Real-World Evidence in the Real World: Beyond the FDA

Abstract

The 21^st Century Cures Act (“Cures Act”)¹ relies on the concept of real-world evidence (“RWE”)² to improve the Food and Drug Administration (“FDA”) approval process. This has amplified interest and furthered momentum in applying RWE more broadly, beyond FDA regulation. In this article, we discuss the understandable appeal of RWE's pragmatic application and its many potential benefits. But we also caution that claims about RWE's wide-ranging, ameliorative impact on the health care system are likely overstated.

The real world of RWE is messy and uncertain. Successfully incorporating RWE into regular health care system decision-making, beyond the FDA, faces considerable obstacles and limitations. We review the reasons to be wary about RWE as a game-changer. These concerns including data reliability, insufficient incentives for stakeholders to generate and engage with high-quality RWE, and lack of comprehensive regulatory oversight. In addition, the push for RWE may impact the enforcement of the health care fraud and abuse laws, perhaps not in necessarily positive ways. Increased reliance on RWE may have significant implications for off-label fraud enforcement, further conflating the distinction between claims that are false for reimbursement rather than for scientific purposes.

I. THE CURES ACT AND INTEREST IN RWE MORE GENERALLY

The Cures Act opens the door to potentially dramatic changes in the evidentiary standards for FDA product approval. Congress directed the Secretary of Health and Human Services (“HHS”) to consider permitting the use of RWE for FDA approval of new indications of previously approved drugs and for satisfying FDA post-approval study requirements.³ RWE, as defined by the Cures Act, is data about the usage, benefits, and risks of a drug derived from sources other than randomized clinical trials,⁴ such as observational data from electronic health records, disease registries, and billing records.⁵ While FDA was relatively silent before the Cures Act about use of RWE in the drug approval sphere, it recently issued guidance for use of RWE in regulatory decisions about medical devices.⁶ Also, FDA was already using RWE in more limited applications, such as supporting safety surveillance of medical products already in clinical use as part of its Sentinel Initiative.⁷

The traditional, gold standard method for evaluating new medical products and treatments is the randomized controlled trial (“RCT”).⁸ However, conventional RCTs have many well-recognized shortcomings for guiding health care decision-making. They are expensive to conduct and can take a long time to complete.⁹ They often analyze efficacy of an intervention by comparing it to a placebo, rather than answering the pragmatically more important question of relative effectiveness. Also, RCTs often use rigid eligibility criteria, such as excluding subjects with secondary illnesses, who take other medications, or who are over a certain age, in order to minimize confounders and isolate the efficacy of the studied intervention.¹⁰ Further, patients from certain racial and ethnic groups, elderly patients, and patients with limited literacy are usually underrepresented in RCTs.¹¹ As a result, the study populations are not representative of many typical patients. In addition, subjects enrolled in RCTs may benefit from a “trial effect” from increased monitoring and contact with health care providers as part of the research process, raising further questions about generalizability of the results.¹²

RWE's considerable appeal is its ability to draw upon diverse data sources that incorporate the experiences of broader patient cohorts and consider treatment in real clinical settings. And through combing of already existing records and observational studies, researchers can gather RWE at lower costs, and with greater ease, than conducting RCTs.¹³ RWE also can be collected in situations where RCTs are infeasible, such as rare diseases where developing large enough sample sizes for adequate randomized study proves near impossible.¹⁴

Health care stakeholders, therefore, can all benefit from careful curating and study of RWE. As RWE expands in use beyond FDA post-approval surveillance, its most common application,¹⁵ it may answer important questions about care value in numerous other needed contexts, including reimbursement decisions, individual treatment choices in the clinic, health care benefit design, quality improvement activities, pricing of products, and product labelling.¹⁶

II. ADJUSTED EXPECTATIONS ABOUT RWE

RWE proponents enthusiastically endorse its wider utilization within the health care system. But excitement about RWE has seemingly crossed the line between pushing hope and pushing hype, with claims that it will “revolutionize informed clinical practice and health service delivery.”¹⁷ Such lofty expectations about RWE need tempering. “RWE is a relatively new kid on the block, and it has its own real-world challenges.”¹⁸ There are numerous reasons to question RWE's role as a game-changer.

A. Really More “Real?”

As a threshold matter, the term “real-world evidence” lends more credence to RWE investigations than may be deserved. Just as “reality television” offers a highly skewed picture of the real world, some RWE can be disappointingly distorted in capturing what occurs in medical practice. Because RWE draws upon data sources from regular clinical practice and health care administration and financing, the presumption is that the evidence will inherently prove more authentic and representative of actual patients' experiences. But some data supporting RWE, such as billing records, are known to be woefully inaccurate when it comes to shedding light on the actual clinical care provided.¹⁹ Even RWE drawn primarily from clinical treatment and medical records data often lacks documentation of important patient outcomes because it is not initially collected with a view to generating secondary evidence for later retrospective or observational analysis.²⁰

Moreover, the sharp contrast between traditional, “artificial” RCTs and modern, “real” RWE may be overstated. Design of RCTs has adapted over the years. A subset of RCTs, known as “pragmatic RCTs,” now feature broader eligibility criteria and larger sample sizes, intentionally draw patients from representative care settings, more readily allow patients to receive standardized care outside of the randomized comparison, and use flexible follow-up and adherence protocols to track the real clinical experiences of patients more closely.²¹

B. Data Sources and Data Collection

The data used to support RWE derives from quite a broad range of sources, ranging from regular clinical records and billing records, to patient reported outcomes, and even mobile health platforms.²² Data of this type may offer a more comprehensive view of patient experiences. However, the very breadth of potential data inputs raises inherent problems. Some of the data will not work well in important contexts and for particular illnesses. For example, slow-severity diseases (e.g., arthritis) and low-prevalence diseases not treated in specialty centers may routinely generate less data, while the data actually collected is likely to be less comprehensive.²³

One would think that the regulatory push to encourage more health care providers to use electronic health records²⁴ would increase the availability of beneficial RWE data sources. But more records to access is helpful only if the electronic health records themselves have sufficient value. At present, the ability of many electronic health records to support high-quality RWE is questionable. Many providers continue to generate electronic health records primarily with billing criteria in mind; the clinically useful information that might inform RWE often remains in less accessible, unstructured clinician notes.²⁵

A supposed advantage of RWE is the ability to build up larger data sets through combination and combing of electronic medical records, disease registries, and other data repositories. However, much of this data, because it was not initially collected with a precise research question in mind, and often under nonrandomized methods, may still not be robust enough to address questions of causal inference, especially when actual treatment effects are will likely be modest.²⁶ Numerous confounders occurring in regular clinical practice, such as sicker patients gravitating toward one type of treatment, may drive seemingly causal associations.²⁷ Because of such concerns, “the very large size of real-world data sets may not matter because there is little advantage to increased precision if the answer has a good chance of being precisely wrong.”²⁸

Further complications arise concerning access to the data underlying RWE. Although RWE is typically generated in the course of regular treatment and health care administration, this does not mean that researchers can feasibly obtain such data. A recent survey of life-science companies indicates the scope of data access challenges. Over half of the companies reported investing in RWE programs to demonstrate the value of their products.²⁹ Despite this interest, they also reported limited progress, identifying access to high-quality data as the biggest impediment.³⁰ Many factors complicate data access, including privacy rules that may require deidentification of certain records or limit access altogether.³¹ Also, data collected from non-traditional sources, such as mobile health platforms and disease registries, will often require cleaning before it is amenable to research purposes.³² Unfortunately, there are no commonly accepted cleaning methods to achieve statistical validity.³³ Other challenges include difficulties linking data sources for the same health care episode, such as billing claims data, medical records data, and patient reported outcome data. Connecting the dots remains difficult because in our still heavily fragmented health care system, stakeholders use different information systems that are not readily interoperable.³⁴

