Lies and Statistics: Statistical Sampling in Liability Determinations Under the False Claims Act.

AuthorKennedy, Patrick

Table of Contents Introduction I. Due Process Problems with Sampling A. The Statutory Framework B. Sampling: A Solution with Its Own Problems C. Statutory Analysis II. The FCA: Patchwork Enforcement, Deterrence Problems, and Misaligned Incentives III. Evaluating the Due Process Problem with Sampling A. Private Interests B. Risk of Error C. Interests of the Government or Relator 1. Direct financial interests 2. Signaling and deterrence interests 3. Negative impacts on government and relator interests D. Balancing IV. Guidelines for Implementing Statistical Analysis A. Modified Bellwether Option B. Class Action Analog Conclusion Introduction

Few issues incite furor and partisanship in the present political climate more than federal health care spending. Indeed, in recent years Republican leaders have stated that Congress should place Medicare spending in its crosshairs, citing growing concerns about cost and debt. (1) While cutting the program is one way to reduce costs, fixing systemic difficulties in prosecuting Medicare fraud could also save the government billions without sacrificing coverage for senior citizens. According to the Centers for Medicare and Medicaid Services (CMS), fraudulent submissions to Medicare cost taxpayers billions of dollars per year. (2) In order to protect the public from this rampant abuse, as well as other fraudulent schemes, Congress has passed a suite of statutes designed to punish fraud, graft, and corruption. (3) The False Claims Act (FCA) is one such statute. (4) The FCA creates civil liability for any person who "knowingly presents, or causes to be presented, a false or fraudulent claim for payment or approval," or "knowingly makes, uses, or causes to be made or used, a false record or statement material to a false or fraudulent claim." (5) This statutory provision is often used to file suits combating Medicare fraud, (6) and represents a potential solution to the Medicare crisis.

The specter of big data produces anxiety in the legal community, but fears about "trial by statistics" are exaggerated. Individualized adjudication is a bedrock of American jurisprudence, and the U.S. Supreme Court has repeatedly recognized this principle to be protected by the Due Process Clause. (7) Statistical forecasting techniques use virtually the opposite approach to traditional legal analysis, focusing on patterns and relationships between a number of similar situations to construct inferences rather than making individualized findings.

Perhaps unsurprisingly, the Supreme Court has looked askance at plaintiffs' efforts to introduce statistical evidence to prove liability in class action cases. For example, the Court rejected statistical evidence about Wal-Mart's discriminatory labor practices in the landmark case Wal-Mart Stores, Inc. v. Dukes, (8) seemingly foreclosing class action plaintiffs' ability to submit statistical proof during the liability phase of a trial. (9) But just five years later, the Court limited its holding in Dukes, holding in Tyson Foods, Inc. v. Bouaphakeo that class action plaintiffs may introduce statistical inference evidence when the evidence would prove an element of the claim in each individual plaintiffs trial. (10) While the Court relaxed its position on individualized adjudication to some extent, it carefully limited its holding in Bouaphakeo to just a subset of class action cases, leaving open the question whether plaintiffs could use statistical inference in any other context. (11)

Demonstrating hospitals', health care facilities', or physicians' liability for Medicare fraud under the FCA one payment at a time is unworkable where tens of thousands of Medicare payments are at issue, leading plaintiffs to attempt to introduce statistical evidence to prove liability. (12) So far, these attempts have met with mixed success, since the Supreme Court's precedent remains fractured and plaintiffs do not file FCA claims as class actions. (13) Unfortunately, the lower courts have also rarely and inconsistently considered the application of statistical analysis to FCA liability, leaving open large questions about the viability of statistical techniques in such cases.

