In a recent Ninth Circuit Court of Appeals case, prosecutors tried to stake their argument on their analysis of anomalies in big datasets. It didn’t work. The proceedings nevertheless highlight how useful data analytics can be in the compliance function.
Plaintiffs and government enforcement agencies increasingly rely on data analysis to support a wide array of allegations, including employment discrimination, insider trading, bid rigging, False Claims Act (FCA) violations and more. Although a recent Ninth Circuit decision in Integra Med Analytics LLC v. Providence Health & Services[1] reminds litigants that statistical anomalies alone are not sufficient to state a claim for fraud, enforcement authorities and civil litigants nonetheless increasingly employ data analytics to build their cases. Compliance-oriented companies should therefore harness their own sources of data to enhance their compliance programs, to identify events worthy of compliance investigations and to ultimately develop an understanding of significant statistical anomalies before prosecutors, qui tam relators or class-action lawyers come knocking.
Lessons from Providence Health
In Providence Health, a qui tam relator filed an FCA claim on behalf of the United States alleging that Providence Health & Services, its affiliated hospitals (Providence) and J.A. Thomas & Associates, Inc. (JATA) filed false claims for government reimbursement under the Medicare program. The relator, Integra Med Analytics, LLC, a data-analytics company, alleged that Providence submitted Medicare claims that were coded for more lucrative secondary diagnoses and that were not supported by patients’ conditions. This claim was based primarily on Integra’s “proprietary statistical analysis” of data received from the Centers for Medicare and Medicaid Services (CMS).[2] The relator claimed its analysis demonstrated that Providence submitted claims that included a higher rate of coding for these secondary diagnoses than other comparable institutions, that it controlled for inter-hospital variation caused by different characteristics in patient populations and that “there was less than a one-thousandth percent chance” that Providence’s greater rate of coding of these conditions was “due to chance.”[3]
The defendants moved to dismiss, and the district court granted their motions in part and denied them in part, allowing Integra’s primary FCA claim to proceed. At the defendants’ request, the district court certified its order for interlocutory appeal. On appeal, the Ninth Circuit concluded Integra had not adequately pleaded the falsity of Providence’s Medicare claims. The court explained that in order to survive a motion to dismiss, a claim must be plausible. And, in evaluating plausibility, courts must consider an “obvious alternative explanation” for the defendant’s behavior.[4] Here, the court found Integra’s allegations “do not eliminate an obvious alternative explanation — that Providence, with JATA’s assistance, was more effective at properly coding for better Medicare reimbursement than others in the health care industry.”[5]
The court accepted Integra’s allegations that:
- Providence used the secondary codes at a higher rate than other comparable institutions,
- this difference was not due to chance or variations in patient populations and
- Providence and JATA staff incentivized doctors to use language conducive to coding higher-paying secondary diagnoses through their documentation tips and queries.
But still it reasoned that it “need not — and cannot — accept the conclusion that these allegations resulted from fraud or that doctors recorded unsupported medical conditions.”[6] Therefore, the Ninth Circuit concluded Integra had not stated a plausible claim for relief and reversed the district court’s order denying the defendants’ motions to dismiss.
The Government’s Increasing Reliance on Data Analytics
The Providence Health decision offers some limitation on the ability of plaintiffs and the government to rely solely on data and statistics to sustain claims for relief, and more specifically, fraud or false claims. However, it does not eliminate the important role that data analytics now plays in shaping government enforcement and private litigation. Even the Providence Health court acknowledges that its decision “does not categorically preclude statistical data from being used to meet Rule 8(a)’s pleading requirement and, when paired with particular details of a false claim, Rule 9(b).”[7]
In 2019, then Deputy Assistant Attorney General Matthew S. Miner explained in a speech delivered to the Government Enforcement Institute that prosecutors were increasingly relying on data analytics to more efficiently identify targets for investigations. He specifically mentioned enforcement actions relating to Medicare claims and to securities and commodities trading. In addition, the DOJ’s Procurement Collusion Strike Force has reported that it is working to bolster the capabilities of data teams across the government to detect bid rigging and other crimes.
Deputy Assistant Attorney General Miner’s remarks explained that not only should companies expect that enforcement teams may rely on publicly available data to identify indicators of anomalies that warrant investigation, they should also expect that if misconduct does occur, “prosecutors are going to inquire about what the company has done to analyze or track its own data resources.”[8] This suggestion was subsequently incorporated into the DOJ’s June 2020 revised guidance for DOJ prosecutors evaluating whether a corporate compliance program was effective at the time of an offense and is effective at the time of a charging decision or resolution. This revised guidance now instructs that while evaluating whether a corporation’s compliance program is adequately resourced and empowered to function effectively, prosecutors should examine the following questions:
- “Do compliance and control personnel have sufficient direct or indirect access to relevant sources of data to allow for timely and effective monitoring and/or testing of policies, controls and transactions?” and
- “Do any impediments exist that limit access to relevant sources of data and, if so, what is the company doing to address the impediments?”
Looking Ahead
Compliance professionals should take some comfort from the Ninth Circuit’s reminder that a statistical anomaly alone is not sufficient to state a claim where there is an obvious alternative explanation. However, the Providence Health decision also serves as a reminder that private plaintiffs, as well as enforcement agencies, are increasingly relying on data analysis to identify potential claims or targets for investigations. Companies should, therefore, use the data they have at their disposal to improve their own compliance efforts. Such efforts might include:
- Employing data analytics to identify anomalies in sales data, customer complaints, payments to consultants or other third parties, reimbursements to employees and other data sources. Compliance professionals should identify the major risk areas for their company and industry and then work with other departments to identify potentially useful sources of data.
- Data analytics can also be used to create dynamic tools that can illustrate patterns over time (i.e., sales by country or reimbursements to third parties), which can help identify new compliance risks as they emerge. For example:
- Sales data can be monitored in real time to ensure unusually large increases in sales to particular customers are monitored.
- Travel data can also be examined to help the compliance department identify activities in new jurisdictions that might raise new compliance issues.
Data analytics should, of course, supplement, not replace, traditional compliance functions. Harnessing data can help target the work of compliance teams, thus enabling compliance professionals to spend more of their time on complex issues like determining the real causes of statistical anomalies that data analytics may surface. As the Providence Health decision underscores, examining and understanding the causes of such anomalies may prove important in future litigation.
[1] No. 19-56367, 2021 WL 1233378 (9th Cir. Mar. 31, 2021).
[2] Id. at *2.
[3] Id. See also id. at n.3 (noting that Integra’s complaint also states that Integra “only considered claim groupings where there was less than a one-in-a-thousand chance that the difference in major complication rate at Providence versus other hospitals was due to random causes”).
[4] Id. at *3 (citing Eclectic Props. E., LLC v. Marcus & Millichap Co., 751 F.3d 990, 996 (9th Cir. 2014), which was quoting Ashcroft v. Iqbal, 556 U.S. 662, 683 (2009)).
[5] Id. at *4.
[6] Id. at *3-4.
[7] Id. at *4 n.5.
[8] Miner, supra n.8.