Another significant challenge concerns representativeness of the study population. Although data in RWE studies comes from the “real” world, the cohort studied may not fairly approximate the wider patient population. For example, claims data and medical records data for a subset of colon cancer patients, even if drawn from different parts of the country, may not account for critical differences between this cohort and the larger group of colon cancer patients in terms of patient demographics, disease course, health insurance status, and other important factors.³⁵

The general problems for RWE investigations in accessing quality data exacerbate the risk that researchers will make use of whatever data sources are more easily accessed, even if non-representative. The wide variety of data sources to choose from also allows for strategic gamesmanship in source selection. Accordingly, a central concern underlying RWE “is whether the most appropriate data sources (rather than simply the most readily available, or the cheapest or the ones that gave the best results …) have been chosen ….”³⁶

Indeed, there is considerable risk that the push for more RWE will generate a lot of noise but questionable value. “[T]he confluence of large data sets of uncertain quality and provenance, the facile analytical tools that can be used by non-experts, and a shortage of researchers with adequate methodologic savvy could result in poorly conceived study and analytic designs that generate incorrect or unreliable conclusions.”³⁷ Moreover, the relative ease with which RWE studies can be developed compared to RCTs runs the risk of stakeholders generating multiple “do it yourself” RWE investigations to justify certain positions.³⁸ This can lead to a “Tower of Babel” of RWE, with important audiences likely tuning out much of the evidence. “Given the plethora of data sources and analytical approaches, differences in RWE study results are inevitable …. [W]ith insufficient technical expertise (or time or inclination) to conduct a critical comparison of the methodological aspects of each study, the average decision maker is likely to ignore both ….”³⁹

C. Risks of Bias

Because RWE draws upon data from real clinical practice, significant biases can undermine the evidence. Perhaps most obvious is selection bias. Unlike RCTs, RWE usually is generated from records and observations of non-randomized groups of patients.⁴⁰ Thus, it remains difficult to know whether effectiveness outcomes resulted from differences in the types of intervention or differences in the patients chosen. For example, physicians may tend to prescribe a new, expensive medication for only very sick patients, and insurers likewise may restrict coverage to very ill individuals. A RWE study that compared outcomes of patients receiving the new medication with patients using an older drug will have confounding bias because the new medication's utilization in the “real world” already skews toward sicker patients less likely to recover.⁴¹

RWE can introduce other biases beyond the initial selection stage. Relying on observational studies, as much RWE does, raises the possibility of performance bias, as patients observed between distinct groupings may adhere to treatment plans differently, including varying degrees of success in taking medications at the right time and the right dosage.⁴² Attrition bias is also a risk, as patients may end up withdrawing from different treatments in skewed patterns⁴³ if, for example, patients in one group encounter more reimbursement difficulties for certain treatments or worry about particular side effects associated with a certain intervention.

Additionally, a long-standing concern for clinical trials has been publication bias. Publication bias refers to the greater likelihood in industry-funded trials that positive outcomes will be published and shared with the medical community, and the lower likelihood that negative trial results will be publicly reported. Unfortunately, the relative ease and lower cost with which RWE studies can be generated—compared to that of RCTs—likely increases the already existing risk of publication bias. Proponents offering a RWE study as evidence must “dispel the suspicion that they ran multiple similar studies and analyses, but published only the one that gave a positive result.”⁴⁴

The FDA has taken helpful first steps for addressing publication bias in its guidance for use of RWE for regulatory decisions about medical devices. The guidance emphasizes that FDA expects that RWE investigations relied upon will have resulted from studies that followed clearly available, prospectively-defined protocols.⁴⁵ This in part limits the ability of investigators to switch out research plans and data sources midstream to pursue and publish only favorable results. Unfortunately, previous efforts to use transparency to counter publication bias in regular clinical trials, such as the requirement that certain studies be prospectively registered with ClinicalTrials.gov,⁴⁶ have had spotty enforcement and limited impact.⁴⁷

D. Standardization/Interoperability

The wide variety of potential data sources for RWE raises multiple coordination problems. For high-quality RWE data to be generated, the many providers, insurers, disease registries, and other parties generating the data in routine clinical care and health administration, must follow minimally consistent procedures and use standardized formats. Yet, despite numerous attempts at increasing data standardization in routine health care, data generation practices are not on the same page. Indeed, “[t]he format, quality, and validity of [real-world data] can vary significantly by practice, insurer, [electronic health record] vendor, and provider, and nascent efforts to address these challenges have yet to align on a path forward.”⁴⁸

Even outcomes measures recorded for patients with the same disease may vary between different clinical sites and different electronic records systems used, contributing to confusing noise that complicates assessment of the effectiveness of an intervention. For example, in oncology records, critically important endpoints that may be of interest to researchers, such as disease progression and tumor shrinkage, are not recorded in consistent, standardized ways in readily accessible data sources.⁴⁹ Moreover, many health care information systems are not sufficiently interoperable, making it difficult to transmit and link data elements, however standardized in format, between different health systems, distinct treatment sites, multiple disease system registries, different insurers' claims files, and other data sources. To increase assurance, the new push for RWE needs to strive toward adoption of standardized methods for accruing and linking real-world data, as well as increased interoperability between the data sources themselves.⁵⁰

E. Oversight/Uptake Problems

The new push for RWE also requires a comprehensive oversight apparatus capable of creating the right incentives and delineating uniform standards for generating high-quality evidence across the health care system. Unfortunately, the Cures Act does not create this type of regulatory platform. While FDA has developed guidance for using RWE in FDA oversight of devices, the standards criteria are still somewhat vague.⁵¹ Moreover, the guidance only applies to how FDA will consider RWE for its regulatory decision-making,⁵² leaving many other potential uses for RWE, from supporting reimbursement criteria to clinical practice guidelines, still largely unregulated.

The need for an effective RWE oversight program is especially important because of the likelihood of more products coming onto the market after earning FDA approval through the agency's increased reliance on RWE, as encouraged by the Cures Act, and through various other accelerated approval programs.⁵³ In all these instances, manufacturers will still need to demonstrate to payers, providers, and other stakeholders, post-FDA approval, that the products add value, even though there may be limited available data to address lingering uncertainties.⁵⁴ As a result, manufacturers will feel inevitable pressures to turn to RWE for quick results, however flawed.⁵⁵

One clear challenge in developing reliable RWE is appropriately incentivizing front-line clinicians to participate in high-quality data generation. This will require providers to do more than simply record and capture data as usual for billing and treatment. Instead, providers will have to pay greater attention to reporting consistent outcomes, including somewhat more attenuated measures not directly related to immediate treatment and billing activities, that are standardized across different records systems. At present, many front-line clinicians lack sufficient financial incentives to develop high-quality RWE in this manner, especially to offset the additional time and cost involved.

