In recent developments that could potentially affect overpayment findings, an administrative law judge (ALJ) invalidated a Medicare auditor’s statistical sampling method because it removed underpayments, and the chief statistician for a Medicare administrative contractor (MAC) came to a similar conclusion in an unrelated appeal. If their point of view catches on, extrapolated overpayment amounts may be smaller in some cases, experts say.
The ALJ agreed with the appellant, a durable medical equipment (DME) supplier, that failing to include unpaid and underpaid service lines violated the Medicare Program Integrity Manual,[1] according to a June 18 decision. While it sounds technical, the ruling sends a message that Medicare auditors should consider both underpayments and unpaid claims in the extraction and extrapolation of audit samples, said attorney Stephen Bittinger, who represented the appellant. “This is a very important opinion,” said Bittinger, with K&L Gates in Charleston, South Carolina. Leaving out unpaid claims, which have a dollar value of zero, and underpaid claims could “significantly skew” the audit at the starting gate. Other ALJs recently have come to the same conclusion in hearings, he noted.
Statistician/auditor Bruce Truitt, a former faculty member of the Medicaid Integrity Institute in Columbia, South Carolina, said he has been wondering for decades when a case like this would be won. “In a nutshell, all overpayments are improper payments, but not all improper payments are overpayments. Some are underpayments,” he explained. “If you don’t include underpayments, you never get to the correct value of the claim. As a result, the true dollar value, and conceivably the claim count of the universe, is not properly determined.”
Truitt said when auditors start out, they “always maintain the single, unique and complete dollar value of the claim. If I have a claim with three line-items on it, one of which is an unpaid zero dollar value item, one of which is a negative dollar adjustment and the other is a positive dollar amount, leaving out the negative dollar adjustment will affect the total dollar value of the claim. And, if I sample at the line-item level, leaving out the zero dollar unpaid item will affect the total size of the sampling frame from which I pull the sample.”
‘Sampling Size Must Include All Underpayments’
The June ALJ decision, which the attorney shared with RMC, focused on claims submitted by a DME supplier mostly for knee orthoses and spinal orthoses that were denied by a zone program integrity contractor (ZPIC) after a postpayment review. The ZPIC used stratified random sampling, with paid amounts serving as a proxy for stratifying by overpayment amounts, which are unknown before sampling.
After making some headway in its appeal to the qualified independent contractor, the DME supplier took its challenge of the sampling method and extrapolation to the ALJ. Although Dr. Cox, the statistician for the DME supplier, argued there were five reasons why the ZPIC’s sampling methodology was flawed, Bittinger said the unpaid claims argument is the game changer.
According to the ALJ opinion, Cox argued that “regarding the composition of the universe of claims,” Chapter 8 of the Medicare Program Integrity Manual “cannot be interpreted to allow the removal of the unpaid or zero-paid service lines from the universe. As a result, the net overpayment was not considered, only the gross overpayment. Sampling size must include all underpayments and zero paid claims and all must be extrapolated to determine the net overpayment. AdvanceMed [the ZPIC] included claim numbers that had multiple individual codes and dates of service, then they removed the individual codes and dates that were not paid. When they remove zero paid line items, the claims are never audited and there is never the opportunity to review for mistakes and potential payment. By removing the zero paid claims they remove the possibility of finding an underpayment. Medicare requires that these zero paid items must be included and reviewed. Not a valid sample due to the lack of the zero paid claims. Sample not properly designed.”
The ALJ agreed with the statistician on this point and four others. The five reasons “for invalidating the statistical sample are accurate,” ALJ Marc Lambert wrote. As a result, “the statistical sample for the claims at issue is considered invalid.”
