Article Summary

In U.S. ex rel. Michaels v. Agape Senior Community Inc., the Fourth Circuit will tackle the proper use of statistical evidence in False Claims Act cases. Arguments will be informed by the Supreme Court's recent opinion in Tyson Foods, Inc. v. Bouaphakeo, which dealt with a similar issue under the Fair Labor Standards Act. The outcome likely will help whistleblowers who reveal a common type of Medicare fraud.

This expert analysis by TELG managing principal R. Scott Oswald and former principal David L. Scher was published by Law360 on October 18, 2016.

Originally published in:

Biggest Test Yet For Statistical Sampling In FCA Cases

On Oct. 26, 2016, the Fourth Circuit will become the highest court yet to hear full arguments on the proper use of statistical sampling in False Claims Act cases that involve a defendant’s submission of thousands — or even millions — of questionable bills to the federal government.

A March 2016 decision from the U.S. Supreme Court will loom large in the debate: In Tyson Foods, Inc. v. Bouaphakeo, a case under the Fair Labor Standards Act, Justice Anthony Kennedy’s 6-2 opinion rejected any blanket prohibition of statistical evidence, and endorsed instead a case-by-case examination of whether sampling is “the only practicable means” to establish a defendant’s liability. In such cases, said Justice Kennedy, courts should not frustrate the FLSA’s remedial purpose by insisting on proof of the “precise extent” of liability if doing so would erect “an impossible hurdle” for plaintiffs.

If extended to the FCA, the logic of Tyson Foods would demand a statistical approach to many cases that allege large-scale fraud against the government’s Medicare insurance program, where a pattern of overbilling might cover literally millions of payment requests that cannot, as a practical matter, be examined individually. The Fourth Circuit case, U.S. ex rel. Michaels v. Agape Senior Community Inc., provides a modest but powerful illustration of this dilemma.

In Michaels, two whistleblowers challenged tens of thousands of insurance claims by their former employer, a network of nursing homes in South Carolina. The parties agreed that each disputed claim could, in theory, be analyzed individually — and U.S. District Judge Joseph F. Anderson Jr. ruled that this fact precluded any statistical shortcuts, saying that sampling evidence isn’t allowed under the FCA unless it is “the only way” for relators (whistleblowers who file suit on behalf of taxpayers) to prove their allegations.

Still, Judge Anderson immediately certified his decision for review because he recognized that a claim-by-claim inquiry would force “a staggering outlay of expenses by the [relators] and a significant drain of [court] resources.” Pretrial costs alone might exceed $35 million, he noted, more than $10 million above the government’s estimate of recoverable damages — an amount that was, in turn, far higher than a settlement amount that was agreed by the parties but rejected by the government.

Michaels is a complicated case, and the Fourth Circuit could decide it without reaching the question of sampling. Still, it promises the highest-level debate yet about the validity of statistical approaches to FCA claims — and the decision will serve as an early indicator of the vitality of Tyson Foods, which could extend well beyond the FLSA.

Michaels: The Purist’s Dilemma

In the order now under appeal in Michaels, which reached the Fourth Circuit before Tyson Foods was decided, Judge Anderson lays out the FCA sampling issue with clarity. The Michaels relators, alleging “a widespread fraudulent scheme” in billing for hospice and other services by Agape, the defendant, had proposed to examine a random sample of insurance claims made by the medical network. The percentage of fraudulent claims then would be projected to the “total universe” of up to 61,643 payment requests to determine the scope of Agape’s potential liability. Pragmatic district courts have allowed such a statistical approach, but Judge Anderson opted for purism, siding with the defendants to find that “each and every claim at issue in this case is fact-dependent and wholly unrelated to each and every other claim”:

Distilled to its essence, each claim asserted here presents the question of whether certain services furnished to nursing home patients were medically necessary. Answering that question for each of the patients involved in this action is highly fact-intensive inquiry involving medical testimony after a thorough review of the detailed medical chart of each individual patient.

While some FCA cases might be suited for statistical sampling, said the judge — for example, if evidence has dissipated, “rendering direct proof of damages impossible” — Michaels “is not such a case” because every patient’s medical chart is “intact and available for review.” He rejected any idea that the disputed claims could be analyzed as a pattern.

Judge Anderson acknowledged that his bright-line approach could result in “a trial of monumental proportions” and huge outlays that might never be recouped. Having experts review the medical facts, for instance, might cost relators as much as $3,600 per patient, he said. The perverse result: Schemes that fleece taxpayers of millions of dollars via an accumulation of small overcharges, a staple of Medicare fraud, could move wholly beyond the reach of the FCA.

The judge labored to avoid such a conclusion by approving a few dozen claims for a so-called bellwether trial, a move that successfully pushed the parties into mediation. But the U.S. government, which previously had declined to intervene in Michaels, forced the issue by unexpectedly vetoing the resulting settlement. A frustrated Judge Anderson, having tied his own hands, sua sponte “implored” the parties to ask the Fourth Circuit to untangle the matter.

Tyson Foods: A Reasonable Balance

While the Michaels parties were briefing the Fourth Circuit, the Supreme Court released its Tyson Foods decision. Unlike Michaels, it shunned bright lines. The difference boiled down to a single word: Under Michaels, sampling is generally disallowed unless it is the “only” way to prove FCA liability — but under Tyson Foods, sampling may be allowed if it is the “only practicable” way to prove FLSA liability (emphasis added).

