Statistical sampling is a method of using a selection of data (a subset or sample) from a larger set in order to provide a reliable estimate of the data as a whole. Courts have allowed statistical sampling in different forms of litigation for decades. The method is now finding its ways into whistleblower litigation under the False Claims Act. In U.S. ex rel. Martin v. Life Care Centers of America, Inc., No. 1:08-cv-251, slip. op. (E.D.Tenn. Sept. 29, 2014), the United States District Court for the Eastern District of Tennessee considered a challenge to the government’s reliance on statistical sampling to prove its case under the False Claims Act.
The United States Government in Martin retained Dr. Constantin Yiannoutsos to serve as plaintiff’s expert and provide a statistically valid sample of data of the claims submitted by the defendant, Life Care Centers of America, to Medicare for payment. As plaintiff’s expert, Yiannoutsos took a sample of 400 admissions to the hospital out of a total of 54,396 admissions. Yiannoutsos testified that he used simulations to gain assurances that his estimates were precise. During a Daubert hearing, Yiannoutsos testified that the patient records associated with the 400 sampled admissions would be “reviewed by medical experts to ascertain whether claims within sampled admissions were submitted for non-covered services and to determine whether any overpayment resulted by submitting such claims to Medicare.” Id. at *7.
Under Daubert v. Merrell Dow Pharms., Inc., 509 U.S. 579 (1993), the United States Supreme Court established two roles for courts when determining the admissibility of expert testimony: (i) determine whether the expert opinions are reliable; and (ii) analyze whether the evidence is relevant. The court found that the government offered Yiannoutsos’ testimony to estimate the number of claims submitted by Life Care Centers for services not covered by Medicare and estimate the amount of loss suffered by the government due to those claims.
Regarding the reliability of statistical sampling, the Martin court noted that other courts have considered statistical sampling and extrapolation and found the methods sound. Citing a string of cases, the court noted that inferential statistics has been considered an “acceptable due process solution” in litigation which is well recognized as acceptable evidence in determining facts at trial. Id. at *15. After considering the various publications Dr. Yiannoutsos had published in academic journals, the margin of error applied to his methodologies and statistical sampling’s acceptance in the scientific community, the court found that Yiannoutsos’ sampling met the “reliability” prong under Daubert. Id. at 17.
Turning next to Daubert’s relevance prong, the court in Martin cited Federal Rule of Evidence 401 which provides that evidence is relevant if: (i) the evidence has a tendency to make a fact more or less probable; and (ii) that fact is of consequence in determining the underlying action. The court concluded that the government offered Dr. Yiannoutsos as “an expert proposing to testify as to scientific knowledge of statistics, which will ultimately assist the trier of fact to determine the facts at issue.” Yiannoutsos’ opinions regarding the number of claims submitted and the resulting damages stemming from those claims, the court reasoned, were certainly of consequence to the underlying action. Because the statistical sampling provided facts regarding the size and scope of the government’s case, the court found the expert testimony and opinions both reliable and relevant. Id. at *19.
Jason Cornell is an attorney who represents whistleblowers with the law firm Clark Fountain LaVista Prather Keen & Littky-Rubin. Clark Fountain represents plaintiffs in various matters throughout the United States. If you have questions regarding the issues addressed in this or other posts, you can reach Jason at email@example.com.