How a New Drug Gets Through the FDA and Why the Trip Takes So Long

Every approved drug has an origin story that begins with a long wait. Somewhere a researcher has a promising compound, and between that compound and a pharmacy shelf stands a stack of paper roughly the size of a small office building, a series of human trials that can take years, and a federal agency whose employees will be blamed if they say yes too soon and thanked by no one if they say yes quickly.

Now, I do not write this to mock the Food and Drug Administration. Its job is hard, and the rules it enforces were written for good reasons, often in the wake of real disasters. I write it because there is an accounting question in this story that almost nobody asks: the rules count the harms of approving too soon, and they do not count the harms of approving too late. Let me explain the whole path first, one step at a time, and then the accounting.

I will make no claim in this post about whether any particular drug works. That is for patients, doctors and the scientific record. My subject is the process and its incentives.

Terms

The FDA is an agency within the Department of Health and Human Services. Its authority over drugs comes from the Federal Food, Drug, and Cosmetic Act, first passed in 1938, and from later amendments.

A sponsor is the company or institution that wants to bring a drug to market.

An investigational new drug, or IND, is a drug that has not yet been approved and that is being tested in humans.

A biologic is a medicine made from living sources, such as vaccines, antibodies and many modern cancer and autoimmune treatments. Biologics are licensed under the Public Health Service Act. Most other drugs, the kind made through chemical synthesis, are approved under the Federal Food, Drug, and Cosmetic Act.

The history that built the rules

The path has been built in layers, each one laid down after a failure.

In 1937 a company marketed a liquid form of a sulfa drug dissolved in a toxic solvent, and more than a hundred people died, many of them children. The result was the 1938 Act, which required that new drugs be shown safe before marketing.

In 1962, after the thalidomide tragedy abroad, Congress passed the Kefauver-Harris Amendments, which added a requirement that drugs be shown not only safe but effective, through adequate and well-controlled studies. The story of the FDA reviewer whose caution kept thalidomide off the American market, and of the law that followed, is told in Frances Kelsey, Thalidomide and the 1962 Law That Slowed Every New Drug. The 1962 amendments are the origin of the modern trial system.

I do not argue that those laws were mistakes. A rule that asks a drug company to prove that its product is safe and effective before selling it to the public is a reasonable rule. The question is the height of the hurdle and the length of the course.

Step one: preclinical work

Before a drug is tested in a person, it is tested in a laboratory. Researchers study how the compound behaves in cells and in animals, how it is absorbed and broken down, and what doses are toxic. This stage can take a year or two, sometimes longer, and it is where the great majority of ideas are abandoned. A compound must show enough promise, and enough safety in animals, to justify the next step.

Step two: the investigational new drug application

The sponsor then files an IND application with the FDA. It describes the compound, the manufacturing, the animal data and the plan for human trials. Under the regulations, the sponsor may begin human trials thirty days after the FDA receives the IND, unless the FDA places a clinical hold and tells the sponsor to wait. The FDA does not "approve" the IND in the way it approves a drug; it permits the trial to proceed or stops it.

An institutional review board, an independent committee at each trial site, must also approve the plan and protect the participants.

Step three: three phases of human trials

Human testing in the standard model proceeds in three phases.

Phase 1. A small group of volunteers, often healthy people but sometimes patients with the target disease, typically a few dozen. The purpose is to learn how the drug is processed in the body, what doses are tolerated and what side effects appear. It is about safety, not effect.

Phase 2. A larger group, up to several hundred patients who have the condition. The purpose is to look for evidence that the drug works and to find the right dose, while continuing to watch safety. Many drugs fail here.

Phase 3. The large trials, often several hundred to several thousand patients, usually randomized and compared with a placebo or with the standard treatment. This is the stage designed to give the evidence of effectiveness that the 1962 law requires, and to find less common side effects. These trials are expensive and can take years.

