In 1863, the False Claims Act (“FCA” or “the Act”) was originally enacted as a result of Congressional actions to stop army contractors involved in the Civil War from defrauding the U.S. government. Under the Act, as amended, a person who knowingly submits or causes others to submit false claims to the U.S. government is liable for such conduct equivalent to treble damages and a per claim penalty that ranges between $10,781 to $21,562.1 The FCA also permits private individuals to make claims of statutory violations as a Relator in “Qui Tam” action on behalf of the U.S. Government.2 Civil litigation against health care companies for alleged FCA violations has become one of the federal government’s most effective legal tools in recovering damages and penalties from individual and corporate health care defendants. For the fourth consecutive year, the Department of Justice (DOJ) reported that in fiscal year ended September 30, 2015, it collected more than $3.5 billion in settlements and judgments from civil litigation involving fraud and false claims, with total cumulative recoveries exceeding $26.4 billion. In 2015, 54% or $1.9 billion of DOJ’s collections came from health care industry defendants (individuals and companies) arising from FCA claims relating to “unnecessary or inadequate care, paying kickbacks to health care providers to induce the use of certain goods and services, or overcharging for goods and services paid for by Medicare, Medicaid, and other federal health care programs.” The DOJ also noted that additional recoveries were made for state Medicaid programs and individuals.3 The economic impact to a health care defendant in FCA cases can be consequential given the U.S. government’s continued trend in bringing or supporting FCA claims, the potential for a defendant’s liability, the magnitude of a defendant’s potential economic damages and penalties, and the costs of defense.

In reviewing recent FCA claims brought in various Federal District Courts, FCA defendants are actively challenging the application of statistical sampling techniques employed by Qui Tam plaintiffs and the federal government to prove liability and to calculate damages. This article discusses key trends involving defense challenges and considerations involving statistical sampling in an FCA claim against health care defendants.

Historically, the federal government proffered damage calculations in FCA claims based on statistical sampling and extrapolation.4 As explained in United States v. Fadul, a 2013 health care fraud case that alleged fraudulent billing practices by a licensed cardiologist, the District Court of Maryland found that “Courts have routinely endorsed sampling and extrapolation as a viable method of proving damages in cases involving Medicare and Medicaid overpayments where a claim-by-claim review is not practical.”5 In a recent FCA claim against a corporation that owns skilled nursing facilities, USA ex rel. Martin v. Life Care Centers of America, Inc., the District Court found that FCA did not specifically preclude the use of statistical sampling and permitted the federal government to prove FCA liability by using statistical sampling.6 The District Court of Eastern Tennessee in Life Care case determined that even though the federal government could provide individualized proof of specific claims made or false statements, such individualized proof would require a level of effort by the federal government that was regarded by the District Court as impractical. The District Court also disagreed with Life Care’s claim that unique patient factors and medical determinations should preclude the use of statistical sampling. Instead, the District Court of Eastern Tennessee opined that Life Care could challenge the weight that a fact-finder would place on conclusions drawn from a statistical sample through cross-examination and alternate witnesses to demonstrate the disparity between the parties’ sampling and testing methods and conclusions.7 In other FCA litigation, the District Courts have applied policy considerations and found that limiting statistical sampling would reduce FCA enforcement because having to perform claim-by-claim reviews would deter the number of prosecuted claims.8 In a 2015 FCA case, U.S. ex rel. Rukh v. Genoa Healthcare, LLC the District Court of the Middle District of Florida applied similar justifications to those in used Life Care, to allow statistical sampling in FCA litigation where the federal government alleged that fifty-three medical facilities overbilled patient charges. The District Court determined that “(c)onsidering a large universe of allegedly false claims in the instant case, it would be impracticable for the Court to review each claim individually … it would consume an unacceptable portion of the Court’s limited resources.”9 Thus, recent FCA litigation has shown that District Courts have permitted the use of statistical samples to prove liability based on Congressional intent and statutory interpretation of the Act, FCA enforcement policy considerations, and concerns relating to efficiency in evaluating the government’s claims where there are large volumes of evidence to consider.

Although District Courts have expanded the use of statistical sampling to prove liability in FCA claims involving health care defendants, these federal court decisions do not suggest that statistical sampling has been given blanket approval. In United States v. Friedman, the District Court of Massachusetts allowed the introduction of statistical sampling as evidence in claims against a defendant that allegedly overbilled Medicare and found that the defendant had violated the FCA. However, the District Court in Friedman denied the use of a sample to extrapolate and calculate damages. Since the District Court was faced with only 676 claims, it preferred the review of each individual claim in order to reach its determination of damages.10 Thus, the size of the total population relative to the claims or damages may be a factor for consideration by the District Court in determining whether sampling is appropriate.

