In the health care debate, most politicians, publications, and activists have been careful to continue to support--loudly--medical advances. Everyone wants to find a way to make health care cheaper, but no one wants be accused of advocating anything less than state-of-the-art medicine. Sunday, The New York Times editorial board comes out boldly with a statement that runs counter to this trend: "The Congressional Budget Office estimates," they begin, "that an astonishing half or more of the increased spending for health care in recent decades is due to technological, surgical and clinical advances." That should actually concern us, they argue, because, though advances are often "cause for celebration ... an expensive new drug is not always better than an older, cheaper drug." Here's the problem, they say:

To win approval, drugs and many devices must undergo tests for safety and effectiveness. For drugs, there is usually no comparison to products already approved. For both, there is no consideration of cost. Once drugs or devices are approved to treat one class of patients or illnesses, doctors can use them for virtually any ailment they please. Manufacturers eagerly promote their most expensive products to doctors and patients.

Patients have few ways to judge what is best for improving health or saving money. They must rely on doctors who may have insufficient information--or economic incentives to pick the costliest treatment.

The health care reform package already passed "makes a start at figuring this out," write the editors, but "after all the cynical demagoguing about 'death panels,'" it falls short of what is needed to streamline the system. "Research that systematically compares the effectiveness of different treatments and drugs is clearly needed." Furthermore, though it is now illegal to deny "Medicare coverage of services 'solely' on the basis of comparative effectiveness research," considering effectiveness "in conjunction with other factors" is allowed. In fact, say the editors, it should be encouraged.