FDA drug approvals rest on fewer trials, and the cracks are showing

The bargain at the heart of drug approval has always been plain: prove it works, then prove it again. Regulatory frameworks in the US and Europe typically require at least two well-controlled studies before a product's effectiveness counts as validated — a safeguard against results that look convincing but are really chance. It is an unglamorous rule. It is also the rule this week's health reporting suggests is quietly fraying.

A retrospective review of every novel drug the FDA approved from 2016 through 2024, built from FDA public records and the ClinicalTrials.gov database to track how many completed trials stand behind each approval and who paid for them, found that novel approvals in the US have been supported by a decreasing number of studies since 2016. The trend, as Neurology Advisor summarized it, points to a system that brings new medications to market by relying on fewer clinical trials and more industry funding.

Hold those two threads together: fewer studies per drug, and more of the money behind those studies coming from the companies selling the product. Neither fact is a scandal on its own. Together they describe an evidence base growing thinner per approval at the very moment its funding grows less independent. The rest of the week's health news reads almost like a stress test of what that means.

When the floor is one study

Start with the starkest case. The Medicines and Healthcare products Regulatory Agency has suspended the new use of the drug avacopan after concluding that the pivotal trial data behind it is no longer reliable. One pivotal clinical study underpinned the license authorization, and that study was found to have compromised integrity and reliability.

The detail worth sitting with is what gets lost. Patients and providers lose access to a treatment that, as Medscape's report put it, showed real-world efficacy but can no longer be supported by regulatory benefit-risk standards. The drug did not stop working. The evidence did.

If a single pivotal study is the floor beneath a medicine, that floor is only as good as the study beneath it. The FDA review says the floors are getting thinner. The avacopan suspension shows what a collapse costs — and who pays: the patients who had a working treatment, and now do not.

Borders going up around the evidence

The evidence behind medicine is being reshaped from a second direction: geopolitics.

Contagion Live's editor-in-chief, Jason Gallagher, highlighted a proposed Office of Management and Budget rule that would restrict federal grants for international research collaborations. The rule takes aim at the cross-border networks infectious disease practitioners use to track new pathogens, resistance mechanisms and outbreaks. The restrictions, the piece warns, threaten the surveillance and knowledge-sharing systems essential for managing diseases that do not respect political or geographic borders.

Those networks are not an abstraction. East County Magazine's health and science highlights for reports ending September 4 included a cyclosporiasis outbreak and a measles surge that has pushed infections to a 30-year high, with two children dead. Outbreak surveillance is exactly the work a rule like this would make harder.

Then the counter-movement, because this week cut both ways. On September 3, the National Institutes of Health ended its ban on funding new research projects in South Africa. NIH Director Jay Bhattacharya issued an internal memo requiring that international projects have a clear scientific rationale and the potential to improve the health of Americans. The decision restores access to one of the world's strongest clinical trial infrastructures for tuberculosis, HIV and other infectious diseases. South African Medical Research Council President Ntobeko Ntusi welcomed the move: "No country can solve global health challenges alone."

Note the shape of the reopening, though. The ban is gone; the test remains. International research must now justify itself in terms of American health. That is not a border being erased — it is a border redrawn with a checkpoint, collaboration routed through a gate marked "clear scientific rationale." Administrators can draw lines around a borderless specialty. Pathogens have never carried passports.

The evidence arrives late

A thinning approval file rarely means no evidence. More often it means the evidence arrives late — after the prescriptions.

The week's clearest example came in Stat's Morning Rounds roundup, which covered a study finding that Paxlovid did not help long Covid. For patients managing chronic post-viral symptoms, that is new clinical data in a space where hope has been doing the work of proof. The same roundup flagged fresh research on how cannabis edibles affect driving.

When the evidence does not settle matters at approval, the filtering moves downstream — to guideline writers and insurers. Updated migraine prevention guidelines were released Monday, and the reason, per Stat, is that a "wealth of new treatment" has emerged faster than clinical standards could absorb it. Physicians will use the framework to navigate the newest prevention medications and, tellingly, to secure insurance prior authorizations — justifying advanced drugs to payers. Guideline committees and insurers become the backstop, doing work that a deeper approval file would have settled.

Meanwhile, money pressure runs through the same system from the other direction. The Morning Rounds entry noted that Donald Trump has introduced new deals on drug pricing as part of his health policy updates, developments with direct consequences for what patients pay. And East County's column flagged rising costs forcing large employers to drop healthcare coverage for their workers. Squeezed from both ends, the system is being asked to prove more with less evidence while costing less — a combination nobody would design on purpose.

Rewiring the plumbing

The quietest story in the week's pile may be the most consequential: government, industry and public health sectors are collaborating to standardize clinical research data — common standards for how it is collected and shared, so that safety and efficacy results stay transparent instead of remaining hidden in fragmented data. As Applied Clinical Trials explains, standardized data would streamline validation, letting regulators more efficiently confirm that clinical results are meaningful and not due to chance.

This is plumbing, not headlines. But plumbing decides whether a second study — that old two-study safeguard — can even be assembled from the world's scattered trial data, and whether the results regulators see add up to something real.

The week's through-line: the evidence behind medicine is thinner per drug, more narrowly funded and more politically governed than it once was. The repairs attempted this week — common data standards, a reopened research channel in South Africa, updated migraine guidelines — are real, but modest against that drift. The two-study rule of thumb exists because a single striking result is the easiest thing in science to be wrong about. An approval that leans on less is not automatically a mistake. It is a bet. And this week showed who is left holding it when the bet fails.