What can be done about an Audit Extrapolation.

By Scott Wilson, CPA, CFE, MBA & Frank Alt, Ph.D.

We have recently seen a trend of payers sending demand letters to providers for substantial amounts which are largely attributed to an extrapolated amount.   An extrapolation is essentially projecting an error rate found in a sample to an ENTIRE population.   With the operating margins for so many providers under immense pressure from so many external forces ranging from regulatory burdens to financial tightening, an audit finding of a 10% (or more) error rate in a sample to an entire population can translate into an overpayment demand in the hundreds of thousands of dollars (or more, much more).

While there are no generally accepted statistical principles that govern extrapolation per se, statistical sampling has been used by the Medicare program since the early 1970s as an effective and acceptable methodology of estimating the amount of overpayments in a population.  By using samples and extrapolations, regulatory auditors can avoid the costly and prohibitive alternative of examining every item in the population.  Chapter 8 of the Medicare Program Integrity Manual (MPIM) includes guidelines for data analysis, statistical sampling, extrapolation and estimating overpayments which are widely used and adhered to within the healthcare industry.

So what does all this mean?  Simply put, it means, as a provider of healthcare services, if you are faced with an overpayment demand letter from an insurance company (or another party representing one) based on an extrapolation of an audit sample error rate, there are ways to challenge and overcome the extrapolation and the alleged overpayment.

Some of the areas in an extrapolation that lend themselves to challenge are as follows:

• Precision and confidence – this relates to the level of uncertainty surrounding the point estimate of the extrapolation.  The MPIM has clear guidance concerning this.  We have used this as a strategy for challenging the extrapolation.
• Sample size – if the sample size if too small, the sample value could have too much sampling error resulting in a sample value inappropriate for extrapolation.  This is another area that we have used to challenge the extrapolation.
• Random and representative – in order for a sample result to be valid, there must not be selection bias in the sample selection such that the sample is representative of the entire population.

The preceding list is not all-inclusive.  The most important thing to keep in mind if you find yourself facing an extrapolation is to consult an expert who is knowledgeable regarding the nuances of statistical science.  If you are currently facing an overpayment demand as a result of an audit sample extrapolation or are interested in obtaining more information, contact Scott Wilson, MBA, CPA, CFE, at swilson@chief-financial-solutions.com or call him directly 443-325-7227.