H013

Is it safer for the FDA to wait for one more study?

A drug regulator can make two kinds of mistakes: approve a drug that turns out to be harmful, or hold back a drug that would have saved lives. Both cost lives. Only one leaves victims with names.

The drug turns out to be harmful
The drug turns out to be helpful
The FDA approves it
Mistake · seen
Vioxx, 1999–2004
88,000–140,000
estimated extra serious heart attacks in the U.S.; about 44% fatal
Recall, hearings, lawsuits, named victims
Right call
Beta-blockers after a heart attack
26% fewer deaths
in the 1981 trial that proved they work
Now standard care
The FDA waits
Right call · seen
Thalidomide, 1960–62
10,000+
children born with birth defects in other countries; 17 in the U.S., from test samples
The reviewer who held it back got a presidential award
Mistake · unseen
Beta-blockers, years later than Europe
~10,000 a year
U.S. heart-attack deaths one critic estimated an earlier approval could have prevented
No names, no hearings, no lawsuits

Both mistakes cost lives, but only one is seen. When an approved drug causes harm, the victims are real people with names, and there are recalls, hearings, and lawsuits. When a helpful drug is delayed, the people who would have lived are only a statistic. No one, including their families, ever knows who they were. So a regulator is blamed for the first kind of mistake and almost never for the second, and the pressure leans toward waiting.

That doesn't make waiting wrong: thalidomide shows how much a careful “not yet” can save. It means a fair judgment counts both kinds of mistakes. Any choice to act or to wait has costs on both sides, and the ones that leave no names are the easiest to miss.

Try it somewhere else

After one bus accident, a school cancels every field trip for the year.

What costs of that decision would never make the news?

How solid are these numbers?

The seen side is easier to count. The Vioxx estimate comes from an FDA scientist's study of real patients who took the drug. The thalidomide count comes from children who were actually born with injuries.

The unseen side can only be estimated. The beta-blocker figure is a 1979 estimate by pharmacologist William Wardell, a well-known critic of drug delays, based on how the drugs performed in Europe. Other researchers have produced larger and smaller numbers, and many of the biggest estimates come from people arguing for faster approval. Even Wardell's example is complicated: practolol, one of the beta-blockers he pointed to, was later found to cause a severe eye and skin reaction. That difficulty is part of the lesson: an invisible harm is hard to see, and also hard to measure.

Is this an argument against the FDA?

No. It's an argument for counting both columns. Some drugs deserve more study, and some patients can't afford to wait. A drug for a deadly disease with no other treatment carries a very different cost of waiting than a new acne cream. Since the 1990s the FDA has added faster paths for drugs that treat serious diseases, partly in response to AIDS activists and other critics of drug delays. People still disagree about where the balance should be.

Related exhibits

Sources: Vioxx: D. J. Graham et al., The Lancet (2005), estimating 88,000–140,000 excess cases of serious coronary heart disease in the U.S. during the drug's 1999–2004 market life, 44% of them fatal. Thalidomide: estimates of 10,000–20,000 affected children worldwide, and 17 confirmed U.S. cases from samples distributed before approval was refused; Frances Oldham Kelsey's 1962 President's Award for Distinguished Federal Civilian Service. Beta-blockers: Beta-Blocker Heart Attack Trial (National Heart, Lung, and Blood Institute, 1981), total mortality 7.2% with propranolol vs. 9.8% with placebo; William M. Wardell, “More Regulation or Better Therapies?” Regulation (1979), estimating at least 10,000 preventable U.S. coronary deaths a year.