Who Owns the FDA's Thinking?

A law firm says that FDA’s rejection letters reveal too much. That may be exactly why the public needs them.

The FDA is charged with approving drugs only once they are shown to be safe and effective. But what does it mean for a drug to be safe and effective? And, equally importantly: how do the drug makers know whether they’ve provided sufficient proof that their drug is safe and effective? This sounds like a simple question, but it turns out to be the toughest one in all of drug regulation. It sits at the messy border of science, public administration, and law, and that creates a lot of tension.

One significant tension is between the desire for the FDA to be both predictable and particular. If you are developing a new drug, you should have some idea of how the FDA will weigh your evidence of safety and effectiveness. But perfect predictability is impossible. Each new drug introduces new scientific and regulatory questions, and it’s difficult to anticipate precisely how the FDA will answer those novel questions.

The FDA has gotten a lot better at being predictable. They issue guidance documents that lay out their approval standards, and they meet with drug companies before they start their studies to set expectations. The improvement is borne out in the approval statistics: in recent years, novel drugs are approved the vast majority of the time. That’s not because the FDA has low standards; it’s because the drugs that don’t meet FDA’s standards don’t make it to submission. Companies can already predict how FDA will review those.

That’s what makes FDA rejections so interesting. Rejections represent prediction failures: cases where a drug company thought they knew what FDA wanted, only to learn that they were mistaken. There is perhaps no richer vein of regulatory wisdom available to us. It’s worth exploring what happened to these applications and why. Often, the rejected application introduced a novel question that FDA had not previously considered. Future drug developers can benefit from knowing what conclusion the agency reached on those questions.

So it is tragic that, for decades, the FDA treated that knowledge as effectively private. Specifically, the agency had concluded that, if they rejected a drug, and that drug remained unapproved, they would not share the letters they sent to drug companies explaining why their drug was rejected: this information was presumed to be “commercial confidential”. In fact, for decades the agency held that even the existence of a drug application could not be revealed unless the drug was ultimately approved. The result was a strange regime in which some of FDA’s most important thinking about drug development remained locked away.

For decades, academics and FDA experts have been frustrated with this state of affairs and pushed for greater transparency. As far back as 2010, the FDA had explored the idea of releasing information about drug rejections, but ultimately demurred, perhaps concerned that certain pharmaceutical companies would object.

But times have changed. This is the “you can just do things” administration, and they just went ahead and did things. In September 2025, FDA announced that it was releasing a batch of 89 “complete response letters” (CRLs): letters that the FDA sends to notify the company that their drug has not been approved, why, and what they need to do to obtain approval. This wasn’t an entirely new idea: FDA already releases CRLs if and when the drug is finally approved. But this batch included letters for drugs that had not yet been approved – a reversal of the FDA’s prior stance that information about unapproved applications would not be published. The releases included letters dating back to 2024, and they promised to continue to release more. Importantly, FDA made redactions that they believed would protect drug companies’ proprietary information, but left in much of the agency’s own advice to companies, particularly related to their clinical trial designs.

Not everyone was happy with this move. One anonymous company hired a top tier FDA law firm, Covington and Burling, to file a petition to FDA on their behalf to stop the practice. Covington’s petition, published on April 20th, attacks the publication of CRLs on two grounds: First, they argue that the release of CRLs is not legal. Second, they raised several “policy concerns” with the practice; in other words, they argue that it is not just illegal, but unwise–for reasons we’ll get into in a moment.

The legal concerns are genuine: it’s not clear whether FDA is actually authorized to release these letters–even with redactions–and other law firms have already raised similar concerns. FDA’s new policy does seem to be a shift from its earlier stance.

As for their policy concerns: this is where the Covington memo struggled to win me over, and may not win over the FDA either. To start, they expressed concerns that the release of the CRLs might “confuse consumers” or “undermine confidence” in the drugs if they are ultimately approved. To which I can only shrug my shoulders. I’m not convinced we should withhold valuable and true information from the public just because some people might misinterpret it, and I don’t think that argument is going to persuade the administration either. I hope the era of withholding information in the name of public health paternalism is over, at least for a while.

But more significantly, they attacked the core logic behind the FDA’s release of the CRLs. And they did it in a strange way: by agreeing with it.