The FDA often meets with the sponsor between phases to agree on what the next trial must show. Here is a point worth marking: the agency is both the adviser and the judge. It tells sponsors what evidence it will want, and later decides whether the evidence suffices. That is efficient in some ways and it concentrates a great deal of power.

Step four: the marketing application

After the phase 3 results are in, the sponsor submits a new drug application, or NDA, under section 505 of the Federal Food, Drug, and Cosmetic Act, or, for a biologic, a biologics license application, or BLA, under section 351 of the Public Health Service Act. These applications are enormous, often containing enormous volumes of data.

The FDA first decides whether the application is complete enough to review. If it is, a team of physicians, statisticians, chemists, pharmacologists and others reviews it. The agency may also ask an advisory committee of outside experts to weigh in. The committee's vote is advisory; the agency decides. The agency may approve the drug, or issue a complete response letter, which says what is lacking, and the sponsor can respond and resubmit.

Approval does not end the process. The FDA can require postmarketing studies, sometimes called phase 4, and the agency monitors adverse event reports after the drug is on the market.

The clock: PDUFA

How long does the review take? Here a statute matters.

By the late 1980s, review times at the FDA had grown long, and the agency complained that it lacked the staff to review applications faster. In 1992 Congress passed the Prescription Drug User Fee Act, known as PDUFA, signed by President George H. W. Bush on October 29, 1992. The deal was this: drug makers would pay fees with each application, and those fees would fund additional reviewers, and in return the FDA would commit to performance goals on how fast it would act.

Under PDUFA the agency sets goals for review. In round terms, a standard application is to be reviewed within ten months of the filing date, and a priority application, for a drug that would be a significant improvement in treatment, within six months. For the newest kinds of drugs the clock starts after a filing period of sixty days, so the full timeline from submission is longer than the headline figure. PDUFA is reauthorized by Congress every five years, and each reauthorization renegotiates the goals.

There is much to say for PDUFA. Review times did shorten after it passed, and the arrangement brought in money the agency needed. There is also a fair criticism that user fees tie the agency's budget to the industry it regulates, and some critics worry that this creates a dependency. I do not think that is a reason to repeal it, but it is a reason to read the reauthorizations carefully. If you want to watch your Congress exercise real oversight of a regulatory agency, the PDUFA reauthorization is where it happens every five years.

Ways to move faster

Congress and the FDA have created several paths for drugs that meet an urgent need.

• Fast track. Created in 1997, it allows more frequent meetings and a rolling submission, so that the agency can review parts of the application as they are completed.
• Priority review. Created in the 1992 act, it sets the shorter six-month review clock for drugs offering a significant improvement.
• Accelerated approval. Established by FDA regulation in 1992 and later written into statute, it allows the agency to approve a drug for a serious condition on the basis of a surrogate endpoint, a measure such as a lab value or a tumor response that is reasonably likely to predict a clinical benefit, rather than waiting for the final clinical outcome. The sponsor must then conduct confirmatory trials. Congress tightened the rules for confirmatory trials in 2022.
• Breakthrough therapy designation. Created by the Food and Drug Administration Safety and Innovation Act of 2012, it provides intensive guidance for drugs that show substantial improvement over available therapy on a clinically significant endpoint in early evidence.

I list them for a reason. The existence of these pathways is an admission. Each was created because the standard process was too slow for patients who could not wait, and each carries rules and conditions that make it harder to use. They are the exception that proves how long the rule takes.

The accounting question

Now to the point I promised.

When the FDA approves a drug that later turns out to cause harm, there is a victim, a name, a news story and a congressional hearing. When the FDA delays a drug that would have helped, there is also a victim, but she is a statistic, and she is rarely in the room. Her loss is real and invisible. The agency's incentives follow from that asymmetry. An official who approves a drug that later fails is blamed. An official who delays a drug that would have succeeded is not.

