Statistical Sampling Growing in Acceptance – and Importance – in False Claims Act Cases (Part 3 of 3)
Sampling Rejected: United States v. Vista Hospice Care, Inc.
Though Tyson Foods, Inc. v. Bouaphakea, 136 S. Ct. 1036, 194 L. Ed. 2d 124 (2016), definitively rejected the common defense argument (also made in Life Care) that statistical sampling is never appropriate in False Claims Act cases, courts are of course not bound to accept sampling in all FCA cases. A post-Tyson case that rejected sampling is United States ex rel. Wall v. Vista Hospice Care, Inc., No. 3:07-CV-00604-M, 2016 WL 3449833 (N.D. Tex. June 20, 2016). In Wall, the relator alleged that the defendant hospice providers violated the False Claims Act in several ways, including by admitting patients who were ineligible for hospice and billing the Government for their care. Id. at *1, 5. Eligibility for the Medicare hospice benefit depends on a patient being terminally ill, mean “the individual has a medical prognosis that his or her life expectancy is 6 months or less….” Id. at *2. “CMS recognizes that prognostication is ‘uncertain’ and not ‘an exact science,’” going so far as telling physicians that “they need not be concerned” with being incorrect and that “‘[t]here is no risk to a physician about certifying an individual for hospice care that he or she believes to be terminally ill.’” Id. at *3 (quoting Program Memorandum Intermediaries/Carriers, Subject: Provider Education Article, CMS-Pub. 60AB (Mar. 28, 2003)).
To prove her claims if improper hospice admissions, relator offered testimony from two experts, a statistician who developed a stratified sample of 291 patient files, and a geriatrician who evaluated each file to determine whether the patient was eligible for hospice care. Id. at *10–11. The relator then offered the statisticians extrapolation from the stratified sample to the entire population as “the only evidence regarding those patients and claims.” Id. at *11.
The Wall court rejected relator’s attempt to prove liability and damages with the testimony of this statistician, finding that “[i]n this context . . . the underlying determination of eligibility for hospice is inherently subjective, patient-specific, and dependent on the judgment of involved physicians. Even if extrapolation could support a False Claims Act claim for submitting MHB claims for patients with greater than six month prognoses, the facts do not justify such extrapolation here.” 1 Id.
The court distinguished other FCA cases that permitted sampling by finding that in those cases falsity did not “depend on individual physicians’ judgment regarding individual patients.” Id. at *12. The court elaborated on this distinction, noting that while other cases involved an inquiry into whether “Defendants provided services that were reasonable and necessary,” i.e. medical necessity, the hospice claims at issue in Wall depended on “whether, ‘based on the physician’s or medical director’s clinical judgment regarding the normal course of the individual’s illness’ a patient had a life expectancy of 6 months or less.” Id. (emphasis in original) (citing 42 U.S.C. § 1395(f)(a)(7). Cases like United States ex rel. Martin, 114 F. Supp. 3d 549, were different, the Wall court reasoned, because they “involved the clinical picture of individual patients, they did not require examination of the subjective clinical judgment of a number of certifying physicians applying the ‘uncertain,’ ‘change[able],’ and ‘[in]exact science’ involved in predicting an individual’s life expectancy.” Id. at *12 (emphasis in original) (quoting Program Memorandum Intermediaries/Carriers, Subject: Provider Education Article, CMS-Pub. 60AB (Mar. 28, 2003)). 2
Litigating False Claims Act Cases with Statistical Samples
As the Supreme Court unequivocally held in Tyson Foods, Inc. v. Bouaphakeo, 136 S. Ct. 1036, 1046, 194 L. Ed. 2d 124 (2016): “A representative or statistical sample, like all evidence, is a means to establish or defend against liability.” Cases like Wall and Michaels should serve as warnings that courts will not always accept the statistical sample that the proponent offers, even if it means litigating the relator’s claims would be impossibly time consuming and expensive. But most courts will appreciate the policy arguments in favor of permitting sampling that were articulated in United States ex rel. Martin v. Life Care Centers of Am., Inc., 114 F. Supp. 3d 549, 571 (E.D. Tenn. 2014) and accept a well-crafted statistical sample.
Cautious litigators will pay close attention to how the population and the sample are selected; seek sampling agreements with opposing parties where possible; obtain early guidance from the court; and, have a contingency plan should their proposed sample be rejected.
While we did not represent any of the parties in these cases, our firm routinely represents whistleblowers in complex healthcare False Claims Act actions. If you are aware of fraud involving healthcare or government contracts, please contact us.
- For another hospice case that rejected proof of liability and damages through statistical sampling, see U.S. ex rel. Michaels v. Agape Senior Cmty., Inc., No. CA 0:12-3466-JFA, 2015 WL 3903675 (D.S.C. June 25, 2015), order corrected, No. CA 0:12-3466-JFA, 2015 WL 4128919 (D.S.C. July 6, 2015), and aff’d in part, appeal dismissed in part sub nom. United States ex rel. Michaels v. Agape Senior Cmty., Inc., 848 F.3d 330 (4th Cir. 2017). The Michaels court certified its statistical sampling holding for interlocutory appeal, but the 4th Circuit dismissed that part of the appeal as improvidently granted. United States ex rel. Michaels v. Agape Senior Cmty., Inc., 848 F.3d 330, 340–41 (4th Cir. 2017). ↩
- The Wall court identified further problems with relator’s particular attempt to rely on sampling, including: her geriatrician’s testimony that “‘certainly you can’t extrapolate’ from how one physician assessed a patient’s eligibility [for hospice] to make conclusions about another physician”; her statisticians failure to select a random sample from the entire population; and, his failure to control for variables identified by the geriatrician. United States ex rel. Wall v. Vista Hospice Care, Inc., No. 3:07-CV-00604-M, 2016 WL 3449833, at *13–14 (N.D. Tex. June 20, 2016). ↩