Posted by Jaime L.M. Jones and Brenna Jenny
Although relators and the government have long leveraged statistical inferences to estimate damages in FCA cases, in two recent opinions, courts have permitted the extension of these approaches to efforts to establish FCA liability. This is a troubling signal for defendants because, particularly when utilized in conjunction with the lower pleading standard of certain circuits, these decisions will make it easier for FCA plaintiffs to fend off a motion to dismiss.
On September 29, 2014, the District of Tennessee significantly expanded the role of statistical sampling in FCA litigation, when it ruled that extrapolation from a small sample can be used not merely to calculate damages, but to establish liability. In U.S. ex rel. Martin v. Life Care Centers of America, Inc., the government alleged that Life Care Centers, which owns a chain of skilled nursing facilities, pressured therapists to overstate the amount and intensity of therapy residents required, resulting in higher daily per diem payments under Medicare Part A. The government did not present the court with any specific examples of patients who received medically unnecessary therapy. Instead, the government sought to select a random sample of 400 Medicare beneficiaries who received high-intensity therapy, examine their medical records to determine whether they any of this therapy was medically unnecessary, and then extrapolate any findings of unnecessary services received by these 400 patients across 54,396 patient admissions, comprising 154,621 claims, to ascertain the number of false claims the defendant submitted.
The defendant strongly disagreed that this novel statistical application could establish falsity or materiality in this case, arguing that the unique nature of each patient’s condition requires an individual assessment of medical necessity, thereby precluding extrapolation. This is particularly so because the per diem payments received by skilled nursing facilities hinge on a patient’s Resource Utilization Group (“RUG”) classification, and even if a portion of a patient’s therapy were medically unnecessary, a patient could remain in the same RUG based on the balance of the necessary portion of his therapy.
The court acknowledged the distinction between using extrapolation to establish damages after liability has already been proven, and using extrapolation to establish liability in the first instance. However, after assessing the case law marshaled by both sides, the court found all to be inapposite and, left to decide the matter in a perceived vacuum, the court determined that the fraud-fighting goals of the FCA would be stymied if the court sided with the defendants and effectively required a “claim-by-claim review” in every FCA suit. The court reasoned that if “Congress intended to preclude statistical sampling from being used in this context, it has had ample opportunity to have that intention reflected in the language of the FCA.” Furthermore, the court viewed defendants as sufficiently protected from specious statistics through other sturdy safeguards, such as the opportunity to cross-examine opposing expert witnesses.
A recent decision in U.S. ex rel. Greenfield v. Medco Health Systems, Inc., demonstrates how reliance on statistical inferences, in conjunction with adoption of the more lenient pleading standard, can resuscitate a qui tam suit that may otherwise struggle to survive a motion to dismiss. The relator alleged that defendants—providers of specialty pharmacy services and hemophilia therapy management programs—tied their charitable donations to hemophilia foundations to the recipients’ patient referrals back to the defendants. In addition, defendants allegedly gave gifts to patients, including Medicare and Medicaid beneficiaries, in order to encourage them to continue to use their services. The court dismissed the second amended complaint without prejudice, ruling that it failed to show that any of defendants’ charitable contributions were tied to federal funds. Indeed because the recipients of the donations used the funds to purchase insurance for financially needy patients, the court ruled that the alleged quid pro quo scheme “demonstrate[s] that defendants’ contributions were used by [the foundations] to avoid the need to avail themselves of any federal benefits program.” The court also concluded the “plaintiff’s math (and his corresponding assumption that federal funds are implicated) is too attenuated and derivative to state a viable claim under the heightened Rule 9(b) standard.”
The relator’s Third Amended Complaint still relied on statistical inferences. For example, the relator concluded that because nationwide, 6% of hemophilia patients are Medicare beneficiaries and one-third are Medicaid beneficiaries, 6% of defendants’ 401 hemophilia patients in New Jersey (24 patients) must be Medicare beneficiaries and 33% (133 patients) must be Medicaid beneficiaries. The relator’s estimates were either widely off—defendants had 53 patients in New Jersey who were Medicaid beneficiaries—or completely unsubstantiated—relator could only show the number of defendants’ Medicare beneficiaries (149) nationwide. Moreover, the relator could not point to any of these Medicare or Medicaid beneficiaries as having received inappropriate gifts from the defendants.
However, the Third Amended Complaint survived defendants’ motion to dismiss in no small part because in the intervening time period, as we previously reported, the Third Circuit Court of Appeals sided with the First, Fifth, Seventh, and Ninth Circuits in adopting a less restrictive pleading standard satisfied by “indicia that lead to a strong inference that claims were actually submitted,” rather than by representative samples of allegedly false claims submitted. The new standard transformed the plaintiff’s “attenuated and derivative math” into “plausible statistical inferences” that adequately pled a “a strong inference that claims were actually submitted for reimbursement for these illegally procured patient prescriptions from federal funds.”
As the Martin court observed, there is little pre-existing case law in this area, and the extent to which statistical sampling can evolve into a tool for establishing FCA liability is a topic we will continue to monitor.