In recent years, the rate of carotid endarterectomy has more than doubled in the United States, with more than 150,000 procedures performed in 1998. However, experts disagree on the best method to select patients for this procedure. European guidelines are based on studies that show benefit only in men with more than 80 percent stenosis and women with stenosis of more than 90 percent. However, North American guidelines recommend that surgery be considered in symptomatic patients with stenosis of 50 percent or more. To overcome differences in measurements between various clinical trials, Rothwell and colleagues re-examined data from the three largest studies to determine the effectiveness of endarterectomy at different levels of stenosis.

Because two of the five randomized controlled trials of endarterectomy are not compatible with current surgical practice, the researchers obtained data from the remaining three studies, the European Carotid Surgery Trial, the North American Symptomatic Carotid Endarterectomy Trial, and the Veterans Affairs trial. These three trials recruited patients with symptoms attributable to demonstrated stenosis of the carotid artery and randomly allocated patients to best medical treatment or immediate carotid endarterectomy and best medical treatment. Data were available for more than 6,000 patients who were followed for 35,000 patient-years. The researchers reassessed original data for each patient to standardize definitions of stenosis and techniques of measurement and to establish common outcome measures for the three trials. After reassessment of each case, data were reanalyzed to assess death during surgery or within 30 days after surgery, time to first stroke, and any stroke within 30 days of surgery.

Overall, surgery increased the risk of stroke within five years for patients with less than 30 percent stenosis (see the accompanying table). In patients with stenosis between 30 and 49 percent, surgery had no effect. Surgery had marginal benefit when stenosis was 50 to 69 percent and was highly beneficial in patients with at least 70 percent stenosis without near occlusion.

The authors stress that carotid endarterectomy can have significant morbidity and is associated with some mortality, and that the morbidity from minor vascular changes in the brain may not be recognized or documented. Patients face a 7 percent risk of death or stroke within 30 days of the procedure. The decision about treatment of symptomatic carotid stenosis must be individualized, but these findings indicate that surgery is beneficial only when the stenosis is at least 50 percent. The greatest benefit is reached in stenoses of more than 70 percent.

editor's note: One of the most challenging tasks for family physicians is to dissuade patients or families from inappropriately aggressive therapy. This study indicates that even in patients who have experienced significant symptoms, medical therapy is more appropriate than surgical treatment unless there is at least 70 percent stenosis. The study does not help with asymptomatic patients whose stenosis is discovered incidentally or by screening. Even in symptomatic patients, subgroup analysis would be highly useful to help us advise patients who are often anxious to “do everything possible to avert a stroke.” For many patients, the surgery seems to make sense, and they do not welcome cautions or the advice that medical therapy can be more effective. Even after a bad outcome, I have had families say they were “glad everything possible was done.” Why is conservative treatment always assumed to be inferior? What happened to “First, do no harm”?—A.D.W.