The reasonableness of statistical sampling methods may also be considered by the District Court in its determinations of whether samples should be admitted as evidence to prove FCA liability or damages. In United States ex rel. Trim v. J.D. McKean, the District Court of the Western District of Delaware found that certain proffered audits performed by Medicare, Medicaid, and other benefit programs were invalid statistical samples since the Court found that the proffered audits did not accurately represent all relevant claims at issue. Some of the McKean audit deficiencies identified by the District Court included, among others, the inclusion of atypical claims in the audit, the failure to establish the auditor’s reliability or the reliability of the audit methods applied, the relatively small audit sample sizes, the varying scope of years in each audit, and finally, the judgmental nature of coding determinations. The Court also found that some of the McKean audit evidence was illegible and in a foreign language. Even though the District Court found the McKean audits were not reliable as statistical samples, it still held that the audits provided evidence that supported the conclusion that McKean had violated the FCA.11 In a 2015 federal court opinion involving FCA claims, United States of America ex rel. Brianna Michaels and Amy Whitesides v. Agape Senior Community, Inc. et al., the District Court of South Carolina denied the federal government’s use of statistical sampling to prove that certain of Agape’s nursing homes had violated the FCA. The District Court determined that Agape’s medical charts remained available for review and were not under threat of being destroyed. In addition, the District Court found the government’s claims to be fact-intensive, including medical testimony to determine whether nursing home patient services provided had been medically necessary. The Agape defendants also asserted that there would be no cost savings to the plaintiffs in using a sample to determine liability since each sampled item would be subjected to a lengthy cross-examination.12 In a 2016 case, United States of America ex rel. Misty Wall v. Vista Hospice Care, Inc. et al., the federal government alleged that a hospice care facility had violated the FCA based on a sample of 291 patient files from a total population of 12,000 patients and calculated economic damages.13 The District Court for the Northern District of Texas found that the statistical sampling performed to establish Vista’s FCA liability and damages was “inherently subjective, patient-specific, and dependent on the judgment of involved physicians… extrapolation is not always appropriate.”14 The District Court relied on a Supreme Court decision, Tyson Foods, Inc. v. Bouaphakeo, 136 S.Ct. 1036, 1046 (2016), where the Supreme Court held that “(t)he permissibility of statistical sampling turns on ‘the degree to which the evidence is reliable in proving or disproving the elements of the relevant cause of action.’” The Court also distinguished the Vista case from Life Care by finding that while Life Care involved the clinical condition of individual patients, Vista involved the “subjective clinical judgment of a number of certifying physicians applying the ‘uncertain… science’ in predicting an individual’s life expectancy.”15 The District Court also stated that “no circuit has resolved whether statistical sampling and extrapolation can be used to establish liability in an FCA case where falsity depends on individual physicians’ judgments regarding individual patients.”16 In Agape, although there was a sizeable sample, the Court noted that there would be no efficiencies gained in cross-examination by sampling and was one of the factors the District Court considered in finding that the entire population of medical charts must be considered by the government in proving its claims. By highlighting the similarities between the Agape and Vista claims, the District Court for the Northern District of Texas noted that in Agape, “‘each and every claim at issue’ was ‘fact-dependent and wholly unrelated to each and every other claim,’ and determining eligibility for ‘each of the patients involved a highly fact-intensive inquiry involving medical testimony after a thorough review of a detailed medical chart of each individual patient,’…the case was not ‘suited for statistical sampling.’”17 The District Court in Vista also found that conclusions made by one physician on a patient’s condition could not be extrapolated to draw conclusions about the conduct of another physician.18 In summary, District Courts have declined the use of statistical sampling in smaller cases where individual claims can be evaluated, where sampling methods are not determined by the Court to be appropriate or that produce unreliable results, or where the Court finds that there are individual fact intensive claims that involve subjective medical judgment on an individual’s condition which cannot be extrapolated to other patients or physicians.

