Frances Kelsey, Thalidomide and the 1962 Law That Slowed Every New Drug

There is a rule for telling a hero from a pretext: honor the first, and then check what was done in the hero's name. Frances Kelsey was a hero. The law that Congress passed in her wake did more than she did, and a good deal more than she needed.

Both sentences are true, which in Washington makes them nearly unsayable. The trouble with most coverage of this story is that it will only hold one at a time. Let me hold both.

Terms, and the thesis

Two terms drive this entire story, and the law distinguishes them carefully.

Safety asks whether a drug is likely to harm the person who takes it, at the dose and for the use intended. Efficacy asks whether the drug works: whether it does what the label says it does. A drug can be safe and useless. A drug can be useful and dangerous. Regulators can ask about either.

My thesis has four parts, and I will number them so that you can hold me to them.

1. Kelsey's refusal to approve thalidomide in 1960 and 1961 was a model of what a regulator should do: she insisted on evidence before a risk was taken on the public's behalf.
2. The 1938 law she was enforcing already gave her the power to do it. The need for her to be a hero was the sign that the old system worked.
3. The Kefauver-Harris Amendments of 1962 added something new, a federal requirement that a drug maker prove efficacy before marketing, and gave the agency a veto in place of a clock.
4. That change has real costs, in time, money and lost options, and a free people is entitled to weigh them even while honoring the woman whose caution inspired the law.

The law before 1962: the 1938 Act

Start with the baseline, because it is the part the legend forgets. The Federal Food, Drug, and Cosmetic Act of 1938 was the product of a catastrophe too. In 1937 a liquid form of a sulfa drug, Elixir Sulfanilamide, was sold with a solvent that was a poison, and more than a hundred people died, many of them children. Congress responded with the 1938 Act, which for the first time required a manufacturer to show that a new drug was safe before selling it, and to submit an application to the Food and Drug Administration to do so.

Here is the feature that matters for our story. Under the 1938 Act, the application would take effect on its own after a set period, sixty days, unless the FDA objected. The agency could extend the clock by saying the application was incomplete, but the default was approval by the passage of time.

A bureaucracy that approves things by forgetting about them is, I grant, an unusual theory of government. Washington has since improved on it: today it delays things by forgetting about them. That is not a perfect system. It put the burden on the agency to find a problem. But it was a safety regime, and it had teeth. It was the law Frances Kelsey picked up when she walked into the FDA.

Kelsey and the Kevadon application

Frances Oldham Kelsey joined the FDA in 1960 as a medical officer. One of the first applications to land on her desk was for a sedative that the company, the William S. Merrell Company of Cincinnati, proposed to market under the name Kevadon. The active ingredient was thalidomide. It had already been sold in a number of other countries, and the company expected an easy approval.

She did not give it one. She found the application's evidence thin and the claims of safety too lightly supported. She asked for more data. Merrell pressed her to approve, and the sixty-day clock was the lever. Kelsey used the other end of the same lever: when the clock was about to run out, she declared the application incomplete and asked for more information. Her particular concern was reports from abroad that the drug seemed to cause a nerve condition called peripheral neuritis in people who took it. She was also not satisfied that the company had shown the drug was safe for use in pregnancy.

Then came the news from Europe. In late 1961 doctors in Germany and elsewhere connected the drug to a wave of severe birth defects in children whose mothers had taken it during pregnancy. The German manufacturer withdrew the drug from sale at the end of 1961. The Kevadon application, by then, was still unapproved in the United States.

That is the heart of the story. Thousands of children abroad were harmed, and the American record, while not spotless because some of the drug was distributed to doctors for investigational use, was far better than it would otherwise have been.

The part the legend gets right

A newspaper story in July 1962, by Morton Mintz in the Washington Post, put Kelsey's name before the public. On August 7, 1962, President Kennedy presented her with the President's Award for Distinguished Federal Civilian Service. It was deserved. Here was a government employee who did her job under pressure from a company with a lot of money at stake and a sixty-day deadline hanging over her, and who did not blink. If you want a picture of what public service ought to look like, you could do worse.

