The False Claims Act (FCA) is silent about whether statistical sampling may be used to prove a violation of the Act and, since the case law is varied, there are no clear rules about when or how statistical sampling evidence may be introduced. Two cases—the Fourth Circuit’s recent decision in Agape and the Supreme Court’s decision last year in Tyson, a wage and hour class action—show that litigants will probably never obtain a bright-line rule on the use of sampling evidence to prove liability. Instead, decisions will turn on the specific facts of each case and the validity of the sampling methodology.

In the context of the FCA, courts are generally willing to permit the use of statistical sampling to prove damages where liability is not contested or has already been established.1 This is consistent with decisions in other complex cases, where expert sampling evidence has enjoyed widespread acceptance in state and federal courts as the basis to prove damages.2 However, courts are split about the use of sampling to prove liability under the FCA. In some cases, courts have rejected the use of sample extrapolation to prove a violation, finding that the falsity element under the FCA cannot be proven through expert medical necessity reviews that involve subjective clinical judgment.3 In other cases, courts have allowed the use of sampling to establish liability, holding that any challenge to the sample could be effectively addressed through cross-examination of the government’s expert and presentation of a competing expert and evidence.4

In prior cases, defendants argued in favor of a per se prohibition on the use of sampling to prove liability, claiming it constituted a violation of due process as a matter of law. But, as Agape and Tyson show, defendants must now instead focus on the specific facts of their cases, as there is no clear prohibition against the use of sampling evidence.

Tyson Foods, Inc. v. Bouaphakeo

Indeed, it was the Supreme Court, outside of the FCA, that last year refused to create a blanket prohibition on the use of sampling evidence to prove liability in the class action context.5 In Tyson Foods, Inc. v. Bouaphakeo, employees at a meat processing plant filed a putative class action against Tyson under the Fair Labor Standards Act, alleging the company failed to pay them overtime for time spent “donning and doffing” their protective gear. The company did not maintain any records of how long each employee spent performing this task each day.

To prove liability, the plaintiffs retained an expert to estimate the average time employees in different departments spent on the task and another expert compared those averages to each employee’s recorded hours to identify alleged violations of the FCA. The key dispute in the case was whether differences in the time employees spent “donning and doffing” gear made the plaintiffs’ expert evidence too speculative to prove classwide liability.

The Supreme Court found that the employees were sufficiently similarly situated, and it refused to create a “categorical exclusion” or ban on statistical sampling to establish liability, explaining the use of sampling evidence would hinge, like other evidence, on its reliability. As the Court explained, the “fairness and utility of statistical methods in contexts other than those presented here will depend on facts and circumstances particular to those cases.”

United States ex rel. Michaels v. Agape Senior Community, Inc.

Earlier this year, in United States ex rel. Michaels v. Agape Senior Community, Inc., the Fourth Circuit addressed a similar question in the FCA context.6 (In this case, the relators alleged that the defendant caused the submission of false claims for hospice reimbursement. The Medicare regulations governing the hospice benefit require physicians to certify that the patient seeking the benefit has a terminal illness with a prognosis of surviving six months or less. At the district court, the relators had tried to use statistical sampling to establish liability.)

The FCA defense bar highly anticipated the decision, hoping it would provide a bright-line prohibition against the use of sampling to prove liability under the FCA. When presented with the issue, however, the Fourth Circuit too took a fact-specific approach and refused to create a categorical exclusion against its use. The Court dismissed the relator’s appeal on the issue as “improvidently granted,” holding that interlocutory review was not appropriate since the statistical sampling is not a “pure question of law.” In other words, sampling may or may not be appropriate to prove liability; it depends on the specific facts of the case.

Conclusion

How does the Court’s guidance in Tyson help us understand when statistical sampling may be used in FCA cases? In FCA cases, sampling evidence is typically used to challenge the medical necessity of the claims for reimbursement submitted to the government. In Tyson, the Supreme Court emphasized that the expert’s testimony regarding average times “did not deprive [Tyson] of its ability to litigate individual defenses,” and thus did not violate due process, because the company did not maintain any records, so “there were no alternative means for the employees to establish their hours worked.” That reasoning would arguably not apply to many FCA cases in which the provider maintains the medical records that support its claims for reimbursement and where the use of statistical sampling evidence would arguably deprive the defendant of the right to present individual defenses to each claim.

Despite hopes that Tyson would provide a definitive answer, there continues to be no blanket prohibition against the use of statistical sampling to prove liability. Therefore, it is critical to consider the specific facts and circumstances of each individual case to determine the appropriateness of using statistical sampling, including the reliability of the sampling evidence, whether there are any potential violations of due process and whether there are other options for obtaining accurate information.