The strange logic behind the petition

Those who advocate for FDA transparency (and I include myself among that group) believe that both companies and consumers are better served when we understand the rationale behind FDA’s decisions. When FDA’s reasoning is public, it makes drug development more predictable and prevents companies from making avoidable mistakes that could prevent drugs from reaching the public.

FDA’s CRLs: a roadmap to approval?

The FDA embraced that view when it released the CRLs. In their announcement of the release, they wrote that “Publishing CRLs offers important benefits for public health, including empowering drug developers to avoid common missteps.”

So when Covington introduced its citizens’ petition in opposition to the CRL release, you might have expected them to dispute this fact. They did not. In fact, Covington’s petition goes to great lengths, citing specific examples, to show exactly how the CRLs provided genuinely useful information to the public about FDA’s thinking that could be used as a “roadmap” (a word they used several times) to others seeking to develop drugs for serious diseases, allowing them to avoid missteps that could slow down drug development. In Covington’s own words:

“FDA’s conclusions in a CRL provide competitors with a detailed roadmap of how to avoid the cited deficiencies. Such information could allow a competitor to ‘eliminate much of the time and effort that would otherwise be required to bring to market a product.’ For example, if FDA finds flaws with a study design in a CRL, competitors could use that information to avoid these pitfalls in designing their own studies.”

Why are they echoing FDA’s reasoning if their goal is to get FDA to reverse their decision? Because, they argue, FDA’s logic runs counter to the intent of the disclosure rules. In disclosing the CRLs, FDA seeks to increase competition and cut drug development timelines. But the disclosure rules are deliberately designed to prohibit the release of information that might help a company’s competitors. So Covington is using the FDA’s own logic against them: the more useful FDA’s advice is to competitors, the greater the risk of competitive harm to the company from releasing it, and the more serious the legal violation.

There’s a certain logic to it. But there seems to be a flaw in the logic. The reason that CRLs are helpful are not because they reveal the drug company’s private information. It’s because they reveal FDA’s own rationale and thinking. Surely it can’t be against the rules to share the FDA’s own rationale and advice–especially if that advice establishes a regulatory precedent that others may be asked to follow. Right?

But Covington is prepared for that objection. They make two points: First, they note that the current law plausibly provides a very broad definition of what qualifies as “confidential” information that is protected from disclosure. Arguably, the law treats as “confidential” any information that a company “customarily and actually” treats as private or any information that might harm a company’s competitive position. FDA’s own feedback to companies might qualify as confidential under either standard, especially if the information is helpful to others.

Second, and significantly, they argue that FDA’s advice is “inextricably linked to the confidential information submitted by the applicant.” In other words, by revealing FDA’s own advice to the sponsor, they’re revealing information about the company’s application. The advice was provided in the context of the application, and the application’s content is confidential.

Frustratingly, they might be right about that second point - FDA’s advice is indeed closely linked to the applications it receives. FDA’s regulatory policies are not primarily established through policy pronouncements and guidance; they are established one review at a time, as a kind of regulatory “case law”. As Covington notes, this is why, when faced with a hypothetical policy question, FDA typically responds that it is a “review issue”. It is genuinely difficult for FDA to determine in advance how it will address specific questions without seeing the entire package that the drug company has submitted. It is through the accumulation of experience with specific applications that both the FDA and the public are able to reach a broader understanding of FDA’s regulatory standards.

For those of us who are fond of the rule of law, this creates a real problem. As the legal philosopher Jeremy Waldron notes: “law should be epistemically accessible: it should be a body of norms promulgated as public knowledge so that people can study it, internalize it, figure out what it requires of them, and use it as a framework for their plans and expectations and for settling their disputes with others.” But the law also demands that companies’ confidential information not be released—and that confidential information arguably includes the FDA’s own feedback and regulatory reasoning. What should we do if FDA’s “case law” is locked away in private companies’ hands?

Some have argued that the FDA simply needs to find a way to cleave the regulatory policy from the specific circumstances in which it was applied; to derive general principles from specific cases. I am skeptical. This presumes that there is some sort of underlying principle that can be surgically extracted from the FDA’s reviews and presented before the public as freestanding “general advice”. The FDA does try to do this: they extract insights from previous reviews and share them back in the form of guidance to industry. And the guidances are helpful. But inevitably, the extracted insight fails to capture the nuance and detail of the original FDA “case law” that informed it. The guidances are too broad and too divorced from real drug development questions. That’s because, for the FDA, the reviews themselves are the policy.