I want to say that plainly, because some critics of the modern FDA are uncomfortable doing it. Kelsey's story is not an embarrassment to those who want a leaner regulator. It is evidence for the argument that safety review, done well and with the evidence on the table, is one of the legitimate functions of the federal government's police power over interstate commerce in drugs.

The Kefauver-Harris Amendments

Now the second half. Senator Estes Kefauver of Tennessee had been investigating the drug industry for years, and his original bill was focused largely on pricing and marketing practices. It was not going anywhere. The thalidomide news changed the politics overnight. Congress moved, with the support of the White House, and the amendments were signed on October 10, 1962.

What did they do? Here is a short list, set out in the form of the law's actual changes.

• Proof of efficacy. A new drug application now had to include "substantial evidence" of effectiveness, drawn from "adequate and well-controlled investigations." For the first time, the federal government would decide not only whether a drug was reasonably safe, but whether it worked.
• A veto instead of a clock. The automatic sixty-day approval was replaced with a requirement that the FDA affirmatively approve a drug before it could be marketed.
• Informed consent. Investigators had to obtain consent from people who received investigational drugs, with limited exceptions.
• Manufacturing and reporting. Manufacturers had to follow good manufacturing practices and report adverse events, and prescription drug advertising came under FDA oversight.
• A review of the past. The FDA was directed to review the efficacy of drugs that had been approved on safety alone before 1962, a project that became known as the DESI review.

Some of that is plainly sensible. Informed consent for experimental subjects is a matter of basic dignity. Adverse-event reporting is the kind of information a safety regulator needs. Even an end to the automatic clock is defensible: a rule that approves a drug because an agency was too busy to answer is a rule built to fail.

Congress, you will notice, responded to a drug that was too dangerous by creating a rule about drugs that were too ineffective. It is like answering a house fire by passing a law about wallpaper. The efficacy requirement is the one worth arguing about, and notice something about it. Thalidomide was not a failure of efficacy review. It was a failure of safety review in other countries, and in the United States it was stopped by the old safety standard, enforced by one careful official. The efficacy amendment answered a question the thalidomide tragedy had not asked.

The steelman

The best case for the 1962 law deserves to be stated in the words its defenders would use.

They would say that a drug that does not work is itself a harm. It costs money. It wastes time during which a real treatment might have been used. It can expose a patient to side effects for no benefit. And a patient, a family and even a doctor cannot be expected to evaluate the evidence for a medicine's effectiveness; that is exactly the kind of expertise a public agency can supply. Before 1962, they would add, the market was full of drugs sold on thin claims, and the post-1962 requirement of controlled studies raised the quality of medical knowledge for everyone. Randomized, controlled trials became the standard, and the standard improved the science.

That is a serious argument, and I do not believe it should be waved off. Controlled trials are one of the real achievements of modern medicine. Nobody should want to return to the world of testimonials.

The answer, in four points

But the steelman proves less than it needs to, for these reasons.

1. It treats the mandate and the method as the same thing. The defenders are right that controlled trials are valuable. It does not follow that the federal government must require them as a condition of sale for every drug, at the agency's standard, before any patient may buy it. Doctors, hospital systems, insurers and professional societies all have reasons to demand good evidence and the means to do it. The question is who gets the veto.
2. The cost of delay is real, and invisible. A drug that is approved too early produces victims with names and faces. A drug that is approved too late produces victims who are statistics: people who died or suffered while a treatment sat in a review queue. Both are real. Only one makes the evening news. A rule that is built around avoiding the visible error will, over time, produce more of the invisible one.
3. The record suggests the cost was not small. Economists studied the consequences. Sam Peltzman, in a well-known paper in the Journal of Political Economy in 1973, argued that the 1962 amendments sharply reduced the introduction of new drugs without a compensating gain in the quality of the ones that came to market. Other researchers in that period, including William Wardell and Louis Lasagna, described a "drug lag" in which new medicines became available in Britain and elsewhere years before they reached American patients. Those studies have been criticized, and the lag has narrowed in later decades. But the debate they began has never been settled in the critics' disfavor.
4. The system hardened into a moat. The cost of clinical trials is now so high that only large firms can afford to run them. That favors incumbents. It is the same pattern a student of regulation will recognize in other fields, where compliance costs fall hardest on the smallest and newest entrants, a point made in big companies love the regulations their small rivals cannot afford.

