It seems that not a week goes by before the safety of another popular prescription drug is being questioned. Vioxx and hormone replacement therapy have gone before; the latest is the popular diabetes drug rosiglitazone (Avandia), which is now linked to an increased risk of heart attacks.

This warning comes from a new meta-analysis, reported recently in the New England Journal of Medicine, which found that Avandia raised heart attack risk by 43 percent. The manufacturer of the drug, GlaxoSmithKline, promptly issued a statement strongly disagreeing with the conclusions of the study.

First, let's place this number of 43 percent in perspective. Prempro (a combination of estrogen and a progestational agent) increased the risk of fatal and non-fatal heart attacks by 23 percent in the postmenopausal women in the Women's Health Initiative.

And Vioxx, a widely-prescribed anti-inflammatory drug, more than doubled the risk of heart attacks and strokes. In 2004, Merck voluntarily withdrew this drug from the market.

Avandia lowers blood glucose in patients with type 2 diabetes. It belongs to the class of drugs called the thiazolidinediones. The cardiovascular outcomes of another thiazolidinedione, pioglitazone (Actos), have also been examined. It turns out that Actos actually protects against cardiovascular disease.

One possible explanation for the differences between these two drugs from the same class: Avandia raises LDL cholesterol (the so-called "bad cholesterol") while Actos has a more favorable effect on blood lipids, especially on triglycerides.

The thiazolidinediones are widely prescribed despite common side effects of weight gain, fluid retention, and even heart failure in certain patients at risk. 

What should you do if you are taking Avandia? Talk with your doctor about the relative benefits and risks of the drug, and do not stop taking it without your doctor's approval.

Now the Food and Drug Administration must decide what action to take based on this new information. Avandia is used by millions of people in the United States and many of them may be at risk. However, the authors of the study and an accompanying editorial noted several significant limitations to the results of this meta-analysis.

For example, the total number of heart attacks and cardiovascular deaths in the study was relatively low, so a small change in the numbers could greatly reduce the risk or even indicate that the finding of a greater risk was a chance observation. The FDA will need to review all the available data on the drug before making a final decision. 

All of these drug recalls and warnings have led many people to distrust the FDA's ability to protect us against potentially dangerous medications. The FDA approves a drug if trials show that it is safe and effective in improving certain markers, such as blood pressure, cholesterol, and blood glucose levels.

The assumption is that improving these markers also improves the long-term health of people taking the medications. The authors of the Avandia study, however, question the use of such markers in the approval of a medication. They point out that lowering blood glucose levels with Avandia has never been proven to prevent the complications of diabetes.

I don't agree that a drug company must prove that a new drug provides long-term benefits in addition to its effect on specific markers before it is approved by the FDA. This requirement would be so costly that companies might never undertake the necessary studies.

As a result, approval of valuable drugs could be delayed, or they might never become available. Blood pressure drugs were approved because they lowered blood pressure, and only later were proven to also prevent heart attacks and strokes. Based on its cholesterol-lowering effect, the FDA approved the first statin drug, lovastatin, in 1987, even though the benefits of the statins on cardiovascular disease were first proven seven years later.

On the other hand, I strongly agree with the authors of the meta-analysis and the editorial who emphasize that the FDA must require drug companies to complete large, long-term clinical trials to evaluate the safety and effectiveness of a medication soon after it has been approved.

One of the operative words is soon; too many years have elapsed since Avandia was approved in 1999, and the recent publication of evidence of its possible cardiovascular dangers.