L arge, randomized controlled trials have shown that carotid endarterectomy is beneficial for selected patients with significant, recently symptomatic carotid stenosis. 1,2 However, surgery itself carries a significant risk of stroke and death. Most perioperative strokes are ipsilateral carotid territory cerebral infarcts and some may result from the temporary interruption of blood flow that occurs while the carotid artery is clamped. The duration of interrupted blood flow can be minimized by bridging the clamped section of the artery with a shunt. Although some surgeons advocate routine shunting, others prefer to use shunts selectively or avoid them altogether. 3,4 Potential disadvantages of shunting include complications such as air and plaque embolism and carotid artery dissection, and an increased risk of local complications such as nerve injury, hematoma, infection, and long-term restenosis. 5 However, reliable data on these risks are limited. 6 We performed a systematic review of randomized controlled trials to determine the effect of a policy of routine or selective shunting on the risk of perioperative stroke, death, and other operative complications. We also tried to identify if any one method of selecting which patients undergoing endarterectomy, under general anesthetic, require a shunt is better than any other.
MethodsSimilar search strategies and inclusion criteria were used in the original (up to 1995) and updated reviews (1995 to 2000). 7 Studies were included if they compared any of the following policies: routine shunting versus avoiding a shunt; selective shunting versus routine shunting; selective shunting versus avoiding a shunt-or if they examined different methods for determining the need for a shunt. Recorded endpoints were all strokes, ipsilateral strokes, all strokes and deaths, wound hemorrhage, wound infection, and ipsilateral cranial nerve injuries. The time periods examined were intraoperative, within 24 hours of surgery, within 30 days of surgery, and during the whole follow-up period. If available, long-term outcomes-such as restenosis of the operated artery, cognitive function at the end of follow-up, and the numbers of shunts inserted in trials comparing one method of monitoring with another-were also recorded. Data were analyzed on an "intention-to-treat" basis. If any of the necessary data were not reported, additional data were sought from the trialists. The analyses of surgical complications (including ipsilateral stroke) were based on all arteries randomized, whereas overall stroke and death rates were calculated per patient.