Further, an uptake problem exists as providers may simply tune out RWE as not sufficiently reliable to resolve important decisions. Because providers lack sufficient incentives to help generate high-quality RWE, they may be less likely to pay heed to the results of RWE investigations, especially in the face of likely conflicting studies.⁵⁶ For example, important professional medical societies, such as the American Society of Clinical Oncology and the National Comprehensive Cancer Network, do not yet use RWE in development of their clinical practice guidelines because of, among other reasons, concerns about limitations in the data.⁵⁷ In short, a high-quality RWE program will “ultimately hinge on establishing a clear business case and value-add for those who create, curate, and improve data for RWE applications. Stronger incentives need to be developed for physicians and other providers … to become fully-vested partners in the development of … RWE.”⁵⁸

F. Lesson From CER

The experience with health care reform's push for more comparative effectiveness research (“CER”) offers cautionary lessons for the future of RWE. CER compares treatments against each other to determine which works best.⁵⁹ Like RWE, CER differs from conventional clinical trials as CER studies test relative effectiveness of interventions, typically include treatments already used in clinical practice, and study populations enrolled with more flexible eligibility criteria that are thought to be more representative of typical patients.⁶⁰ CER is one example of the many possible applications for RWE. CER investigations, depending on their design, can rely upon RWE to compare treatment outcomes in actual practice. The Affordable Care Act encouraged greater use of CER through record new funding⁶¹ and the creation of a new federal oversight entity, the Patient-Centered Outcomes Research Institute (“PCORI”), to direct the nation's first comprehensive CER program.⁶²

But attempts to incorporate more RWE into CER studies have encountered problems. One significant challenge is access to the right evidence, consistent with the data access concerns, previously noted, that arise with RWE generally. CER researchers have found it very difficult to access real-world data that has sufficient detail as to individual patient characteristics.⁶³ RWE often involves aggregate data, which may be reported into patient disease registries. But to evaluate questions of comparative effectiveness it is often necessary to explore individual patient characteristics behind each data field. Prominent CER investigations that sought to incorporate RWE, such as the Innovative Medicines Initiative-GetReal Project, have stumbled in gaining access to needed patient characteristics data.⁶⁴

Further, while the Affordable Care Act has generated more CER, it is not clear that front-line clinicians are engaging with it and using it sufficiently in clinical decision-making.⁶⁵ Translating CER into actual medical practice proves quite difficult and requires carefully calibrating the incentives for physicians, including overcoming flawed reimbursement signals and physicians' suspicions about the applicability of the research.⁶⁶ A successful RWE program will likewise need to do much more than just develop the evidence. It will be necessary to create robust incentives for frontline clinicians and other stakeholders to help generate and engage with the data.

A final lesson is the critical need for a comprehensive oversight entity. Like with RWE, CER activities have been around for many years.⁶⁷ But CER did not really take off until passage of the Affordable Care Act. Similar to RWE, private stakeholders typically generated CER.⁶⁸ But before the Affordable Care Act, their efforts were modest, poorly coordinated, and the results often not made publicly available.⁶⁹ And like with RWE, previous CER efforts raised the “Tower of Babel” problem, as CER researchers did not use common data sources or follow uniform standards, undermining the ability to share work, build upon previous studies, and achieve consensus.⁷⁰

Similarly, various medical societies and research groups have proposed best practices for generating and using RWE in the past decade.⁷¹ However, these efforts have not always aligned and “there is little agreement among researchers and other stakeholders on how to optimally do this.”⁷² It may take something as dramatic and comprehensive as the Affordable Care Act to jump-start the health care system's investment in high-quality RWE, as well as to make evidence generation and curation activities more transparent in order to increase trust in RWE more generally.

III. IMPLICATIONS OF RWE FOR HEALTH CARE FRAUD ENFORCEMENT

While it is premature to conclude that the momentum for more RWE will dramatically improve the health care system, it may, in the short term, have troubling implications for the health care fraud and abuse laws. Since the enactment of the Health Insurance Portability and Accountability Act of 1996 (“HIPAA”),⁷³ nearly all federal health care legislation has included provisions expanding the laws and resources devoted to reducing fraud in the federal health care programs.⁷⁴ The Cures Act, by comparison, contains few provisions directly addressing health care fraud. The major change is a new set of penalties applicable to individuals and entities who violate grants, contracts, or other agreements for federal funding.⁷⁵ Modeled on the Civil False Claims Act (“FCA”),⁷⁶ these provisions impose civil monetary penalties for, inter alia, knowingly presenting a false or fraudulent claim under a grant, contract, or agreement, and knowingly making or using a false statement, omission, or misrepresentation in an application, proposal, bid, or progress report.⁷⁷ While enhancing the government's ability to prosecute fraud in federal grants and contracts, these provisions do little to address fraud concerns stemming from the Act's FDA-related reforms. Nonetheless, the ascendance of RWE may well have implications for fraud enforcement, particularly against manufacturers who engage in “off-label” drug promotion. By the same token, recent experience with health care fraud enforcement may offer lessons for the potential fraud implications of increased reliance on RWE.

A. Fraud in the Pharmaceutical Industry

1. Background

Significant resources are spent each year to combat health care fraud and abuse. In fiscal year 2016, for example, the Department of Justice opened 930 new civil and 975 new criminal health care fraud investigations, and won or negotiated over $2.5 billion in health care fraud settlements and judgments.⁷⁸ Since 1997, these efforts have returned approximately $31 billion to the Medicare Trust Funds.⁷⁹ In recent years, the pharmaceutical industry has been a prime fraud enforcement target. Between 1991 and 2015, pharmaceutical manufacturers entered into 373 fraud settlements totaling $35.7 billion, the vast majority of them civil.⁸⁰ Kickbacks, unlawful promotion of pharmaceuticals, and overcharging of government health care programs comprised the lion's share of allegations, many of them brought under the FCA.⁸¹

Although rooted in efforts to prevent fraud on the Union Army during the Civil War, the FCA has evolved into a potent weapon against health care fraud.⁸² The statute prohibits a variety of activities, including submitting false or fraudulent claims to the government and making or using false records or statements material to a false or fraudulent claim.⁸³ The basic false claims provision prohibits a defendant from presenting or causing to be presented a claim for payment or approval, where the claim is false or fraudulent and the acts are undertaken “knowingly”—a standard that includes not only actual knowledge, but also deliberate ignorance and reckless disregard of truth or falsity.⁸⁴ As of February 3, 2017, violations are subject to civil penalties of $10,957 to $21,916 per claim, plus three times the government's damages.⁸⁵

The FCA can be applied not only to “factually” false claims for services that were not provided, but also to “legally” false claims where services were provided, but the claimant also violated a law, regulation, or contract provision.⁸⁶ Legal falsity rests on the concept that claimants make certain promises of truthfulness when they submit claims. Sometimes they do so “expressly,” making explicitly false certifications of compliance; other times, the mere “act of submitting a claim for reimbursement itself [will imply] compliance with governing federal rules,” even without a certification.⁸⁷ If the certification is untrue, the theory holds, the claimant is not entitled to payment.⁸⁸ The certification theories permit FCA cases to be based on violations of a variety of other federal program requirements, including Medicare and Medicaid coverage rules.⁸⁹

Drug promotion also may be actionable under the Federal Anti-Kickback Statute (“AKS”), which prohibits offering, paying, soliciting, or receiving any “remuneration” to induce someone to refer patients or to purchase, order, or recommend any item or service that may be paid for under a federal health care program.⁹⁰ As of February 2018, an AKS violation is a felony subject to up to 10 years of imprisonment and a fine of $100,000, as well as exclusion from the federal health care programs and civil penalties of $100,000.⁹¹ A claim for items or services provided in violation of the AKS also “constitutes a false or fraudulent claim” under the FCA.⁹² This broad criminal statute, designed to limit the influence of financial incentives on health care referral and purchasing decisions, has been particularly important in the pharmaceutical context.⁹³

2. Fraud in Off-Label Promotion

In recent years, one of the most successful targets of both the FCA and the AKS has been pharmaceutical manufacturers' promotion of their drugs for indications beyond the FDA-approved labeling (or “off-label”).⁹⁴ The theory begins with the premise that an approved drug may be considered “misbranded” if its label is false, misleading, or fails to contain adequate directions for use—and, by definition, the label will not contain directions for uses that have not been approved.⁹⁵ An approved drug also may be considered “new” (and unapproved) if it is promoted for an unapproved use.⁹⁶ A manufacturer that promotes a drug for an unapproved use potentially violates both of these requirements.

