Redetermination request, to Noridian, for all denied or partially denied claims in the sample. All such claims must be included in a single request. It is not acceptable to submit a separate request for each claim as the full sample is required to extrapolate the final amount. Each claim is reviewed and a determination is made. The outcome for all appealed sample claims may positively or negatively impact the provider's/supplier's dollars.

All extrapolated appeals are reviewed by a statistician to assure the methodology was implemented properly and the calculations are correct.

Resources

Last Updated $dateUtil.getDate( $modifieddate , "MMM dd , yyyy" , $locale , $tzone )