The Quiet Inversion
The FDA announced this week it will make single clinical trials the default standard for approving new drugs, reversing a 60-year practice that required pharmaceutical companies to prove their products worked twice, according to a New England Journal of Medicine article by FDA Commissioner Dr. Marty Makary and Dr. Vinay Prasad published this week. The shift redefines what counts as adequate proof in drug regulation.
This isn't an expansion of existing exceptions for rare diseases or emergency treatments. The agency is inverting its burden of proof: where two trials were once required unless circumstances justified otherwise, one trial now suffices unless the FDA decides more evidence is needed. Makary and Prasad frame the change as reflecting "modern advances in drug research" that make the old standard obsolete, according to their NEJM article.
The policy follows a pattern emerging across federal health and environmental agencies. In recent months, the EPA revoked its climate endangerment finding and the NIH narrowed its clinical trial registry requirements, as reported by STAT. Each change redefines the threshold of evidence itself rather than simply adjusting how existing standards apply.
Why Two Trials Existed
The two-study requirement traces to the early 1960s, when Congress passed legislation requiring FDA review of data from "adequate and well-controlled investigations," according to the agency's historical guidance documents. The plural mattered. For decades, the FDA interpreted this as meaning at least two studies, preferably with large patient populations and significant follow-up time.
The logic was statistical, not bureaucratic. A single trial can produce promising results through chance, flawed design, or publication bias. The second trial exists to confirm the first wasn't a statistical artifact. If a drug shows a 30% improvement in one study but fails to replicate that result in a second, the initial finding was likely noise.
This replication principle underpins scientific validation across fields. It's why physics experiments get repeated in different labs, why psychology has faced a replication crisis, and why the FDA historically demanded confirmation before allowing drugs into the market. The second trial doesn't just add more data; it tests whether the effect is real.
Dr. Janet Woodcock, who led the FDA's drug center for more than 20 years before retiring in 2024, told STAT the change makes sense. She noted the agency has already moved toward relying on one trial, combined with supporting evidence, for various life-threatening diseases including cancer. But that shift occurred within a framework where two trials remained the standard and exceptions required justification.
Many drugmakers already submit just one pivotal clinical trial for approval, according to FDA data. The question is what changes when the exception becomes the rule, and who bears the consequences when drugs approved on single trials fail to deliver their promised benefits.
What's at Stake for Patients
The practical impact falls on specific patient populations. Consider cancer patients, who already face drugs approved on accelerated pathways: according to a 2021 JAMA study, of 93 cancer drugs granted accelerated approval between 1992 and 2017, only 19 showed improved survival in confirmatory trials. The remaining patients received treatments that never proved the survival benefits suggested by initial studies. Under the new default, this pattern could extend across all therapeutic areas.
Patients with rare diseases will get faster access to experimental drugs, a genuine benefit when no alternatives exist. But they'll also become the confirmatory trial, testing in real-world use whether the single pivotal study's results hold up. When drugs approved on one trial fail to deliver, these patients bear the cost: time spent on ineffective treatments, exposure to side effects without corresponding benefits, and in some cases, financial burden from expensive medications that don't work as advertised.
The FDA's own data shows the stakes. According to agency records cited by STAT, approximately 30% of drugs approved between 2001 and 2010 carried black box warnings or were withdrawn due to safety concerns identified after approval. The two-trial standard caught some of these problems before drugs reached patients. Making one trial the default shifts more of this safety screening to post-market surveillance, meaning patients become the test population for problems the old system might have identified earlier.
The Statistical Power Argument
Makary's defense of the policy rests on a technical claim about statistical power. "You can achieve the same statistical power with one trial as you would with two trials when it's designed and controlled appropriately," he told STAT. The agency will still require two trials in some cases, but the default is now one.
Statistical power measures whether a study is large enough to detect a real effect if one exists. A single trial with 2,000 patients can have more statistical power than two trials with 500 patients each. Makary's argument is mathematically correct but sidesteps what the second trial actually does.
Power and replication serve different functions. A well-powered study reduces the chance of missing a real effect. A replication study confirms the effect you found wasn't a false positive. The first addresses sample size; the second addresses reproducibility. Conflating them treats the problem of statistical noise as solved by adding more participants to one trial rather than confirming results hold across different populations, sites, and conditions.
Who Bears the Risk
The practical effect of this shift depends on how the FDA exercises its discretion about when to require a second trial. If the agency demands additional studies for drugs with marginal benefits or serious safety signals, the policy might function as Woodcock describes: a more flexible version of existing practice.
But the framing suggests a different trajectory. Makary and Prasad predict the change will produce "a surge in drug development," according to their NEJM article. The FDA is simultaneously offering one-month reviews for medications serving "national interests," as reported by STAT. Speed is the explicit goal, with the agency retaining discretion about what qualifies as requiring extra scrutiny.
This reverses the historical burden of proof. Under the old standard, pharmaceutical companies had to justify why one trial should suffice. Under the new default, the FDA must justify why two trials are necessary. That shift changes the negotiation between regulators and industry, particularly for drugs treating conditions where patient desperation creates pressure to approve treatments quickly.
The Pattern Across Agencies
The FDA's approach contrasts with its handling of other products. The agency maintains a "more restrictive approach to other products, including vaccines," according to Makary and Prasad's article, creating an inconsistency in how different categories of medical interventions face regulatory scrutiny. A vaccine given to healthy people to prevent disease faces higher evidentiary bars than a drug given to sick patients to treat existing conditions.
That distinction has logic: the risk-benefit calculation differs when treating illness versus preventing it in healthy populations. But the contrast reveals how regulatory philosophy varies based on product category and political pressure rather than applying consistent principles about what constitutes adequate proof.
The broader pattern suggests agencies are redefining scientific rigor as regulatory obstacle. The EPA's climate decision, the NIH's trial registry changes, and now the FDA's approval standard all share a common thread: they don't dispute the underlying science so much as adjust what level of certainty justifies regulatory action.
What Replication Catches
The history of drug approvals offers examples of what second trials reveal. Some medications show dramatic benefits in initial studies that diminish or disappear when tested again in different populations or settings. Others produce safety signals that only become apparent with larger or longer trials. The replication requirement exists because the first study's conditions, patient selection, trial site expertise, outcome measurement, don't always generalize.
Making one trial the default assumes modern trial design has solved these problems. But "designed and controlled appropriately" is precisely what's contested. Industry-sponsored trials face documented pressures around endpoint selection, patient enrollment, and publication of unfavorable results, according to research published in medical journals. The second trial, often conducted by different researchers in different settings, provides a check on those pressures.
Without that check, the FDA's post-market surveillance becomes more critical. The agency will need to catch problems after approval that the old system identified before drugs reached patients. That shifts risk from pharmaceutical companies, who must invest in confirmatory trials, to patients who receive medications approved on thinner evidence.
Makary frames the change as eliminating outdated requirements that slow innovation. But the two-trial standard wasn't arbitrary bureaucracy; it was a response to drugs that looked promising once and failed to deliver when tested again. The new default bets that statistical power can replace replication as a safeguard against false positives.
Whether that bet pays off depends on how the FDA uses its discretion to require second trials when one isn't enough. The policy makes that judgment call after companies have invested in a single study rather than before, changing the incentive structure around confirmatory research. The surge in drug development Makary predicts may come at the cost of certainty about whether those drugs actually work as well as their first trial suggested, a cost that will be measured in patient outcomes, not regulatory efficiency.