A revealing notice appeared last month in the Federal Register, the US government compendium of agency rules and notices. The Food and Drug Administration (FDA) was reporting on compliance by pharmaceutical companies with its requests for studies of the clinical safety and efficacy of drugs already on the market. Of nearly 1,200 such studies committed to by drug firms but not completed, some 70% have yet to begin.

This is an alarming reflection of the state of the government's vigilance. Last year, the FDA belatedly faced up to research showing that the painkiller Vioxx, which it approved in 1999, markedly increases the risk of heart attacks and strokes. It has been estimated that more than 25,000 people died before Merck pulled the drug from the market in September (see pages 554 and 557). A better approach to assessing the safety of marketed drugs is badly needed.

Under the FDA's existing system, known as MedWatch, doctors voluntarily report suspected side-effects — but epidemiologists estimate that this captures only 10% of adverse events. The agency used to complement this with university-run studies, but this modest effort dried up in the 1990s when it was forced to devote more resources to speeding up drug approvals.

The upshot is that the FDA depends on companies for post-market safety studies but has no legal authority to force firms to do them. The results are sobering. For instance, after 16 years of use in tens of millions of people, it is still not known whether selective serotonin reuptake inhibitors, the blockbuster antidepressants, cause an increase in suicide attempts in some adults. The FDA warned last year of such a risk in young people, but the question still hasn't been adequately explored in adults. Nor will it be if the current system prevails.

Lack of information is not the problem. After a drug comes into use, gigabytes of epidemiological data become available from governments and healthcare organizations. Nor is it a matter of sorting through the 10,000-odd prescription drugs to find one with dangerous side-effects: the most likely culprits are already well known.

What is lacking is both the money and the mandate. The United States needs a government body that can not only determine what drug safety studies are needed, but demand that they are done. It must also have the authority to dictate — rather than negotiate with drug-makers, as the FDA currently does — that beefed-up warning labels are used when evidence of new risks emerges. About $300,000 annually would pay for a high-quality pharmaco-epidemiological study. Surely the richest country in the world can find the funds for that.

It can, but will it? Congress is unlikely to implement such a change. The Republicans control Congress and the White House, and are loath to alienate an industry that has given them twice as much as it has given the Democrats in political contributions over the past decade.

Tens of thousands of people have almost certainly died because of Vioxx. Observers are left to wonder if it will take an even bigger tragedy to force the US government to do the right thing on drug safety.