ORDER
STEVEN D. MERRYDAY, District Judge.
The qui tam relator, Angela Ruckh, sues (Doc. 75) under the Federal False Claims Act and the Florida False Claims Act. Ruckh alleges that the defendants defrauded the United States and Florida by "upcoding" and "upcharging" for patients at fifty-three medical facilities in Florida. Ruckh argues that individually analyzing each claim from each of the fifty-three facilities is impractical and moves (Doc. 172) in limine to admit expert testimony based on statistical sampling that has yet to occur. The defendants respond (Doc. 179) in opposition, and the United States submits (Doc. 190) a brief in support of the motion.
BACKGROUND
On June 10, 2011, Ruckh sued (Doc. 1) one defendant, La Vie Health Care Centers, Inc., in a twenty-two page complaint. The complaint alleged that Ruckh worked for La Vie — a company operating 122 nursing and rehabilitation centers — at two facilities from January 2011 until May 2011, when she "depart[ed]" from her job. (Doc. 1 ¶ 27) Ruckh alleged that, while working for La Vie, she witnessed false claims submissions and "flagrant upcoding" by which La Vie charged Medicare and Medicaid more for a patient than the circumstance justified.1 Thus, Ruckh sued La Vie under the Federal False Claims Act (Counts I-III) and under the Florida False Claims Act (Counts IV-VI).
On August 2, 2012, after the United States declined to intervene,2 Ruckh amended the complaint to identify herself as the relator and to change the defendant from La Vie to Genoa Healthcare Consulting, LLC, (a company that "does business as `LaVie Care Centers'") and Salus Rehabilitation, LLC, (a company that "does business as `LaVie Rehab'"). (Doc. 16 ¶ 8) Genoa operated the two facilities at which Ruckh worked, and Salus offered rehabilitation services at each. The complaint otherwise remained largely unchanged; the complaint comprised twenty-two pages and alleged the same six counts.
Soon after the amendment, Genoa and Salus each moved (Docs. 19, 20) to dismiss the complaint. The two defendants argued (1) that, under United States ex rel. Clausen v. Laboratory Corporation of America, 290 F.3d 1301 (11th Cir. 2002), Ruckh failed to allege each claim with particularity, (2) that Ruckh's allegations targeted neither defendant but instead Marshall and Governor's Creek (i.e., "related" companies that "share the same parent company" as the two defendants), and (3) that Ruckh failed to adequately plead vicarious liability.
A March 5, 2013 order (Doc. 48) grants the motions and dismisses the complaint. The order explains that, although Ruckh alleged a general scheme to defraud the government, under Clausen Ruckh failed to state a claim with the particularity required by Rule 9(b), Federal Rules of Civil Procedure. Most importantly, the complaint specifically described only one instance of fraud, and that description lacked "details about the fraudulent substance of the submission or about the time of the submission or about the government's overpaying the claim." (Doc. 48 at 6) Accordingly, the order stayed discovery (except on the one specified instance of fraud and on the defendants' corporate structure), dismissed the complaint, and directed the relator to file an amended complaint that included "`specific, discrete facts of the who, what, when, and where variety,' to support [any] allegations of fraud." (Doc. 48 at 8-9 (citation omitted))
On April 30, 2013, the relator amended (Doc. 63) the complaint, but an order (Doc. 71) resolving a "discovery tiff" sanctioned the defendants and permitted the relator to amend the complaint again. On June 3, 2013, the relator amended (Doc. 75) the complaint again, and the amended complaint remains the operative complaint. Although the complaint retains contains six counts, the complaint has grown from twenty-two pages to eighty pages and greatly expands each count. Rather than alleging that two defendants defrauded Medicaid and Medicare at Ruckh's two facilities, the complaint alleges that five defendants defrauded Medicaid, Medicare, and TRICARE3 at fifty-three facilities, most of which Ruckh has never visited.4 In response, the five defendants submitted four motions (Docs. 82-85) to dismiss, but a January 14, 2014 order (Doc. 92) denies each motion and lifts the discovery stay.
Citing the voluminous discovery in this action and arguing that producing and processing the relevant medical records at the "fifty-three . . . [medical] facilities and some fifty-three . . . off-site storage locations" within a reasonable time is impossible, Ruckh moves (Doc. 172) in limine to admit expert testimony on statistical sampling, but the sampling has yet to occur. The defendants respond (Doc. 179) in opposition, and the United States submits (Doc. 190) a brief in support of the motion.
The expert explains the statistical sampling that Ruckh moves to admit:
In this case, sampling and extrapolation procedures will produce an estimate of total overpayments to the defendant[s] by the Medicare and Medicaid programs. For example, if 1% of the population is sampled and reviewed, the total overpayment in the population is probably about 100 times the overpayment in the sample.
(Doc. 172-1 ¶ 16) The expert describes the "general steps of sampling . . . as follows":
a. Define the population of interest (e.g., period of interest, facilities, and residents, insofar as their stays were reimbursed by Medicare or Medicaid);
b. Identify a data source from which the sample will be drawn (e.g., list of residents, billing records). The goal is for this "sampling frame" to comprehensively reflect the population;
c. Define a sampling unit (e.g., resident, bill, resident-day);
d. Select a sample type (e.g., simple random sample, stratified sample, cluster sample) and draw the sample;
e. Review the sampled units and determine the outcome of interest for each sampled unit (e.g., overpayment); and
f. Extrapolate the results of the sample review to the population.