The Case for Releasing the Rejection Letters

Let’s step away from legal philosophy to take a concrete look at why this matters. Any evaluation of the impact of CRLs should start with the question: how will CRL release affect innovation? Both the petitioner and the FDA seem to agree that the release of CRLs could improve the speed and efficiency of drug development. But Covington suggests it might lead competitors to free-ride on the innovators’ work, reducing the incentive to invest in drug development in the first place. Could that happen? And will that ultimately reduce innovation overall?

I would love it if we had a better answer to this question. For now, there is little empirical research to lean on: the few studies done on this topic are not cleanly applicable to this situation. That said, it would be very surprising to me if more regulatory transparency reduced innovation on net. One thing Covington and the FDA do agree on is that when the FDA provides a “roadmap” for future drug developers, it reduces regulatory uncertainty, making future innovation easier and less costly. And when the FDA, by releasing CRLs, takes away a company’s exclusive access to its regulatory feedback, it doesn’t take away the many other competitive advantages that the company possesses: including patents, data exclusivity, and the benefits of their own investments in drug discovery, manufacturing, and early-phase clinical development. I believe that FDA sought–and achieved–a reasonable balance: the redacted CRLs remove regulatory barriers to drug development while retaining incentives for innovators.

And releasing CRLs doesn’t just help improve innovation in a specific therapeutic area. Broadly speaking, regulatory transparency makes the entire pharmaceutical industry more competitive. Too often, drug companies with more regulatory experience get an advantage solely because they have a larger stockpile of correspondence and feedback from FDA; access to a secret body of case law that other companies lack. Releasing CRLs helps companies compete on a level playing field, armed with a shared understanding of what evidence FDA expects.

Also, drug developers are not the only group we should be concerned about here. Companies’ submissions are built on data collected in clinical trials, and we should consider the interests of patients and volunteers who participate in those trials. Are we making the best possible use of their time and effort? Covington argues that the CRLs provided a “roadmap” to competitors that could allow them to develop drugs “without needing to invest in the trial and error that is a necessary part of the innovation process”. If that trial and error took place solely in a biology lab, that might be reasonable enough. But in most cases, the “trial and error” they cite refers to clinical trials: experiments with human volunteers. If an FDA “roadmap” exists that could prevent failed trials, don’t we owe it to those volunteers to share that roadmap and use it?

In fact, we’ve had this debate before. Many of the kinds of secrets that Covington ostensibly seeks to protect are already revealed on the government’s clinical trials registry, clinicaltrials.gov. When the government created that website, they heard many of the same objections that Covington raises in this petition: that revealing trial designs and results could result in companies facing a competitive disadvantage. In that case, they decided the interests of patients who participate in trials outweighed companies’ interests in protecting their trade secrets.

The disclosure rules make FDA’s job harder

I should mention at this point that the petitioner did not ask the FDA to stop releasing CRLs entirely. Instead they proposed that FDA create a new process in which, before each CRL is released, companies would have the opportunity to propose their own redactions and, if needed, challenge the CRL release in court. That sounds reasonable, but I’d be concerned with how it might play out in practice. That’s because the proposal does not resolve the fundamental tension that the petition laid out: the very information that would be the most useful to innovation—FDA’s advice and regulatory reasoning—is the information that companies will fight hardest to redact.

We shouldn’t assume that Covington’s client will be successful. Despite the objections, the FDA has taken the position that it has the legal authority to release these documents. The FDA may choose to deny the petition and, if litigation follows, they may prevail in court. But the petition still shows how awkwardly our disclosure rules fit the actual structure of drug regulation. FDA sought to make its reasoning more public while protecting trade secrets, but the law makes FDA’s job more difficult than it should be.

In a better regime, it would be easier for FDA to strike a reasonable balance in which it discloses its regulatory reasoning while protecting truly proprietary information. I hope policymakers take this moment as an opportunity to take another look at the disclosure rules. Drug companies should compete by developing better medicines—not by gaining privileged access to what the FDA thinks.