Federal courts have held defendants responsible for challenging the methods and approaches used by the federal government and its experts in selecting statistical samples, testing the sample population, and providing conclusions on the testing. In Life Care, the District Court concluded that “statistical sampling is permitted to prove FCA claims brought by the federal government; however, the Court cannot control the weight that a fact finder may accord to the extrapolated evidence.”19 As defined, sampling involves the selection and testing of less than one hundred percent of items in order to draw conclusions about the characteristics or amounts of a particular population.20 Extrapolation has been viewed as “a statistical method in which a sample of data is used to draw inferences about a larger population.”21 In Life Care, the District Court discussed various aspects of statistical sampling in a federal case involving various FCA claims of overbilling, false claims and false statements concerning skilled nursing facility payments where statistical sampling methods were challenged. The Court explained that,

“the general purpose of statistical sampling is to ‘provide a means of determining the likelihood that a large sample shares characteristics of a smaller sample…In order to ‘draw reliable conclusions’ about the sample universe, the statistical sample must be of a sufficient size to support the conclusions…statisticians account for any discrepancies by calculating a margin of error.”22

The District Court relied on the Reference Manual on Scientific Evidence to set forth the elements of a reliable sample, including “when a sample method ‘defines an appropriate population, uses a probability method for selecting the sample, has a high response rate, and gathers accurate information on the sample units.’”23 As noted by the District Court, reliable conclusions drawn in statistical sampling are a direct result from selecting an appropriately sized sample from a defined population. All the factors regarding the method of sampling should be documented by the federal government’s expert in a sampling plan that is scrutinized by defendant’s counsel and experts.

Beyond the sampling plan, the defense must perform a thorough and careful evaluation of the statistical sample’s testing approach and conclusions. In Ruckh, the District Court noted that statistical sampling evidence could be excludable if there were “defects in method, among other evidentiary defects.”24 Through analysis and testimony proffered by the defense, a fact finder may conclude that a statistical sample was improperly selected or improperly tested, a statistical sample was not truly representative of the population, or the conclusions related to the extrapolated results were inappropriate through an alternative analysis of the findings or application of relevant medical standards. In their Health Law & Policy Blog, James Segroves and Kelly Carroll effectively summarize key strategies for defense counsel to consider in challenging the government’s use of statistical sampling:

  • Challenge the Need for Statistical Sampling: Consider whether other reasonable options exist for analyzing the claims at issue that would eliminate the need for statistical sampling.
  • Challenge the Validity of the Sampling Technique: Highlight defects in the sampling methodology, including small sample sizes, unrepresentative samples, sample selection biases and randomness of the sample.
  • Challenge the Extrapolation Method and Conclusions: Scrutinize the estimation method employed and extrapolation conclusions reached, paying close attention to the confidence (degree of certainty) and precision (range of accuracy) levels.
  • Challenge the Admission of Statistical Sampling Evidence: ….In Daubert proceedings, a court determines the admissibility of expert testimony or scientific evidence under Federal Rule of Evidence 702 by analyzing whether the evidence is both relevant and reliable.
  • Challenge the Findings: Closely review the factual findings and examination processes used regarding the sample claims, conducting an independent examination of the sample claims as appropriate. This is a critical step, as allowing incorrect or questionable determinations about sample claims to go unchallenged has significant ramifications when multiplied exponentially as a result of extrapolation. Providers may also demonstrate uncertainty by challenging the credentials or the findings of the reviewers or by providing evidence of the subjectivity of the medical decisions underlying the submitted payment claims.”25

The Memorandum Opinion and Order by the District Court in the Northern District of Texas in Vista provide valuable insight into the benefits of thorough diligence in evaluating the plaintiff’s sampling plan, sample selection, and methodology that can result in successful exclusion of an expert on a pre-trial motion in District Court rather than relying on fact-finder opinions in trial.

The District Court found that the federal government’s expert used against the defendants in Vista had acknowledged errors in the sample selection, including having selected duplicate items, permitted random exclusions of patients from the total population, performed misclassifications of patient groups, failed to differentiate across geographies, physicians, and disease type, and failed to appropriately stratify the population. Although the government’s expert claimed the errors were corrected, the Ph.D. did not provide corroborating evidence of the corrections, and precluded evaluation of the corrections by opposing counsel. The expert’s sampling errors and failure to account for relevant sampling variables caused the District Court to lose confidence in the expert’s extrapolation opinions, and ultimately, the Court found that the government expert’s conclusions were unreliable, which precluded the extrapolation of his results to the total patient population.26 Through health care defense counsel insights and examining a recent District Court decision on the exclusion of a sample’s conclusions from FCA health care litigation, it is evident that both defense and plaintiff experts should expect a high level of scrutiny on their sampling plans, testing methods, and conclusions. Documenting each phase of the statistical sampling process, and its related errors or limitations, are critical components in being able to successfully persuade the District Courts on pre-trial motions or a fact finder that an opposing party’s sample is inappropriate, inaccurate, or irrelevant as extrapolated against a total population of medical cases or claims to determine FCA liability and/or economic damages.