The economist Sam Peltzman studied exactly this. In 1973 he published a paper in the Journal of Political Economy, "An Evaluation of Consumer Protection Legislation: The 1962 Drug Amendments." He compared the introduction of new drugs before and after the 1962 law and estimated that the amendments had reduced the number of new drugs introduced by roughly 60 percent, from about 41 a year to about 16, and that he could find no offsetting benefit large enough to match the loss. His conclusion was that the cost of the lost drugs, which he called a cost of delay and absence, was larger than the benefit of keeping ineffective drugs off the market.

I will be careful. Peltzman's paper is an argument, written fifty years ago and about an earlier era. Later scholars have disputed parts of it, and the system has changed a great deal since 1973, in part because of PDUFA and the expedited pathways I described. I do not offer his numbers as a current measure. I offer his framework, which is right: a regulator should count both kinds of error. A drug approved too early can hurt people. A drug approved too late can hurt people too, and the second kind of harm is the one the process hides.

This is what economists call the difference between a Type I error and a Type II error. In a courtroom, a Type I error is convicting an innocent man, and a Type II is acquitting a guilty one. In the FDA, one error is approving a bad drug; the other is rejecting or delaying a good one. A system that guards only against the first will make the second constantly, and the second will never make the news.

What I would argue for, and what I would not

Let me be plain about what this argument does and does not support.

It does not support abolishing the FDA, or letting sellers market untested drugs. The Founders understood that a free market needs rules against fraud, and a patient cannot inspect a molecule. The case for a federal gatekeeper on safety is a serious one. I accept it.

It does support five specific reforms that conservatives of different kinds could agree on:

1. Count the cost of delay. Require the agency to publish, for each approval, how long the review took and how the time compares with its stated goals. What gets measured gets managed.
2. Lean on the pathways that work. Expand and simplify expedited approval, and make the confirmatory trial requirements firm, so that earlier access does not become permanent uncertainty.
3. Honor the access laws. The Right to Try law of 2018 gave terminally ill patients a legal path to experimental drugs that have completed phase 1; see The Right to Try Law of 2018 and the Families Who Pushed It Through.
4. Keep the user-fee bargain honest. Congress should use each PDUFA reauthorization to demand faster performance and public data, not just to renew the fees.
5. Make Congress, not the agency, answer for the big choices. An agency that writes guidance with the force of law raises the questions about administrative power that appear throughout this blog; see The Administrative Procedure Act of 1946 Was a Truce, and It Still Shows.

Speed alone is not the goal. A faster process that approves unsafe drugs would be a failure by the same logic. The goal is a process that is as fast as the evidence allows and honest about what it costs when it is slower.

The economics of the other side of the pipeline, why a lower expected return means fewer drugs started, is in Price Controls on Medicines Today Mean Fewer New Medicines Tomorrow. The two arguments fit together: the longer and more uncertain the road, the more it costs to walk, and the fewer who start.

A family's view

Picture a hypothetical family whose daughter has a rare condition. A promising drug has completed phase 2 and shows some signal, and the next trial is planned for two years from now. The family is not asking for a guarantee. They are asking whether a process built to protect them can also count the years that they are waiting. The honest answer from the system is: we have rules that let some patients in, under conditions, and we have a clock that runs on its own schedule. That answer is better than it was in 1962, and worse than it needs to be.

For a sense of what else people believe about the system and how it pays, see Six Things People Believe About Health Insurance That Are Simply Not So.

The case in brief

1. A drug passes through preclinical testing, an IND application, three phases of human trials and a new drug or biologics license application, and each step was added in response to a real failure, from 1937 to the thalidomide era.
2. The Prescription Drug User Fee Act of 1992 tied industry fees to review-time goals, and reauthorization every five years is Congress's chance to demand faster, more transparent review.
3. Fast track, priority review, accelerated approval and breakthrough designation exist because the standard path is too slow for urgent cases, and each has conditions.
4. Peltzman's 1973 paper on the 1962 amendments gives the right framework: count the harm of approving too late as well as too early, and publish the data to do it.

What you can do: read the FDA's own plain-language page on the drug development process, then ask your representative or senator what they will demand in the next PDUFA reauthorization, using the congressional lookup.