Legally, physicians are free to prescribe approved drugs for both approved and unapproved uses,⁹⁷ and quite often they do: one oft-quoted study estimated that 21% of outpatient prescriptions written in 2001 were for off-label uses.⁹⁸ Whether those uses are medically appropriate, however, remains controversial. That same study concluded that 73% of identified off-label uses “had little or no scientific support,” although the authors acknowledged that they could not measure the full “gradient of evidence” and thus might have overlooked uses with a lesser degree of scientific support.⁹⁹ A similar investigation concluded that the risk of adverse drug events was 44% higher for off-label compared to on-label drug uses.¹⁰⁰

The FDA's attempt to prevent manufacturers from promoting their drugs for off-label uses has come under attack in recent years, with critics arguing that these limitations are an unconstitutional restriction on free speech.¹⁰¹ Those criticisms reached a crescendo in 2012, when the Second Circuit held in United States v. Caronia that the FDA could not prohibit a pharmaceutical sales representative's truthful statements about off-label uses for one of his company's drugs.¹⁰² The case did not fully resolve the debate, however, as the same court later acknowledged that “Caronia left open the government's ability to prove misbranding on a theory that promotional speech provides evidence that a drug is intended for a use that is not included on the drug's FDA-approved labeling.”¹⁰³

Even if actionable as a violation of FDA rules, however, off-label promotion does not necessarily implicate the fraud and abuse laws. But FDA approval status also is relevant to Medicare and Medicaid coverage, although the programs have different rules. Medicaid, for example, allows states some leeway with regard to drug coverage, while Medicare has different coverage rules for inpatients (Part A), outpatients (Part B), and drugs purchased through private Medicare plans (Part C) and voluntary prescription drug plans (Part D).¹⁰⁴ However, coverage is generally restricted to “medically accepted” indications, which include not only those for which the drug has been approved by the FDA, but also those recognized in specific compendia of medical research.¹⁰⁵ Absent an exception, other uses of the drug are not covered.

The interaction of the FDA off-label restrictions and the Medicare and Medicaid coverage rules potentially brings off-label promotion within the ambit of the FCA. Because some off-label uses are not reimbursable, claims submitted for those uses request payment to which claimants are not entitled. As one district court explained, “the alleged FCA violations arises—not from unlawful off-label marketing activity itself—but from the submission of Medicaid claims for uncovered off-label uses induced by Defendant's fraudulent conduct.”¹⁰⁶ While manufacturers themselves do not generally submit Medicare or Medicaid claims, the FCA also applies to defendants who “cause” false or fraudulent claims to be submitted.¹⁰⁷ When a manufacturer promotes a drug off-label, convincing a physician to prescribe the drug for a use not covered by Medicare or Medicaid, and claims are later submitted for that use, those claims may be considered false. If the manufacturer acted with knowledge of that falsity—or at least with reckless disregard or in deliberate ignorance thereof—the manufacturer may have violated the FCA by causing the submission of false claims.¹⁰⁸

Despite the lingering First Amendment debate, off-label FCA settlements continue.¹⁰⁹ Of the 239 pharmaceutical fraud settlements between 1991 and mid-2012, off-label allegations accounted for 14 of the 20 largest recoveries.¹¹⁰ Numerous off-label settlements were negotiated in the months both leading up to and after Caronia; while some were near completion before the decision was announced in December 2012, others appear to have been undertaken later.¹¹¹ The rate of off-label FCA settlements slowed somewhat in 2014-2015, but it is unclear whether this represents a permanent decline or merely a temporary lull.¹¹² Despite Caronia, numerous factors continue to drive the filing of these cases, including incentives for both the government and manufacturers to settle FCA litigation, the fact that qui tam relators have little reason to refrain from filing these suits, and the economics of FCA recovery, which provide a boost to federal coffers even where the government declines to intervene.¹¹³

B. The Potential Effect of Rwe

The RWE provisions in the Cures Act have no direct impact on the FCA or on current FDA policies regarding off-label promotion.¹¹⁴ The Act provides little assistance in answering the key question of whether FCA cases based on off-label promotion implicate the First Amendment. Nonetheless, the invitation to use RWE to support supplemental approval may have significant implications for current theories of fraud enforcement, particularly cases based on off-label promotion.

From a fraud perspective, the plan to rely on billing records as important sources of RWE should give us pause. As noted earlier, billing records are notoriously unreliable in key respects.¹¹⁵ The Medicare Fee-For-Service 2016 Improper Payments Report, for example, found that 11% ($41.1 billion) of Medicare payments from July 2014 to June 2015 were made incorrectly.¹¹⁶ Almost two-thirds of the errors were due to documentation problems, preventing the auditors from confirming “that the billed services were actually provided, were provided at the level billed, and/or were medically necessary.”¹¹⁷ While the report is quick to point out that improper payments do not necessarily equate to fraud, at the very least they indicate sheer sloppiness.¹¹⁸ If billing records lack complete information on the basic services provided, expecting them to serve as useful sources of more comprehensive RWE may be wishful thinking, at best.

Similarly, experience suggests that permitting the FDA to consider RWE in support of supplemental approval may invite manufacturers to game the system in developing such evidence. Even under the current “gold standard” of RCTs, questions have arisen regarding the legitimacy and transparency of study design, conduct, and publication. The literature is replete with examples of “seeding trials”—trials designed to “seed” the market for a new drug under the guise of studying a legitimate scientific research question—as well as those that are “ghost-managed” by a manufacturer and “ghost-written” by company staff, all while attributing the research to independent medical experts.¹¹⁹ The current system raises concerns that pharmaceutical companies may misrepresent the scientific support for a new use, or commission supplemental clinical studies so inherently biased as to be nearly useless. Without a strong system of FDA oversight and evaluation, an increased focus on non-RCT data may invite companies to cloak an ever-expanding range of marketing efforts as scientific evidence.¹²⁰ The expansion of RWE, combined with alleged kickbacks and the offering of other benefits to physicians to entice them to prescribe particular drugs for off-label uses (behaviors that seem to continue unabated despite increased enforcement), poses significant risks.¹²¹

With regard to the FCA, the Cures Act might affect off-label litigation in two diametrically opposed ways. At one end of the spectrum, the Act might lead to an eventual reduction in off-label FCA cases. If the FDA does in fact create a workable process through which supplemental approval can be granted based on RWE, it should be far easier than it is now for manufacturers to secure approval for additional indications—potentially reducing the universe of off-label uses. Fewer off-label uses, in theory, should mean fewer resources devoted to encouraging physicians to write prescriptions for those uses, and fewer grounds for the filing off-label FCA cases. If this comes to pass, the Cures Act might well reduce off-label FCA litigation, perhaps to a greater extent than the ever-growing enforcement threat.