(Doc. 172-1 ¶ 19)
DISCUSSION
The defendants argue that the proposed expert testimony is inadmissible in this action. A Daubert hearing will best resolve those issues. However, the defendants also argue that statistical sampling is impermissible in a qui tam action and that, even though statistical sampling is sometimes admissible, Ruckh's motion is premature. This order resolves those arguments.
1. Statistical Sampling in a Qui Tam Action
Citing United States v. Friedman, 1993 U.S. Dist. LEXIS 21496 (D. Mass. July 23, 1993) (Mazzone, J.), the defendants argue that "statistical sampling and extrapolation cannot form the basis for liability in a [qui tam] case due to the lack of individual proof." (Doc. 179 at 12) In footnote 1, Friedman stated:
The government offered evidence at trial to demonstrate through expert testimony that the random sample, 350 out of the total of 676 submitted by the defendant contained a material misrepresentation. The government then urges that an extrapolation from the random sample of the over 50% of the claims submitted by the defendant to all of the claims submitted would result in an overpayment of $83,766.84. While I recognize the validity of the mathematical and statistical projections based on a review of a smaller number of claims I have declined to extrapolate in the manner urged by the government. My declination is based on the existence at trial of discrete claims which were analyzed and discussed and subjected to cross examination. I was able therefore to review each claim in reaching my conclusions. While I am mindful of the government's efforts to shorten the trial and present its evidence efficiently and clearly, I am reluctant to accept a statistical sampling as the basis for doubling the alleged overpayment without the same scrutiny and support.
However, Friedman is inapplicable in this action. United States ex rel. Martin v. Life Care Centers of America, Inc, 2014 U.S. Dist. LEXIS 142660, at 33 (E.D. Tenn. Sept. 29, 2014) (Mattice, J.) — the only opinion that cites Friedman — explains that "Friedman recognized the validity of statistical sampling even though it was not applied in that case, indicating that the case does not stand for the proposition that statistical sampling cannot be used in large-scale [qui tam] cases." Like Martin, this action differs from Friedman:
Friedman is distinct from the instant case because there was a sufficiently limited universe of claims for the court to review each one individually rather than relying on extrapolation. Considering the large universe of allegedly false claims in the instant case, it would be impracticable for the Court to review each claim individually, as the court did in Friedman. Indeed, if the Court were to individually review each allegedly false claim or statement in this action, it would consume an unacceptable portion of the Court's limited resources.
Martin, 2014 U.S. Dist. LEXIS 142660, at 45-46. Further, even if Friedman applied in this action, Friedman is an unpublished, district court opinion from another circuit that contains little analysis, cites no precedent, and is cited favorably by no one. Accord Martin, 2014 U.S. Dist. LEXIS 142660, at 33 ("[Friedman] is not binding on this Court and provides little analysis regarding the propriety of statistical sampling in a [qui tam] case."); see also United States v. Robinson, 2015 WL 1479396, at *10 (E.D. Ky. Mar. 31, 2015) (Van Tatenhove, J.) (admitting statistical sampling and collecting cases supporting the finding that "statistical sampling methods and extrapolation have been accepted in the Sixth Circuit and in other jurisdictions as reliable and acceptable evidence in determining facts related to [False Claims Act] claims as well as other adjudicative facts").
2. Prematurity of the Motion
The defendants "characterize[] [the relator's motion] as an effort to obtain this Court's `advisory opinion' or `comfort order' indicating that Relator is `on the right track' in preparing its expert analysis." (Doc. 179 at 6) The defendants cite Jahn v. Equine Services, PSC, 233 F.3d 382, 393 (6th Cir. 2000) (Cudahy, J.), which holds, "A district court should not make a Daubert ruling prematurely, but should only do so when the record is complete enough to measure the proffered testimony against the proper standards of reliability and relevance."
The defendants explain that until the expert completes the statistical sampling, the margin of error is unavailable. The expert describes the margin of error:
That [statistical sampling] estimate will have a certain precision, often expressed in terms of a margin of error and an associated confidence level. The smaller the margin of error, the more precise an estimate is. Statistical science provides the tools to quantify the margin of error. It depends, among others, on sample size (the larger the sample, the smaller the margin of error) and on the variability of the outcome of interest (here: overpayments). For intuition: if the overpayment varies widely across sampled units (high variability), a single sample member can substantially affect the average and thus the estimated total overpayment. In contrast, if the overpayment in each sampled unit has little variability, then the overpayment in the rest of the population will likely be similar and the margin of error much smaller.
(Doc. 172-1 ¶ 16) Unlike the sample size, the variability of the overpayments is not known until after the sampling is complete. Thus, as the expert admits, "it is not yet possible to precisely calculate the margin of error that will result from a certain sample size." (Doc. 172-1 ¶ 29) Because the margin of error is unknown and because the defendants have not moved in limine to exclude any expert testimony, an Daubert hearing is premature.
CONCLUSION
To the extent that Ruckh requests a Daubert hearing before the expert completes the statistical sampling and without an objection based on Daubert, Ruckh's motion (Doc. 172) is DENIED. On the other hand, no universal ban on expert testimony based on statistical sampling applies in a qui tam action (or elsewhere), and no expert testimony is excludable in this action for that sole reason (although defects in method, among other evidentiary defects, might result in exclusion).
