Overpayment recovery audits are becoming more commonplace. However, the process utilized by payers to calculate the overpayment amount is both poorly understood and rarely challenged.

Healthcare consultant Frank Cohen of Frank Cohen and Associates is working to educate providers on the number manipulation game going on during overpayment audits. During a recent presentation on post-audit mitigation, Cohen explained that there are a number of different ways that auditors can bias the statistical analysis against the provider, resulting in a higher overpayment amount than might be statistically justifiable.

Statistical extrapolation is the method of calculating a provider’s overall claim payment error rate based on the error rate of a smaller representative sampling of claim payments. Unfortunately, extrapolation calculations can vary significantly from one audit to the next depending largely on the starting point – the data sample.

Cohen explained that data sampling methods can vary significantly among auditors. He points to a couple of basic indicators which should be assessed to determine the potential for bias against the billing organization.

First, Cohen advised that the statistical sampling be reviewed for outliers. Outlier payments deviate significantly from the normal range of payments and, therefore, often impact the overpayment calculation. Cohen states that he routinely demands that outlier payment be excluded from a statistical extrapolation because such payments are not really representative of overall payment patterns. Cohen also says that it is important to make sure that the statistical sample does not include unpaid claims which, again, are not representative of claim payment amounts and can unfairly skew the overpayment calculations.

Second, Cohen advises comparing the data sample statistical mean with the statistical median. Cohen explains that auditors are typically trying to calculate an “average” payment amount in the statistical sample. Such calculations are more accurate if the data is well stratified within the data sample. However, many billing organizations will find that claim payment data is not necessarily well stratified and that, instead, there will be a concentration of high volume procedures with the remaining data composed of somewhat less representative claims data of less frequently performed procedures. To determine the effect of the poor stratification, Cohen will often calculate both the mean and the median of the statistical sample. If there is a lot of variation between the mean and median, the data may not be well stratified and the lack of overall stratification can be used to negotiate a reduction in the overpayment amount.

Cohen conducts training courses aimed at assisting medical providers identify auditor bias and challenge an unfairly skewed audit extrapolation and has developed a number of resources for post audit mitigation. See a more complete description of his services at www.frankcohengroup.com.

However, the first step the reviewing the audit results is to obtain the data used for the overpayment calculation.

Download our Audit Results Information Request Letter for the recommended wording to use.

## Leave A Response