The more likely result, however, at least in the short term, is a continuation of off-label FCA filings similar to that seen post-Caronia. Rather than providing clarity, the recognition of RWE may further complicate one of the most confounding aspects of this litigation: What, exactly, must be false to support a suit under the FCA? FCA liability rests on the submission of a false or fraudulent claim, which turns simply on whether the services were eligible for reimbursement.¹²² Some off-label uses are ineligible for Medicare or Medicaid payment; in fact, even some on-label uses are excluded from coverage.¹²³ For FCA purposes, coverage is the crucial question. However, both commentators and judges have conflated that inquiry with the separate question of the truth of the scientific statements made about the off-label use.¹²⁴ The question is not the truth or falsity of what the manufacturer said to a physician, but rather whether the resulting claim was eligible for payment. And yet the truthful/untruthful speech divide permeates the FCA off-label debate.

It has long been recognized that the FCA is not an appropriate tool for resolving issues of scientific merit. In the words of the Ninth Circuit, “‘known to be false’ … does not mean ‘scientifically untrue’; it means ‘a lie.’ … What is false as a matter of science is not, by that very fact, wrong as a matter of morals.”¹²⁵ FCA suits that shift the focus from the accuracy of the claim for payment to the scientific truth of the manufacturer's statements regarding off-label use are precisely the types of scientific disagreement the statute is least capable of resolving. Yet, because Medicare and Medicaid coverage rests in part on FDA approval status, it is nearly impossible to extricate the concept of a false claim for payment from questions about the falsity of the off-label statements that may have contributed to submission of that claim in the first place.

Widespread consideration of RWE has the potential to confuse the situation even more. FDA currently requires “substantial evidence” to justify drug approval—“evidence consisting of adequate and well-controlled investigations” demonstrating the drug's effects.¹²⁶ Off-label FCA cases often founder on the problem of assessing a manufacturer's good or bad faith regarding data that, by definition, have not met the substantial evidence threshold.¹²⁷ Just as FDA approval does not guarantee that a drug will work for an individual patient, however, the lack of approval does not always mean the opposite.¹²⁸ The fact that the supporting studies are deemed insufficient for approval does not prove that the drug never works—rather, it means that for the aggregate patient population, the threshold for safety and efficacy has not been satisfied.

The FDA has long been criticized for applying that same standard to statements about off-label uses. As one judge remarked, by “asserting that any and all scientific claims about … prescription drugs are presumptively untruthful or misleading until the FDA has had the opportunity to examine them, FDA exaggerates its overall place in the universe.”¹²⁹ By opening the door to consideration of RWE, the Cures Act serves as at least a partial response to critics who argue that the FDA should be willing to accept evidence beyond traditional RCTs to support supplemental approval.¹³⁰ Yet, at least in the short term, the proliferation of RWE will likely only exacerbate the tendency to conflate scientific with reimbursement truths. A manufacturer that has developed RWE and tries to share the relevant data with physicians, will argue that it merely seeks to share a type of “truthful” information that is now broadly recognized by Congress as relevant to drug safety and efficacy. As the categories of acceptable data broaden and manufacturers compile new forms of RWE, the argument that their representations to physicians are truthful may gain additional traction, possibly emboldening them to finally challenge FCA suits on First Amendment grounds.¹³¹ In short, even under the current RCT system, courts and commentators have had a difficult time confining the proper role of the FCA to reimbursement disputes. When the evidence of “truth” expands to incorporate RWE, that difficulty may become insurmountable.

IV. CONCLUSION

While the Cures Act opens the door to greater acceptance of RWE by the FDA, at least in limited circumstances, successfully incorporating RWE into health care decision-making beyond the FDA requires overcoming considerable challenges. Further, past experience sounds a note of caution. In the context of CER—a similar Congressionally-backed effort to develop data beyond the traditional RCT for assessing the effectiveness of medical treatment options—efforts to integrate RWE have faced significant obstacles. Similarly, in the health care fraud context, greater use of RWE runs the risk of exacerbating problems that already plague enforcement efforts. Without a thoughtful and comprehensive approach to addressing the full range of issues raised by RWE, this evidence may turn out to have limited utility in the real world.

Footnotes

1

21st Century Cures Act, Pub. L. No. 114-255, 130 Stat. 1033 (2016).

2

See infra notes 3-6 and accompanying text.

3

See 21 U.S.C. § 355g(a)(1)-(2) (2016).

4

Id. § 355g(b). FDA expanded on the definition in its guidance about RWE and regulatory decisions for medical devices, describing RWE as “clinical evidence regarding the usage, and potential benefits or risks, of a medical product derived from analysis of [real-world data].” U.S. Food & Drug Admin., Use of Real-World Evidence to Support Regulatory Decision-Making for Medical Devices: Guidance for Industry and Food and Drug Administration Staff 4 (2017), [hereinafter FDA Guidance on Use of RWE]. The real-world data sources include electronic health records, disease registries, billing records, and numerous other data sources that can inform on health status. Id.

5

FDA Guidance on Use of RWE, supra note 4, at 4.

6

FDA Guidance on Use of RWE, supra note 4.

7

FDA's Sentinel Initiative, U.S. Food & Drug Admin., https://www.fda.gov/safety/fdassentinelinitiative/ucm2007250.htm [https://perma.cc/3768-Q8C6]; see also Bipartisan Policy Ctr., Using Real-World Evidence to Accelerate Safe and Effective Cures 5 (2016), (last visited Feb. 25, 2018).

8

Daniel Blumenthal et al., Real-World Evidence Complements Randomized Controlled Trials in Clinical Decision Making, Health Aff. Blog (Sept. 27, 2017), https://www.healthaffairs.org/do/10.1377/hblog20170927.062176/full/. In a RCT, “participants are randomly assigned to receive either the treatment under investigation or, as a control, a placebo or the current standard treatment.” Tom Frieden, Why the ‘Gold Standard’ of Medical Research is No Longer Enough, Stat (Aug. 2, 2017), https://www.statnews.com/2017/08/02/randomized-controlled-trials-medical-research/ [].

9

Frieden, supra note 8.

10

Harriette G.C. Van Spall et al., Eligibility Criteria of Randomized Controlled Trials Published in High-Impact General Medical Journals: A Systematic Sampling Review, 297 JAMA 1233, 1237 (2007).

11

Blumenthal et al., supra note 8.

12

See, e.g., Existence of ‘Trial Effect’ in HIV Clinical Trials Confirmed, Sci. Daily (July 13, 2011), https://www.sciencedaily.com/releases/2011/07/110713182219.htm [].

13

Marc Berger et al., Duke Margolis Ctr. for Health Policy, A Framework for Regulatory Use of Real-World Evidence 5 (2017), .

14

Id.

15

Real World Evidence, Duke Clinical Research Inst., https://www.dcri.org/insights/real-world-evidence/ [].

16

Marc L. Berger & James Harnett, Are Real-World Data and Evidence Good Enough to Inform Health Care and Health Policy Decision-Making?, in Decision Making in a World of Comparative Effectiveness Research: A Practical Guide 97, 98 (Howard G. Birnbaum & Paul E. Greenberg eds., 2017).