Pre-trial and trial litigation related to statistical sampling will continue to remain a key aspect of FCA litigation involving health care defendants. Litigation of sampling issues is necessarily case and fact specific, driven by such elements, among others, as the scope and nature of the federal government claims, the government expert methodology, analysis, and conclusions, population size and characteristics, and the strength of the analysis of defense counsel on the government’s sampling evidence and conclusions, and its own proffer of sampling evidence, if it chooses to do so. One thing is certain; without diligence and detail, it is difficult to formulate effective strategies and opinions that are persuasive to the federal Court or other fact finders about whether statistical sampling or its results are accurate and representative of the total population, or should be excluded from consideration in determining FCA liabilities or damages.

Endnotes

  1. Federal Register, Vol. 81, No. 126 (June 30, 2016);
  2. https://www.justice.gov/sites/default/files/civil/legacy/2011/04/22/C-FRAUDS_FCA_Primer.pdf. See also United States of America ex rel. Glenda Martin and State of Tennessee ex rel. Glenda Martin v. Life Care Centers of America, Inc., 114 F. Supp. 3d 549, 557. See also 31 U.S.C. Section 3730(b).
  3. https://www.justice.gov/opa/pr/justice-department-recoversover-35-billion-false-claims-act-cases-fiscal-year-2015.
  4. See Chaves County Home Health Servs. v. Sullivan, 931 F.2d 914 (D.C. Cir. 1991). See also United States v. Jones, 641 F.3d 706, 712 (6th Cir.2011) and United States v. Rogan, 517 F.3d 449,453 (7th Cir.2008).
  5. See United States v. Fadul, Civil Action No. DKC 11-0385 (D. Md. Feb. 28, 2013), pages 1 and 15.
  6. Life Care at 571-572.
  7. Life Care at 571.
  8. See United States v. Friedman, No. 86-610-MA, 1993 U.S. Dist. LEXIS 21496 (D. Mass. July 23, 1993). See also Life Care at 571.
  9. U.S. ex rel. Ruckh v. Genoa Healthcare, LLC (“Ruckh”), No. 8:11-cv- 1303-T-23TBM (M.D. Fla. Apr. 28, 2015), page 3.
  10. United States v. Friedman, No. 86-610-MA, 1993 U.S. Dist. LEXIS 21496 (D. Mass. July 23, 1993).
  11. United States ex rel. Trim v. J.D. McKean, 31 F. Supp. 2d 1308, 1314.
  12. See United States of America ex rel. Brianna Michaels and Amy Whitesides v. Agape Senior Community , Inc. et al., C/A No. 0:12-3466-JFA, Order Resolving Two Interrelated Issues and Certification for Interlocutory Appeal Pursuant to 28 U.S.C. 1292(b), pages 17 and 18.
  13. See United States of America ex rel. Misty Wall v. Vista Hospice Care, Inc. et al. (“Vista”), No. 3:07-cv-00604-M, Section V, pages 21-23.
  14. Vista, page 21.
  15. Vista, pages 9, 23-24.
  16. Vista, page 22.
  17. Vista, page 22-24.
  18. Vista, page 25.
  19. Life Care at 572. See also United States of America ex rel., et al. v Aseracare Inc., et al., Civil Action No. 2:12-CV-245-KOB, pages 17 and 18.
  20. See AU Section 350, Audit Sampling, paragraph .01. See also Segroves, James F. and Kelly A. Carroll, “Numbers Never Lie… or Do They? The Use of Statistical Sampling in False Claim Act Claims, HLB Health Law & Policy Blog, where they reference United States v. Cabrera-Diaz, 106 F. Supp. 2d 234, 240 (D.P.R. 2000).
  21. See Rhoad, Robert T., Crawford, Jason M., and Mary Kate Healy, “Feature Comment: Extrapolation in FCA Litigation: A Statistical Anomaly or a Tactic Here to Stay?” The Government Contractor, Thomson Reuters, Vol 58, No. 2, January 13, 2016, page 1.
  22. Life Care at 559.
  23. Id.
  24. Ruckh, page 4.
  25. See Segrovies, James F., pages 2 and 3.
  26. Vista, pages 25- 27.

Marta Alfonso, CPA\CFF, JD, CIRA is a principal at MBAF, LLP, a top 40 CPA Firm. Marta represents providers as a forensic accounting, litigation and valuation expert in civil and criminal litigation, transactional matters, and in internal fraud and abuse investigations.

Reprinted with permission from: Health Law Journal, Winter 2016,  Vol. 21, No. 3, published by the New York State Bar Association, One Elk Street, Albany, New York 12207.