The Law, Economics, and Medicine of Off-label Prescribing

• Jurisdiction: United States,Federal
• Citation: Vol. 91 No. 1
• Publication year: 2021

THE LAW, ECONOMICS, AND MEDICINE OF OFF-LABEL PRESCRIBING

William S. Comanor(fn*) & Jack Needleman(fn**)

Abstract: There is a major dissonance in the current structure of regulating new drugs that have more than one medical indication. Physicians are authorized to prescribe these drugs for all indications including those beyond their approved purposes. However, product manufacturers are expressly prohibited from marketing or promoting their drugs for any purpose other than those which have been specifically indicated. While prescribing physicians are encouraged to gain medical information on any additional indications, they cannot obtain it from one of its most likely sources: the drug's supplier.

The Second Circuit Court of Appeals' recent opinion in United States v. Caronia has challenged this regulatory structure. For the three states in the Second Circuit, although not the rest of the country, the FDA's regulations prohibiting promotion of non-approved indications have been restricted.

In this Article, we review the legal, economic, and medical aspects of the FDA's current regulatory approach, and explore the likely consequences of a widespread adoption of the Caronia rule.

INTRODUCTION ................................................................................ 119

I. PHARMACEUTICAL EFFICACY AND EFFECTIVENESS ..... 120

II. THE LAW AND REGULATION OF PRODUCT LABELING ... 125

III. THE LAW AND REGULATION OF ADVERTISING AND PROMOTION ................................................................................ 128

IV.THE ECONOMICS OF OFF-LABEL PRESCRIBING ................ 133

V. THE MEDICINE OF OFF-LABEL PRESCRIBING .................... 137

CONCLUSION .................................................................................... 143

INTRODUCTION

There is a major dissonance in the current structure of regulating new drugs that have more than a single medical indication. Physicians are authorized to prescribe these drugs for all indications including those beyond their approved purposes. However, product manufacturers are expressly prohibited from marketing or promoting their drugs for any purpose other than those which have bee n specifically indicated.(fn1) Thus, while prescribing physicians are encouraged to gain medical information on any additional indications, the information that physicians can obtain from the most likely source-the drug's supplier-is substantially constrained.(fn2)

Although the Food and Drug Administration (FDA) originally accentuated this dissonance, it has more recently retreated from that posture; first under pressure from the statutory admonitions of 1997,(fn3) and subsequently due to the Second Circuit's opinion in United States v. Caronia.(fn4) However, the issue remains in flux and is the subject of this Article.

In succeeding Parts, we review the legal, economic, and medical aspects of this dissonance: between what physicians are authorized to prescribe and what information drug manufactures are permitted to provide about their products. A critical feature of this dissonance is its connection to the two separate types of information about the therapeutic properties of pharmaceuticals, so we start with a discussion of this distinction. Finally, we suggest some policy conclusions to be drawn for this discussion.

I. PHARMACEUTICAL EFFICACY AND EFFECTIVENESS

The U.S. drug approval process is a multi-stage process involving the identification of a potential drug and various trials that must be met to discern its safety and efficacy. The formal approval process requires manufacturers to submit a New Drug Application (NDA), which the FDA reviews in its decision-making process on whether to approve a drug for sale. Critically, drugs are approved only for the specific indications disclosed in the firm's NDA.

An essential part of the NDA is its report on the three formal stages of testing required by the FDA. Phase I, usually conducted on healthy volunteers, focuses on safety and potential side effects, and may also be used to understand how the drug is metabolized.(fn5) Phase II examines whether the drug appears to be effective for a specific indication, where the proposed drug is compared to a placebo or another drug.(fn6) Safety and side effects continue to be assessed in these trials.(fn7) Phase III is a much larger trial which assesses the efficacy of the drug in different subpopulations and at different dosages.(fn8) Such trials can vary in their complexity, but their inferences of efficacy are fundamentally based on the statistical tests of the differences in outcomes in the patients treated with the drug and those treated with placebos or alternatives.(fn9) Given the expense of Phase III trials and the numbers of patients required to assure that differences in outcomes are unlikely to be the results of sampling variation between the treated and control groups, the outcomes and indications studied in these trials are often quite limited.(fn10)

At the heart of the ongoing policy debates concerning off-label prescribing lies the distinction between pharmaceutical "efficacy" and "effectiveness." That distinction follows from the different types of information that can potentially be gleaned on the therapeutic benefits gained from taking pharmaceuticals. Consider the difference between the information obtained from a formal clinical trial of a prospective drug and the information gathered from medical practice and experience resulting largely from observational studies.

The clinical trials required by the FDA to be included in a company's NDA make little use of any substantive knowledge of the drugs being studied. The judgment that a drug is efficacious or not is based on the results of a randomized control trial, in which judgments on efficacy are made by ruling out, via statistical theory, that difference in outcomes between the treatment and control group are simply due to sampling variation.(fn11) Randomization is presumed sufficient to balance the observable and unobservable factors that might influence outcomes.(fn12) And confidence in the results is enhanced by including only a narrow group of patients with limited variation in key characteristics and by maintaining high standards for protocol fidelity.(fn13) To a great extent, the fundamental discipline underlying the trials is not pharmacology but statistics.

In contrast, assessments of a drug's effectiveness rely on experience and medical observation in patient populations.(fn14) Understanding the mode of action of the underlying active ingredient can be critical in a clinician's judgment about whether a particular use is appropriate, and these judgments are refined by extension to other settings. Note that this reliance typically requires a clear understanding of the drug's pharmacology.

Both methods have their strengths and their weaknesses. The clinical trials used to demonstrate a drug's efficacy depend critically on the sample of patients being tested. Clinical trials strictly pertain only to the population from which the sample is drawn. If the results are extrapolated or generalized to apply to populations beyond those included in the clinical trial, the therapeutic effects found in the clinical trials may not apply. Furthermore, statistical tests are generally applied to mean values which can be misleading when the variance of individual outcomes is large. For drugs which are effective only for a limited segment of the patient population, moreover, the positive effect on that segment may be obscured by the drug's unresponsiveness in the rest of the population.

In addition, statistical tests require the selection of a particular level of statistical significance, which in effect defines the trade-off between Type I and Type II errors.(fn15) Because of the influence of random or individual factors, minimizing the risk of approving an inefficacious drug means tolerating increased risks of disapproving efficacious drugs. The need to select among these types of error is an inevitable attribute of employing statistical methods for drug approval. Moreover, the level of statistical significance is typically fixed by standard practice without regard to the potential risks and benefits of a particular drug.

In contrast, relying on observational data has its own problems. Outcomes invariably depend on the particular patients observed, and one never knows whether a specific patient is typical or not. In addition, patients in observational studies are not selected randomly so that judgments of a drug's effectiveness may require dealing with substantial variation among patients along with differences in dosages as well as between planned and actual use of dosage regimens.(fn16) As a result, the patient outcomes in observational studies may not represent the typical response to the drug. The relevant information includes both case reports and trials noted in the medical literature. These studies rely on an understanding of medical modes of action so there is more than mere statistics involved. In determining drug effectiveness, pharmacological understanding plays a major role.

For new pharmaceuticals, the clinical trial data contained in the NDA is the only available basis for assessing efficacy. On the other hand, for drugs already on the market that may have been used extensively by physicians for non-indicated purposes, the medical literature is a prominent source of...