17

Anushka Patel et al., Reality and Truth: Balancing the Hope and the Hype of Real-World Evidence, 136 Circulation 260, 262 (2017); see Julie C. Locklear, What is Real-World Evidence, and Why Do We Need It?, Stat (Feb. 2, 2016), https://www.statnews.com/sponsor/2017/02/16/real-world-evidence-need-2/ [].

18

Thomas Reinke, Real-World Evidence Faces Some Real-World Challenges, Managed Care (May 5, 2017), https://www.managedcaremag.com/archives/2017/5/real-world-evidence-faces-some-real-world-challenges [].

19

Spencer B. King III, “Real-World Evidence?” Get Real!, 9 JACC: Cardiovascular Interventions 204, 205 (2016); see infra notes 130-133 and accompanying text.

20

See Amr Makady et al., Practical Implications of Using Real-World Evidence (RWE) in Comparative Effectiveness Research: Learnings from IMI-GetReal, 6 J. Comp. Effectiveness Res. 485, 487 (2017).

21

See, e.g., W. Schuyler Jones et al., The Changing Landscape of Randomized Clinical Trials in Cardiovascular Disease, 68 J. Am. C. Cardiology 1898, 1900 (2016) (describing the various characteristics of pragmatic RCTs).

22

In its guidance for use of RWE in regulatory decisions about medical devices, FDA explains that the real-world data to support RWE is data that informs on health status or relates to care delivery and that can be routinely collected from a wide variety of sources, including electronic health records, disease registries, and mobile devices. FDA Guidance on Use of RWE, supra note 4, at 4. Other definitions of “real-world data” are similarly broad. See, e.g., Duke Margolis Ctr. for Health Policy, supra note 17, at 4 (providing various definitions of real-world data).

23

Duke Margolis Ctr. for Health Policy, supra note 13, at 7.

24

See, e.g., Electronic Health Records (EHR) Incentive Programs, Ctrs. for Medicare & Medicaid Servs., https://www.cms.gov/Regulations-and-Guidance/Legislation/EHRIncentivePrograms/index.html?redirect=/EHRIncentivePrograms/ [] (describing electronic health records incentive programs).

25

Zachary Brennan, Real World Evidence: FDA Commits to Advancing Its Use, Reg. Focus (Sept. 19, 2017), http://prod1.raps.org/Regulatory-Focus/News/2017/09/19/28491/Real-World-Evidence-FDA-Commits-to-Advancing-its-Use/ [].

26

See Patel et al., supra note 17, at 261.

27

See Richard White, Building Trust in Real-World Evidence and Comparative Effectiveness Research: The Need for Transparency, 6 J. Comp. Effectiveness Res. 5, 5 (2016) (explaining the risk of significant bias if “treatment options are compared without adjustment for confounders such as patient demographics, disease severity, comorbidities and concomitant medications.”).

28

Patel et al., supra note 17, at 261.

29

Brett Davis et al., Deloitte Ctr. for Health Sols., Getting Real With Real-World Evidence 1 (2017), .

30

Id. at 9.

31

See Makady et al., supra note 20, at 488 (“[P]atient-level data is subject to strict privacy rules ….”); see generally FDA Guidance on Use of RWE, supra note 4, at 15-16 (explaining how the FDA will consider patient data privacy when reviewing sources of RWD).

32

See Davis et al., supra note 29, at 10 (“Researchers often have to clean RWD, especially from nontraditional sources, since most of it is not collected for research purposes.”).

33

Id.

34

Id.

35

Cf. Yale Collaboration for Research Integrity & Transparency, Public Comment on the Food and Drug Administration's Draft Guidance “Use of Real-World Evidence to Support Regulatory Decision-Making for Medical Devices” 6 (Oct. 24, 2016), (“Previous work has shown that participants in postmarketing [RWE] studies for carotid artery stenting have different characteristics and lower mortality than nonparticipants and, thus, results for post-marketing studies may not be reliably extrapolated to the real-world population of patients receiving the device.”).

36

White, supra note 27, at 6.

37

Rachel E. Sherman et al., Real-World Evidence—What is It and What Can It Tell Us?, 375 New Eng. J. Med. 2293, 2293 (2016).

38

See Thomas E. Hubbard & Rebecca Paradis, Real World Evidence: A New Era for Health Care Innovation 2, Sept. 2015, .

39

White, supra note 27, at 6.

40

See id. at 5 (“The observational nature of RWE means that, unlike RCTs, patients cannot be randomized to different treatment options ….”).

41

See id. at 5.

42

See generally Richard L. Schilsky, Finding the Evidence in Real-World Evidence: Moving from Data to Information to Knowledge, 224 J. Am. C. Surgeons 1, 3 (2017).

43

Id.

44

White, supra note 27, at 6; see infra notes 134-136 and accompanying text.

45

FDA Guidance on Use of RWE, supra note 4, at 8.

46

Clinical Trials Registration and Results Information Submission, 81 Fed. Reg. 64982, 64982-83 (Sept. 21, 2016).

47

See, e.g., Monique L. Anderson et al., Compliance with Results Reporting at ClinicalTrials.gov, 372 New Eng. J. Med. 1031, 1031 (2015); Chris Cai, New Rule on Clinical Trial Reporting Doesn't Go Far Enough, Stat (Jan. 17, 2017), https://www.statnews.com/2017/01/17/clinical-trial-reporting-new-rule/ [].

48

Duke Margolis Ctr. for Health Policy, supra note 13, at 13.

49

Schilsky, supra note 42, at 6.

50

Blumenthal et al., supra note 8.

51

See, e.g., FDA Guidance on Use of RWE, supra note 4, at 12 (“Whether evidence is sufficiently relevant and reliable for use will, in part, depend on the level of quality necessary to make a particular regulatory decision.”).

52

See id. at 9.

53

See Fast Track, Breakthrough Therapy, Accelerated Approval, Priority Review, FDA (Feb. 23, 2018), https://www.fda.gov/ForPatients/Approvals/Fast/ucm20041766.htm [].

54

Sherman et al., supra note 37, at 2295.

55

Id.

56

See supra notes 37-39 and accompanying text.

57

Reinke, supra note 18.

58

Duke Margolis Ctr. for Health Policy, supra note 13, at 13.

59

See Richard S. Saver, Health Care Reform's Wild Card: The Uncertain Effectiveness of Comparative Effectiveness Research, 159 U. Pa. L. Rev. 2147, 2147 (2011) [hereinafter Saver, Health Care Reform's Wild Card].

60

Id. at 2151.

61

42 U.S.C. § 1320e(b)(3) (2010).

62

Id. § 1320e(b)(1), (c).

63

Makady et al., supra note 20, at 486.

64

See id.

65

See generally Saver, Health Care Reform's Wild Card, supra note 59, at 2175-98 (discussing several of the reasons physicians may be hesitant to rely on CER information).

66

Id.

67

Alvin I. Mushlin & Hassan M.K. Ghomrawi, Comparative Effectiveness Research: A Cornerstone of Healthcare Reform?, 121 Transactions Am. Clinical & Climatological Ass'n 141, 141 (2010) (explaining that CER has “existed for a long time” but has had various names throughout its time).

68

See Ali Riaz et al., Comparative Effectiveness Research in the United States: A Catalyst for Innovation, 4 Am. Health Drug Benefits 68, 68-69 (2011).