Unfortunately, as it stands now, the Department of Justice (DOJ) finds it exceedingly difficult to prosecute more than a handful of Medicare fraud cases when care providers habitually submit similar fraudulent claims. This leaves serial abusers free to squeeze the federal coffers, with little chance of punishment. (14) While there is a qui tam (15) provision allowing private plaintiffs to step in and prosecute Medicare fraud on the government's behalf, (16) actually prosecuting such cases can cost more than a plaintiff would expect to recover. This is because an expert, usually a physician, must examine each purportedly fraudulent submission. (17) Needless to say, this analysis is costly and may quickly swamp a qui tam plaintiffs expected gains, (18) particularly in the common event that the defendant's alleged fraud was programmatic and continual. (19) Nevertheless, randomly sampling from a large pool of potentially fraudulent claims to determine a hospital's or physician group's probability of liability presents a potential solution. (20) Given the Supreme Court's mixed messages in Dukes and Bouaphakeo, lower courts have little guidance in deciding whether to allow sampling. (21) Unfortunately, sampling also presents a major problem under the Due Process Clause: Can a defendant be made to pay without a finder of fact determining actual liability for each count under the FCA? (22)

Ultimately, these due process issues, discussed in Part I, present a clear set of questions that the courts should answer more definitively. First, should liability in FCA cases be proven for each claim, or does statistical sampling of a portion of claims suffice? Second, if sampling is allowed, under what conditions should it be allowed?

The rest of this Note addresses these questions, concluding that sampling is appropriate in cases against a single defendant with large numbers of claims at issue where the alleged fraud is systematic and the variability between claims is relatively limited. Part II unpacks the requirements for bringing a case under the FCA, as well as the application of due process precedent to FCA cases. Part III examines the characteristics of the cases where questions about sampling arise. Part IV discusses the procedural due process interests at stake in allowing statistical extrapolation. Part V discusses options for legal rules in light of this evaluation of the process due to defendants, examining the Mathews v. Eldridge balancing factors. (23)

While commentators are increasingly grappling with these issues, (24) consideration of the actual process a statistical analyst would use to assess liability in FCA claims demonstrates that sampling need not offend due process. Perhaps unsurprisingly, more abstract considerations of "justice" (25) and high-level discussions of sampling (26) conclude that sampling violates due process or does not comport with the scienter requirement in the FCA. (27) In contrast, I consider the question from the perspective of a specific forecasting strategy, which is how a well-informed court determining the constitutionality of extrapolation would approach the question. While this Note largely focuses on due process considerations, much of the analysis also responds to the claim that the FCA bars statistical analysis. (28)

  1. Due Process Problems with Sampling

    Medicare fraud is usually prosecuted under the FCA because the statute makes unlawful the knowing submission of fraudulent or false claims to the government. (29) This Part unpacks the systemic problems hampering effective civil actions and deterrence under the statute, as well as the due process concerns with the existing solutions.

    1. The Statutory Framework

      A claim under the FCA proceeds in four parts: The plaintiff must identify specific claims made to the government, prove the claims or records are false or fraudulent, prove the defendant's knowledge of falsity, and, finally, prove the fraud or falsity to be material to the government's decision to pay the claims. (30) The Supreme Court recently clarified that the materiality requirement is meant to ensure that FCA claims go beyond "garden-variety breaches of contract or regulatory violations," and that violating minor or irrelevant regulations would not constitute material fraud. (31) Traditionally, each false claim allegation is proven separately, with specific documentary or expert evidence. (32) Since the statute contains a scienter requirement, plaintiffs must prove actual or constructive knowledge. (33) Proving knowledge of fraud in Medicare reimbursement cases often requires expert testimony to show that a provider should not have ordered a procedure submitted for reimbursement, or did not perform the tests or procedures claimed. (34)

      The FCA contains two different enforcement mechanisms for pursuing those who knowingly submit fraudulent claims for government reimbursement or payment: First, the government can directly file a complaint against the tortfeasors; (35) second, there is a qui tam provision, allowing whistleblowers (termed "relators") to file suit and collect a portion of the damages, usually between 15% and 30%, depending in part on whether the government intervenes in the suit. (36) Since qui tam plaintiffs "step into the shoes" of the government, the government can choose to intervene within sixty days to take control of a privately initiated claim. (37) In either case, victorious plaintiffs receive treble damages, (38) in line with Congress's attempt to encourage suits and discourage fraud.

      Some alleged FCA violations, particularly those in Medicare cases, involve patterns of fraudulent claims, often running into the tens of thousands. (39) For example, a hospital may routinely send patients with simple claims for unnecessary...

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT