The issue of using statistical sampling in federal False Claims Act (FCA) cases has come to the fore in the 4th Circuit Court of Appeals, following a U.S. District Court decision denying its use in a case brought in South Carolina against a network of 24 nursing home providers (collectively, “Agape”).
The case, U.S. ex rel. Brianna Michaels and Amy Whitesides v. Agape Senior Community, et al., was brought in 2012 by two former employees of Agape as relators (“plaintiffs”) under the whistleblower provisions of the FCA. Plaintiffs alleged that Agape engaged in a scheme to submit claims to Medicare, Medicaid and Tricare for hospice and nursing home inpatient services that were false because the care was not medically necessary or the certifications required to obtain reimbursement were falsified. As with all FCA cases, the U.S. government investigated the allegations to determine whether it would intervene in the case on behalf of the private party plaintiffs. The Government ultimately elected not to do so, and the plaintiffs and defendants moved forward with discovery. As a result of a dispute between the parties regarding the scope of discovery, the matter came before U.S. District Court Judge Joseph F. Anderson, Jr. on the question of whether the plaintiffs would be permitted prove liability and damages based on a statistical sampling model.
In the context of health care fraud and abuse, the process of statistical sampling takes a sample of claims relating to a small group of patients which are reviewed to determine which of the sampled claims are allegedly false. The results of that analysis of the sample may then be extrapolated to the much larger universe of claims and patients to prove a plaintiff’s claim of liability and damages under the FCA. In the Agape case, Judge Anderson noted that the case presented well over 50,000 individual non-related claims, involving medical charts of between 10,000 and 20,000 nursing home patients. Nevertheless, despite the extensive amount of time that would be required for the parties to review and present of all of those claims at trial, Judge Anderson determined that statistical sampling would not be appropriate in this case, because (1) each claim asserted in the case presented the question of whether services furnished to nursing home patients were medically necessary; (2) answering the question for each of the patients would involve a highly fact-intensive inquiry involving expert testimony after a review of each patient’s medical chart; and (3) the medical charts of each patient for which the false claims were alleged were intact and were available for review by the parties. Thus, while the review could conceivably involve thousands of claims and patients, Judge Anderson was satisfied that those claims should be adjudicated individually, rather than using an extrapolation from a statistical sample that may not accurately reflect the non-sampled cases.
Following Judge Anderson’s initial ruling on the statistical sampling issue, the plaintiffs and Agape entered into settlement discussions without the government’s involvement. They ultimately reached an agreement whereby Agape would pay the plaintiffs $2.5 million in settlement of all claims. The settlement was then submitted to the government for approval, as all FCA settlements must be approved by the court and the U.S. Department of Justice (the “Government”). The Government rejected the settlement, based in large part on its own statistical sampling analysis that put the value of the case at $25 million.
The plaintiffs and Agape objected to the Government’s refusal to approve the settlement, but Judge Anderson ruled that the plain language of the FCA required the Government’s approval, despite the fact that it had not intervened in the case and despite the fact that he had already found that a statistical sampling methodology — which formed the basis for the Government’s objection — could not be used in the case to determine liability or damages.
However, rather than move forward with the trial, Judge Anderson certified both of his rulings for immediate appeal, reasoning first that if the Government’s objection is overturned by the Court of Appeals and, upon remand, Judge Anderson determined that the objection was unreasonable, the case would end with an amicable settlement. Conversely, Judge Anderson reasoned that if the trial proceeds without the use of statistical sampling in determining liability or damages, the parties would face a trial of monumental proportions, involving a staggering outlay of expenses by the parties, which would possibly be unnecessary if the Court of Appeals reversed his rejection of the plaintiffs’ proposed statistical sampling methodology. Thus, Judge Anderson concluded, it would be much more judicially efficient to have a ruling on both questions before starting the trial.
Plaintiffs, Agape and the Government have all recently filed briefs with the Court of Appeals. In addition, the American Hospital Association (AHA), the Catholic Health Association (CHA) and the American Health Care Association (AHCA) filed amicus curiae briefs in support of Agape, asking the Court of Appeals to affirm Judge Anderson’s ruling disapproving statistical sampling. The AHA/CHA joint brief argues that when the falsity of a claim depends on a doctor’s medical judgment about a patient’s condition, plaintiffs cannot prove liability through statistical sampling. Otherwise, plaintiffs in such cases would only have to provide evidence that there was no reasonable basis for a doctor’s medical judgment for treatment of the patients in the sample, and could ignore individual treatment issues present in the non-sampled patients. Therefore, liability must be proved on a claim-by-claim basis. The joint brief further argues that the end result would be that the larger the number of patients and claims covered by a plaintiff’s allegations, the lower his burden of proof would become. This “combination of lowering the burden of proof and truncating a defendant’s ability to defend itself would only further incentivize the filing of questionable and meritless qui tam suits.”
The ACHA brief also focuses on the need for a claim-by-claim analysis where the FCA claims are “based on physicians’ medical judgments concerning their patients’ conditions, prognoses, and medical needs.” ACHA goes on to argue that allowing plaintiffs in this type of case to prove liability for unspecified claims using statistical sampling would essentially shift the burden of proof to providers to have to disprove the elements of FCA liability for each such unspecified claim.
The parties are now awaiting a hearing date for oral arguments on the appeal.