69

Saver, Health Care Reform's Wild Card, supra note 59, at 2158.

70

Id.

71

See, e.g., Marc L. Berger et al., A Questionnaire to Assess the Relevance and Credibility of Observational Studies to Inform Health Care Decision Making: An ISPOR-AMCP-NPC Good Practice Task Force Report, 17 Value Health 143, 143 (2014).

72

Jennifer Graff, Opinion, The Use of Real-World Evidence in Health Care Decision-Making, Morning Consult (Sept. 13, 2017), https://morningconsult.com/opinions/understanding-use-real-world-evidence-health-care-decision-making/ [].

73

See, e.g., Health Insurance Portability and Accountability Act of 1996, Pub. L. No. 104-191, §§ 241-49, 119 Stat. 2016-21 (codified at 18 U.S.C. §§ 24, 1347, 669, 1035, 1518, 1956(c)(7), 1345(a)(1)-(2), 3486, 982(a)) (revising criminal law provisions relating to health care fraud).

74

42 U.S.C. §§ 1320a-7, 1320a-7a, 1320a-7b, 1320a-7d (2012) (expanding administrative penalties); id. § 1320a-7c (establishing the Health Care Fraud and Abuse Control Program).

75

21st Century Cures Act, Pub. L. No. 114-255, § 5003, 130 Stat. 1188 (2016) (codified at 42 U.S.C. § 1320a-7a(o)-(s) (2016)).

76

31 U.S.C. §§ 3729-3733 (2009).

77

42 U.S.C. § 1320a-7a(o) (2016).

78

Dep't of Health & Human Servs. & Dep't of Justice, Health Care Fraud and Abuse Control Program Annual Report for Fiscal Year 2016 1 (2017), .

79

Id.

80

Sammy Almashat et al., Twenty-Five Years of Pharmaceutical Industry Criminal and Civil Penalties: 1991 Through 2015, Public Citizen 10 (Mar. 2016), .

81

Id. at 55-56.

82

See Act of Mar. 2, 1863, ch. 67, 12 Stat. 696; see also S. Rep. No. 99-345, at 6 (1986), reprinted in 1986 U.S.C.C.A.N. 5266, 5273.

83

31 U.S.C. § 3729(a)(1)(A) (2009); id. § 2739(b).

84

31 U.S.C. § 3729(b)(1) (2009). One reason the FCA has been so effective is the law's qui tam provision, which permits private “relators” (including competitors, employees, and patients) to sue on the government's behalf and receive a portion of the proceeds. See id. § 3730(d).

85

28 C.F.R. § 85.5 (2018).

86

See, e.g., United States ex rel. Conner v. Salina Reg'l Health Ctr., Inc., 543 F.3d 1211, 1217 (10th Cir. 2008).

87

United States ex rel. Mikes v. Straus, 274 F.3d 687, 698-99 (2d Cir. 2001), abrogated by Universal Health Servs., Inc. v. United States ex rel. Escobar, 136 S. Ct. 1989 (2016) (affirming validity of the implied certification theory). For a thorough discussion of the certification theories, see Joan H. Krause, Reflections on Certification, Interpretation, and the Quest for Fraud that “Counts” Under the False Claims Act, 2017 U. Ill. L. Rev. 1811 (2017).

88

See generally Escobar, 136 S. Ct. at 2000-02 (alleging that mental health clinic's Medicaid claims contained implied representations of compliance with regulations, but in reality clinic had violated state licensing and supervision rules).

89

See id.

90

42 U.S.C. § 1320a-7b(b) (2015).

91

Id.; see also id. §§ 1320a-7(a) (mandatory exclusion upon conviction), 1320a-7(b)(7) (permissive exclusion), 1320a-7a (civil monetary penalties); Bipartisan Budget Act of 2018, H.R. 1892, § 50412.

92

42 U.S.C. § 1320a-7b(g) (2015).

93

See, e.g., Press Release, U.S. Dep't of Justice, Amgen Inc. Pleads Guilty to Federal Charge in Brooklyn, NY.; Pays $762 Million to Resolve Criminal Liability and False Claims Act Allegations (Dec. 19, 2012), http://www.justice.gov/opa/pr/amgen-inc-pleads-guilty-federal-charge-brooklyn-ny-pays-762-million-resolve-criminal [] (alleging that manufacturer offered kickbacks designed to influence physicians' prescribing decisions).

94

For a detailed analysis, see Joan H. Krause, Truth, Falsity, and Fraud: Off-Label Drug Settlements and the Future of the Civil False Claims Act, 71 Food & Drug L.J. 401 (2016) [hereinafter Krause, Truth, Falsity, and Fraud].

95

21 U.S.C. §§ 331(a), 352(a), 352(f) (2016).

96

21 C.F.R. § 310.3(h)(4) (2018) (explaining that a drug may be “new” with regard to the treatment of one condition, even if approved to treat another).

97

See, e.g., Citizen Petition Regarding the Food and Drug Administration's Policy on Promotion of Unapproved Uses of Approved Drugs and Devices; Request for Comments, 59 Fed. Reg. 59,820, 59,821 (proposed Nov. 18, 1994).

98

David C. Radley et al., Off-Label Prescribing Among Office-Based Prescriptions, 166 Arch. Internal Med. 1021, 1021 (2006); see also American Society of Clinical Oncology, Reimbursement for Cancer Treatment: Coverage of Off-Label Drug Indications, 24 J. Clinical Oncology 3206, 3206 (2006) (estimating that approximately half the uses for cancer drugs are off-label); see generally John N. Joseph et al., Enforcement Related to Off-Label Marketing and Use of Drugs and Devices: Where Have We Been and Where Are We Going?, 2 J. Health & Life Sci. L. 73, 79-80 (2009) (describing common off-label use of older drugs).

99

Radley et al., supra note 98, at 1022-23.

100

Tewodros Eguale et al., Association of Off-Label Drug Use and Adverse Drug Events in an Adult Population, 176 JAMA Internal Med. 55, 58 (2016).

101

See United States v. Caronia, 703 F.3d 149, 160 (2d Cir. 2012); see also Washington Legal Found. v. Henney, 56 F. Supp. 2d 81, 86-87 (D.D.C. 1999) (finding that restrictions on a manufacturer's ability to share information about off-label uses of pharmaceutical drugs unconstitutionally restricts protected commercial speech).

102

Caronia, 703 F.3d at 168-69.

103

United States ex rel. Polansky v. Pfizer, Inc., 822 F.3d 613, 615 n.2 (2d Cir. 2016) (emphasis added).

104

See, e.g., 42 U.S.C. § 1396r-8(d)(1) (2016) (permitting states to limit Medicaid drug coverage); see also, e.g., Ctrs. for Medicare & Medicaid Servs., Medicare Benefit Policy Manual § 30 (Rev. 234, Mar. 10, 2017), (explaining Medicare Part A coverage of drugs in the hospital setting).

105

See, e.g., 42 U.S.C. §§ 1396b(i)(1), 1396r-8(k)(3), 1396r-8(k)(6) (2018) (defining a medically accepted indication as a use approved by the FDA or “which is supported by one or more citations included or approved for inclusion in” the specified compendia); id. § 1395w-102(e)(4)(A)(ii) (applying same definitions to Medicare Part D).

106

United States ex rel. Franklin v. Parke-Davis, 147 F. Supp. 2d 39, 52 (D. Mass. 2001).