Tyson Foods concerned the failure of Tyson Foods, the giant meat processor, to pay proper overtime to a class of more than 3,000 hourly employees at its hog-slaughtering plant in Storm Lake, Iowa. A jury had awarded the workers a total of $2.9 million in damages, based partly on an extrapolation from sampling evidence offered by a time-and-motion expert. Tyson Foods appealed the verdict, arguing that such extrapolation can’t prove liability or damages owed to any individual worker — and complaining that the method would prevent it from identifying instances where damages weren’t owed.

In response, Justice Kennedy prescribed a case-by-case inquiry into the permissibility of statistical sampling, saying that a blanket prohibition in class actions “would make little sense.” He continued:

A representative or statistical sample, like all evidence, is a means to establish or defend against liability. Its permissibility turns not on the form a proceeding takes — be it a class or individual action — but on the degree to which the evidence is reliable in proving or disproving the elements of the relevant cause of action.

Invoking the landmark 1946 case Anderson v. Mt. Clemens Pottery Co., Justice Kennedy said that a proper inquiry into statistical sampling will weigh available alternatives along with the “remedial nature” of the underlying law and the “great public policy which it embodies,” and then adjust evidentiary burdens accordingly.

Notably, neither Tyson Foods nor Mt. Clemens involved any direct evidence that could show FLSA liability in individual instances: In both cases, sampling wasn’t just the “only practicable” way to prove the plaintiffs’ claims, it also was the “only” way. Yet the Supreme Court chose not to limit its rulings to such bright-line situations, twice opting instead for a fact-heavy test that weighs statutory intent and the likelihood of a just outcome. Justice Kennedy further indicated that, once a district court finds statistical evidence to be admissible and scientifically valid — a standard established by the Supreme Court’s 1993 ruling in Daubert v. Merrell Dow Pharmaceuticals Inc. — the role of sampling in a liability determination is “the near-exclusive province of the jury.”

Applicability to FCA and Michaels

None of this means, of course, that statistical evidence must be admitted in Michaels or in any other FCA case. Applied to the facts of Michaels, however, Tyson Foods strongly suggests such an outcome. Like the FLSA, the FCA is a remedial statute. And while the parties in Michaels agreed that tens of thousands of disputed Medicare billings could be examined individually, Judge Anderson’s order made clear that such a requirement is not “practicable” and easily could kill the relators’ claims — no matter how valid.

Apart from the government’s unusual settlement veto, Michaels is a large-scale but typical case of nickel-and-dime Medicare fraud under the FCA. In such cases, proving liability patient-by-patient is, as in Mt. Clemens, “an impossible burden” for plaintiffs because it requires huge upfront investment. It also takes no notice of the underlying fraudulent pattern: Today’s most noxious Medicare fraudsters ignore their patients’ individual needs in favor of rote care that is designed to maximize repayment — ordering certain therapies for every patient, for instance, regardless of the facts. To ask plaintiffs to pay upfront for countless proper medical diagnoses is an unjust asymmetry, and could turn the FCA into a dead letter for an entire category of wrongdoing.

Judge Anderson was hardly blind to the danger that a ban on sampling “would allow widespread fraud to go unpunished,” as he noted. Indeed, he said he likely would have permitted statistical evidence in a recent non-Medicare FCA case, but the parties settled before he had a chance. That case was an easier call, however: As in Tyson Foods and Mt. Clemens, sampling was the “only way” to reconstruct what had happened.

Other courts have allowed the use of sampling in FCA litigation, even when claims could be examined individually. Judge Anderson listed 10 supportive cases, including United States v. Rogan, a case in which the Seventh Circuit rejected any claim-by-claim requirement as “a formula for paralysis,” albeit in dicta. But Michaels also cited six cases that seemed to suggest the opposite, including a more tangential Seventh Circuit decision and United States v. Friedman, a 1993 case from the U.S. District Court for the District of Massachusetts in which the judge, in a footnote, acknowledged the validity of sampling but said he was nonetheless “reluctant” to accept it.

In the end, said Judge Anderson, “the cases are legion on each side of the issue, and ultimately, it is this court’s responsibility to determine the fairest course of action based upon the facts presented and the claims asserted in this case.” True enough — except he then used the Friedman footnote to construct a hard-and-fast rule based on “the existence at trial of discrete claims that may be analyzed, discussed and subjected to cross examination.” Any facts about the cost of such analysis, he implied, are irrelevant to fairness.

Even if the Fourth Circuit reverses Judge Anderson and sets a looser rule for allowing statistical evidence in FCA cases, the sampling question will likely remain contentious — at least until another appellate court weighs in. A robust application of Tyson Foods, however, should serve as the keystone for an emerging consensus: In FCA cases with an aggregation of small frauds against the government, trial judges should admit statistical evidence that meets the Daubert standard — and then stand back and allow their juries to draw the appropriate conclusion, exactly as Justice Kennedy recommends.


R. Scott Oswald is the managing principal at The Employment Law Group, P.C. David Scher is a former TELG principal.