1)People with diabetes have too much sugar in their blood, so a drug that lowers blood sugar ought to be a good treatment, right? 2) Maybe not. Consider the diabetes drug Avandia, or rosiglitazone, which was approved in 1999．it lowers blood sugar, and about a million people in the United States have been talking it for Type 2 diabetes, the most common form. of the disease. But last week, doctors reported that Avandia might increase the risk of heart attacks. 3) Heart disease is a major complication of diabetes, so a drug that could make the risk even worse is bad news indeed. 4) The jury is still out on Avandia. Meanwhile, patient advocates and some politicians and researchers are already denouncing it, and the Food and Drug Administration has issued a tepid 'safety alert' telling patients to ask their doctors what to do while the agency 'is carefully weighing several complex sources of data.' Avandia's manufacturer, GlaxoSmithKline, insists it is safe. Personal injury lawyers are advertising on the Internet for clients who think they were injured by the drug. 5) What happened here reflects a larger question—the tricky problem of how to judge whether a drug is safe and effective. Avandia was approved because it lowered blood sugar, and seemed safe in clinical trials. 6) But the real test of whether a drug is any good is how are the patients? Not their blood tests or X-rays or EKGs, but the people themselves, and not after just six months, but after years, especially if they have a chronic disease and will be talking medicine for the rest of their lives. Are those talking the drug more or less likely than people not talking it to have heart attacks, die or develop heart disease or other illnesses? 7) The problem is, it can take a long time and a lot of patients—and, therefore, a lot of money — to get a real picture of health and survival. That is especially true for something like heart disease, which develops slowly and is so common that it may be hard to detect a small increase in risk. Studies might have to go on for years instead of months, and include far more than the few thousand patients in whom drugs are typically tested before they get approved. 8) So instead of waiting to see if people die or have heart attacks, drug companies have looked for other traits that seem to correlate with health and survival and that could stand in as a yardstick—objective measures like blood pressure, cholesterol (胆固醇) levels, blood sugar or tests of heart function. Researchers call these measurements 'surrogate endpoints,' and the F. D．A．has encouraged companies to find surrogates that could reliably predict how patients would fare. These kinds of tests are seen as a way to streamline the drug approval process. 9) But reliable surrogates are hard to find. There are plenty of endpoints that in theory should do the job, but do not. Tumor size, for instance: there are drugs that Can shrink tumors without prolonging a patient's life. Bone density is another example. Fluoride can increase it in people whose skeletons have thinned from osteoporosis (骨质疏松症) , so fluoride should prevent fractures. But it doesn't, in fact, it makes fractures more likely, because it turns bones brittle. 10) Heart rhythm can also be deceptive. Certain medicines can stabilize dangerous, abnormal heartbeats in people who have had heart attacks—and yet have been found to increase their odds of dying. Cholesterol levels do not always tell the whole story, either. Hormone treatment in women after menopause (绝经期) can raise HDL (高密度脂蛋白) , the so-called good cholesterol, and so was expected to prevent heart disease— but does not. Similarly, researchers had high hopes for an experimental drug that raises HDL, but instead of preventing heart attacks the drug wound up increasing the risk. 11) Part of the problem is that surrogat A．the whole society denounced it severely