When Should Labs Fight Overpayment Determinations Based on Extrapolation of Claims?

A recent district court decision on overpayments based on extrapolation of claims contains important lessons for health care providers, including clinical laboratories, on how and when to challenge overpayment requests. The U.S. District Court of Nebraska on Jan. 22 ruled on an overpayment based on an extrapolation of a sampling of claims. The court denied the plaintiff’s assertion that the extrapolation was invalid even though two separate statistical experts opined that it was unreliable and should not be used for extrapolating to the larger universe of claims. In upholding a motion for summary disposition in favor of the Department of Health and Human Services, the court said, “A court reviewing an agency’s adjudicative action should accept the agency’s factual findings if those findings are supported by substantial evidence on the record as a whole.” In the case of a clinical laboratory where thousands of patients and tests are often involved, overpayments are most often based on an extrapolation from a relatively small sample. History of Case Schuldt Chiropractic Wellness Center (Hasting, Neb.) is a provider of chiropractic services for Medicare beneficiaries. Wisconsin Physician Services (WPS) is a Zone Program Integrity Contractor (ZPIC) that conducted an expanded, postpayment review of claims submitted by Schuldt on 75 Medicare beneficiaries, with dates of service from January 2008 through April 2010. The review was conducted on a statistical sample of claims, a common auditing practice, and caused WPS to conclude that there was a 99.55 percent error rate. Based on that, WPS projected overpayment findings of $126,041.13 extrapolated from the $11,376.14 of actual overpayments in the audit sample. Schuldt exercised its appeal options of redetermination and reconsideration, to no avail, and requested an administrative law judge (ALJ) hearing. During the Sept. 19, 2011, telephone hearing, ALJ William J. Cowan found that 344 of the 445 services billed were correctly billed based on the issues of coverage, liability, and overpayment waivers and Schuldt should be paid for them. The ALJ also considered the opinions of two statistical experts, Schuldt’s and the ALJ’s own expert, and concluded that the ZPIC’s results “were insufficiently reliable to be used for the purpose of estimating an overpayment to a larger universe than the sample itself.” On Aug. 6, 2012, the Centers for Medicare and Medicaid Services (CMS) filed a referral memorandum with the Medicare Appeals Council (MAC) asserting that the ALJ decision contained an error of law material to the outcome of the claim. Based on the single issue of whether Schuldt met its burden of proving that the statistical sampling methodology was invalid, the MAC reversed the ALJs decision in respect to the validity of the statistical sampling methodology and did not address the ALJs findings with respect to coverage, liability, or overpayment waiver. On Jan. 31, 2013, WPS issued a recalculation of the projected overpayment applying the ALJs findings with respect to coverage, liability, and overpayment waiver to the sample of 445 billed services and came up with a $37,580 overpayment, which Schuldt paid. Presumption of Validity Schuldt was not wrong to challenge the sampling and extrapolation, but if its appeal had been based solely on that challenge, it would have failed and the $126,041.13 overpayment would have been upheld. Schuldt got its overpayment amount overturned based on the coverage and liability findings of the ALJ, which reduced the number of claims in error and consequently the amount owed. CMS may have wanted to establish the validity of extrapolation for overpayment purposes using smaller samples for future reference. The court ruling explains that sampling creates a presumption of validity as to the amount of an overpayment. This has the effect of shifting the burden to the provider to disprove the validity of the extrapolation. The key element is the error rate. Once the error rate is determined to be less than 99.55 percent, the extrapolation calculations change. Effectively, Schuldt most likely could have obtained the reduced overpayment amount based on the reduction in the error rate that changed the number of claims for services that the ALJ found to be actually incorrect based on coverage, liability, and overpayment waivers. One other important message from this case is that the Medicare manual instructions allow the use of smaller samples and less precise results in certain circumstances and notes that 42 C.F.R. 405.1062 requires that ALJs and the MAC give substantial deference to manual instructions. What We Learned From This Case What this case really provides that is valuable to laboratory compliance officers and other health care providers is guidance on what needs to be done to attack the validity of the sampling and extrapolation used by a Medicare auditor, and if that is a feasible approach to get overpayments reduced or overturned. First, it should be noted that attacking the accuracy of the extrapolation is a difficult endeavor and often fails. A provider would have to present evidence that a different random sample from the universe of claims produces a lower error rate or rate of denials. That would require a new sampling and redetermination of the error rate in the new sample. Another option would be to review the entire universe of claims from which the sample was taken to establish the validity of the error rate of all of its actual claims to demonstrate that the auditors’ projection is factually impossible. Both of these options are onerous and potentially costly. The provider really needs to attack the error rate in the sample. By challenging the errors in the sample based on coverage and other Medicare regulations, the provider can change the overpayment estimation calculation because the error rate is a key element in the extrapolation calculation. Takeaway: When faced with an overpayment based on an extrapolation of a sample of claims, challenging the statistical correctness of the auditor’s methods and sample size should be the challenge of last resort, used only if there are no other options available and the provider has the ability to resample from the universe of claims.