107

See 31 U.S.C. § 3729(a)(1)(A)-(B) (2018).

108

See id.

109

There has been no definitive ruling on whether such suits implicate the First Amendment, although some courts have expressed skepticism regarding the underlying FCA theory. See, e.g., United States ex rel. Solis v. Millennium Pharms., Inc., No. 2:09-cv-03010-MCE-EFB, 2015 WL 1469166, at *7 (E.D. Cal. Mar. 30, 2015) (denying motion to dismiss off-label FCA allegations). Cf. United States ex rel. Polansky v. Pfizer, Inc., 822 F.3d 613, 619-20 (2d Cir. 2016) (affirming dismissal of FCA claims; expressing skepticism that “anyone could be identified who actually submitted a false claim” under the facts, but acknowledging that a relator might be able to satisfy pleading requirements “in a case in which it would be obvious to anyone that the use promoted is one that is not approved ….”).

110

Sammy Almashat & Sidney Wolfe, Pharmaceutical Industry Criminal and Civil Penalties: An Update, Public Citizen 45 (Sept. 2012), .

111

See Krause, Truth, Falsity, and Fraud, supra note 94, at 417-18, tbl.1.

112

See Almashat & Wolfe, supra note 110, at 5, 13-17 (discussing possible factors for decline in settlements); see also John Bentivoglio et al., Recent Settlements Suggest Off-Label Cases Aren't Extinct, Law360 (Aug. 30, 2017, 12:30PM), .

113

See Krause, Truth, Falsity, and Fraud, supra note 94, at 427-30; see, e.g., Press Release, U.S. Dep't of Justice, Pharmaceutical Companies to Pay $67 Million to Resolve False Claims Act Allegations Relating to Tarceva (June 6, 2016), https://www.justice.gov/opa/pr/pharmaceutical-companies-pay-67-million-resolve-false-claims-act-allegations-relating-tarceva [].

114

The Cures Act did expand an existing exception permitting “health care economic information” about off-label uses to be shared with certain entities involved in drug coverage and reimbursement decisions in limited circumstances, but that provision is not relevant to most physician interactions. See 21 U.S.C. § 352(a) (2016); see also Sam F. Halabi, Off-Label Marketing's Audiences: The 21st Century Cures Act and the Relaxation of Standards for Evidence-Based Therapeutic and Cost-Comparative Claims, 44 Am. J.L. & Med. 181 (2018).

115

See supra note 19 and accompanying text.

116

Ctrs. for Medicare & Medicaid Servs., Medicare Fee-For-Service 2016 Improper Payments Report 1 (July 27, 2017), [hereinafter CMS Improper Payments Report 2016]; see also Joan H. Krause, Following the Money in Health Care Fraud: Reflections on a Modern-Day Yellow Brick Road, 36 Am. J.L. & Med. 343, 345-49 (2010) (analyzing error rates).

117

CMS Improper Payments Report 2016, supra note 116, at 14.

118

Id. at 1. See also Malcolm K. Sparrow, License to Steal: How Fraud Bleeds America's Health Care System 2 (Westview Press updated ed. 2000) (“What you see is not the problem. It's what we don't see that really does the damage ….”).

119

See, e.g., Kevin P. Hill et al., The ADVANTAGE Seeding Trial: A Review of Internal Documents, 149 Annals Internal Med. 251, 251 (2008) (describing seeding trial for Vioxx); Samuel Krumholz et al., Study of Neurontin: Titrate to Effect, Profile of Safety (STEPS) Trial, 171 JAMA Arch. Internal Med. 1100, 1100 (2011) (describing seeding trials); see generally Sergio Sismondo, Ghost Management: How Much of the Medical Literature is Shaped Behind the Scenes by the Pharmaceutical Industry?, 4 PLoS Med. 1429 (2007).

120

See, e.g., FDA Guidance on Use of RWE, supra note 4, at 13-14 (describing criteria for relevance and reliability of RWE in the medical device context); see generally Jonathan P. Jarow et al., Multidimensional Evidence Generation and FDA Regulatory Decision Making: Defining and Using “Real-World” Data, 318 JAMA 703 (2017) (describing types of data sources that might qualify); Aaron S. Kesselheim & Michelle M. Mello, Prospects for Regulation of Off-Label Drug Promotion in an Era of Expanding Commercial Speech Protection, 92 N.C. L. Rev. 1539, 1585 (2014) (raising concern that manufacturers might produce studies of dubious quality to support off-label uses); see generally Hubbard & Paradis, supra note 38.

121

See, e.g., United States ex rel. Solis v. Millennium Pharms., Inc., No. 2:09-cv-03010-MCEEFB, 2015 WL 1469166, at *6 (E.D. Cal. Mar. 30, 2015) (alleging manufacturer violated Anti-Kickback Statute and FCA by funding grants, paying excessive fees to speakers, honoraria, meals, attendance at CME events controlled by the company, and funding advisory boards and preceptorship opportunities, all to entice physicians to write off-label prescriptions).

122

See, e.g., United States ex rel. Franklin v. Parke-Davis, 147 F. Supp. 2d 39, 52-53 (D. Mass. 2001) (describing FCA liability theory).

123

See Ctrs. for Medicare & Medicaid Servs., Medicare Prescription Drug Benefit Manual § 20.1 (Rev. 18, Jan. 15, 2016), [hereinafter CMS Medicare Prescription Drug Benefit Manual] (describing classes of drugs excluded from Part D coverage, including drugs used to treat anorexia, weight loss, or weight gain; fertility drugs; and drugs used for cosmetic purposes or hair growth).

124

See, e.g., Franklin, 147 F. Supp. 2d at 52-53 (refusing to dismiss Medicaid off-label allegations, but noting “a much closer question” would have been raised if the allegations involved unlawful but truthful company statements).

125

Wang v. FMC Corp. 975 F.2d 1412, 1421 (9th Cir. 1992), overruled by United States ex rel. Hartpence v. Kinetic Concepts, Inc., 792 F.3d 1121 (9th Cir. 2015); see also Krause, Truth, Falsity, and Fraud, supra note 94, at 431-37.

126

21 U.S.C. § 355(d) (2017).

127

See Kesselheim & Mello, supra note 120, at 1583 (noting evidentiary difficulties in proving promotional statements are false and misleading).

128

See, e.g., Sandra H. Johnson, Polluting Medical Judgment? False Assumptions in the Pursuit of False Claims Regarding Off-Label Prescribing, 9 Minn. J.L. & Tech. 61, 117 (2008) (“The absence of clinical trials does not mean that [the drug] Neurontin … is not effective in treating these highly similar pain states just as FDA approval in 2002 did not make the drug effective for treating pain.”).

129

Washington Legal Found. v. Friedman, 13 F. Supp. 2d 51, 67 (D.D.C. 1998).

130

See, e.g., Kesselheim & Mello, supra note 120, at 1598-1603 (encouraging expanded acceptance of “substantial clinical experience” to support certain drug advertisements and encouraging expansion of “substantial evidence” to encompass certain observational research); Krause, Truth, Falsity, and Fraud, supra note 94, at 435 (discussing other standards); 21 U.S.C. § 352(a) (2016) (permitting “[h]ealth care economic information provided to a payor, formulary committee, or other similar entity” to be “based on competent and reliable scientific evidence ….”).

131

See generally Krause, Truth, Falsity, and Fraud, supra note 107.