I should be fair about what I am not claiming. I am not saying that the FDA should be abolished. I am not saying the efficacy standard should never have existed. I am saying that Congress, in the emotional aftermath of a safety scare, expanded the agency's power over a question that was not at issue, and that no one in the room was assigned to count what that would cost.

The history of the thalidomide drug itself

There is a coda that illustrates the complexity. Thalidomide was never approved in the United States in the 1960s. But the drug did not disappear from medicine. In 1998 the FDA approved it for the treatment of a complication of leprosy, with strict controls on its use designed to prevent exposure in pregnancy, and in 2006 for a type of blood cancer. The lesson is not that thalidomide was safe. It is that a drug can be dangerous in one setting and valuable in another, and that the right answer depends on the patient, the use and the controls. That is a case for risk management by informed people, which the modern system does attempt, rather than for blanket prohibition or blanket permission.

Measure it against the record

A conservative does not have to choose between Kelsey and skepticism. The Constitution gives Congress the power to regulate commerce among the states, and a safety standard for drugs shipped in interstate commerce sits comfortably within it. That is an old and defensible use of the power. What a conservative should insist on is that each extension of a power is justified by its own evidence, not carried along on the moral authority of the last one.

James Madison wrote in Federalist No. 51 that if men were angels, no government would be necessary. He did not say that if regulators were angels, no limits on regulators would be necessary. The point of the separated structure is to assume that even good people in government will sometimes carry a good principle further than the facts justify. Her own power, under the 1938 law, was enough.

If you want to see how the modern process works, how long the trip from laboratory to pharmacy takes and why, read How a New Drug Gets Through the FDA and Why the Trip Takes So Long. If you want the patient's side of the same question, read The Right to Try Law of 2018 and the Families Who Pushed It Through. For the broader argument that regulation tends to protect the regulated, see the regulation bill a ten-person company pays that a giant can shrug off.

What a better balance looks like

A reform agenda that honors Kelsey would not repeal safety review. It would ask a few disciplined questions.

• Should there be a faster, conditional pathway for drugs aimed at fatal diseases with no treatment, with a requirement of continued data collection after release?
• Should foreign approvals from regulators with comparable standards be given weight, rather than starting from zero?
• Should the agency be required to publish, for each major rule, an estimate of the time and the patients' cost of delay, the way it publishes the safety rationale?
• Should Congress, rather than the agency, decide how much evidence is enough for each class of drug, so that voters can hold someone accountable?

Several of these have been tried in some form, and the details matter. The principle that unites them is that a safety agency should be judged by both kinds of error, the drug it let through and the drug it held back, and that the second kind should be counted in public.

The case in brief

1. Frances Kelsey's refusal to approve thalidomide in 1960 and 1961 was an act of exemplary public service, and she earned the 1962 President's Award.
2. She did it under the 1938 Food, Drug, and Cosmetic Act, which already required proof of safety. The system worked because one official insisted on evidence.
3. The Kefauver-Harris Amendments of October 10, 1962 went further than safety: they required proof of efficacy and replaced the sixty-day clock with an affirmative approval.
4. Economists and clinicians have argued since the 1970s that this slowed the arrival of new drugs, and a free people should weigh that cost alongside the benefit.
5. The better rule judges a regulator by both errors, the harm it permitted and the help it delayed.

What you can do: look up the most recent FDA report on drug approvals on the agency's website, and ask your representative how long the average review takes and what Congress has done to measure the cost of delay.