A series of thoughtful articles in the spring tackled some basic questions about the drug development and approval paradigm. I’ll be reviewing other SSRN postings in later posts, but the articles described below work well together — presenting fundamental questions about how drugs should be studied (when and by whom), how the resulting information is processed, and what information (how much and what type) should be required before drugs are permitted on the market. And they tee up what might be a spirited debate about what, exactly, a “public health” approach to new drug approval might mean.
- Cynthia Ho, First Amendment Overprotection of “Alternative Facts”: The Case of Cognitive Biases with Pharmaceutical Marketing.
In this article forthcoming in the Indiana Law Journal, Professor Ho argues that pharmaceutical companies present “alternative facts” when marketing their products, that doctors are subjective to cognitive biases making them susceptible to these facts, and that judges are subject to cognitive biases making them unable to see the vulnerability of doctors to manipulation. Ho — who has written before on cognitive bias affecting pharmaceutical law and policy — questions the “schema” that doctors are adequately sophisticated to evaluate drug marketing (as inconsistent with cognitive bias literature) and the “schema” that disclosure (more information) is inherently desirable (as contrary to literature about how people process information). She argues for changes to commercial speech law, so that speech about unapproved (off-label) uses receives less First Amendment protection. FDA already thinks a fair amount about how people process information, but she believes that the agency might resist pressures to modify its policies (in response to various recent First Amendment rulings) if it understood those rulings were not consistent with cognitive bias science.
I lump the Ho article with the next article because Professor Ho concludes with a brief call to “reconsider” the current approach to drug development and approval, that is, where responsibility to study new drugs falls to the companies that will market those drugs.
- Nicholson Price, Drug Approval in a Learning Health System.
In this highly readable article published in the Minnesota Law Review, Professor Price concerns himself with the search for a better drug approval process. He describes a broad move in the healthcare system to a “learning health system” — one in which data about clinical care are collected continuously and used (also continuously) to learn about and improve that care. And he shows that the current drug development and approval paradigm is, somewhat, a learning paradigm. For instance, accelerated approval permits market entry on the basis of fewer data with studies after approval, and since 2007 FDA has had statutory power to require postmarket research. Rather than a hard line (a particular type of complete data set always required at the moment of drug approval), there is some “line blurring” — which, he argues, trades earlier access with less information for the promise of later information.
Like other scholars, whose work he discusses in a helpful literature review, Price concludes that the tradeoff is not quite working; the gathering of later information remains a challenge. Nevertheless, he argues persuasively that the system is moving (slowly) towards a “lifecycle” approach to evidence development. The interesting stuff comes when he moves to discussion of a “learning health system” more broadly (including capture and use of data from clinical care) and — in section IV — describes the informed consent rules and privacy rules that are likely to hamper and bias the development and integration of information needed to make such a system really effective and efficient. If policymakers are persuaded that our health care system should include rapid integration of continuously developed data from clinical practice, section IV is worth their attention.
- In this context, it’s worth noting a paper published by George Mason’s Mercatus Center — The Proper Role of the FDA for the 21st Century, written by Jason Briggeman, Ethan Roberts, and Joseph Gulfo.
This paper — which is a white paper rather than a law review article — also takes up the tradeoff between data and access, but the authors see things very differently. Where Professor Price sees the glass half full, they see it half empty
Briggeman and co-authors argue that FDA’s premarket approval process has become unduly conservative. They argue that FDA’s focus on overall risk-benefit balance in approval decisions is relatively new and that it leads to data requirements that protract premarket timelines and add to post-approval research burdens. In their view, a “public health” approach to new drug approval decision-making at FDA would simply focus on whether a submitted application shows the drug is safe and effective for the labeled uses. A drug would be “effective” so long as it had “biological activity related to a disease.” FDA would not — in their model — look at overall risk-benefit for patients generally speaking, nor would it speculate about clinical utility in real-world settings.
To be fair, though, the agency has always focused on risk-benefit balance. I think they have some of the history wrong — for instance, in claiming that FDA used to focus on pharmacologic activity rather than clinical utility. It’s true that the agency has used surrogate endpoints for approvals for nearly all of its history, but I am not sure it’s true that the agency is relying on them less now. Indeed, there has been increased interest in use of novel (not validated) surrogate endpoints since the 1980s. But the authors are correct that the agency has in recent years asked some companies for large outcomes trials, delaying approval and adding to post-approval burden. I talk about some of this history in my article, The Drug Innovation Paradox (see pages 66 to 77).
What Briggeman (and colleagues) and Price share is a realization that the modern medical marketplace is dynamic, with valuable real-world evidence emerging from clinical practice and continuing clinical research. Briggeman would have the information pushed out at medical meetings and in literature, allowing the market (physicians and payers) to respond in natural fashion. Professor Ho, I think, although I am putting words in her mouth, would probably raise questions about the ability of physicians to respond appropriately, particularly if the industry is involved in pushing the information out. Compare Briggeman (“there is no doubt that some doctors feel that post-approval studies performed by industry, as well as independent clinical investigators, are well-suited for providing evidence of clinical benefit”) with Ho (“even doctors who are skeptical about whether industry information is accurate may nonetheless rely on inaccurate information due to the cognitive bias of accessibility”). These authors are talking about different things (the information that doctors want versus whether doctors are competent to assess this information), but on a basic normative question (the desirability of a model in which physicians make treatment decisions relying heavily on post-approval research performed by industry) they come out in a very different place.
The bottom line in the Briggeman piece is revolutionary. Among other things, the authors argue, Congress should define safety for purposes of drug approval with regard to likelihood of death, disability, or severe harm. The marketplace, they explain, can sort out appropriate uses for medicines that present other safety issues. Also Congress should define effectiveness only as “positive activity on the disease” which includes an impact on biomarkers. Finally, they argue, there should be a “strengthened norm against undue criticism of the FDA by Congress.”
- But consider, in response, a very thoughtful piece in the Food & Drug Law Journal by my co-blogger Patricia Zettler, with Margaret Foster Riley and Aaron Kesselheim: Implementing a Public Health Perspective in FDA Drug Regulation.
Motivated in part by the public health crisis arising from opioid misuse and addiction, these authors argue precisely the opposite: that FDA’s focus in drug approval on the benefits and risks of a proposed drug is too narrow and that the agency should consider the real-world use and public-health impact of the drug when making the approval decision.
As Zettler and colleagues point out, FDA’s evaluation of new drugs is understood as “drug specific” — the agency is viewed as focusing on the benefits and risks of a product when used for the specified indication (in accordance with its labeling) as defined in the premarket trials. (They think this is unduly narrow; Briggeman and colleagues think this is too broad.)
For drugs like opioids and antibiotics, with externalities, Zettler and colleagues argue, the agency should consider additional information, taking a “public health” approach. After all, the agency has a public health mission and the relevant institutional expertise. Readers will want to pay close attention to section II of the article, especially pages 239 to 247, in which they argue that FDA already has sufficient statutory authority to take a broader perspective. Starting with a Scalia quote that “[s]tatutory construction  is a holistic endeavor,” they build an argument from statutory and regulatory language as well as several canons of construction. It would undoubtedly be possible to write an equally thoughtful brief on the other side. And they are careful to acknowledge weaknesses in their argument. But the analysis is very carefully done, and FDA-law practitioners and scholars will enjoy the read. They also cite numerous helpful precedents, such as FDA’s consideration of herd immunity when evaluating vaccines, and its focus on actual use for OTC status.
When should FDA take this broader approach? According to Zettler and colleagues, opioids invite a public health approach to approval — as do other drugs with potential for misuse (such as benzodiazepines and gabapentin). Antimicrobial drugs might warrant it, and — indeed — so might any prescription drug with common (or expected-to-be-common) “off label” uses. And what data should FDA use for implementation of this public health approach? Not just data from randomized controlled trials but also data regarding population health impact — real-world evidence and patient experience data, for instance. Thus, in a footnote, they note additional uses for these data — citing Professor Price’s article (above) on drug approval in a learning health system.
In sum, then, these present two very different views on what might constitute a “public health” approach to new drug approval at FDA — and thus two very different perspectives on the direction that the agency should take. Briggeman and colleagues argue that FDA has strayed from its statute and propose that Congress step in to effect the changes (narrowing of scope) they recommend. Zettler and co-authors argue that the agency already has the authority to take the steps (broadening of scope) they recommend, and indeed that the agency is moving in the right direction already.