Six RCTs (n=1,177) met the inclusion criteria of the review. Five trials were included in the meta-analysis (n=1,154); one (n=23) was excluded as the sample size and number of events were too small to be included in a valid analysis.
The median quality score according to the Jadad scale was 3 (range: 1 to 3).
No statistically significant difference was observed between CAS and CEA for the 1-month composite stroke or death rate (RR 1.3, 95% CI: 0.6, 2.8, P=0.5), based on 1,154 participants in 5 RCTs. There was evidence of statistical heterogeneity (P=0.07). The sensitivity analyses found that the results were similar when using a fixed-effect model, when excluding 2 RCTs accounting for most patients, and when excluding 1 RCT using distal protection devices. A subgroup analysis of studies of symptomatic patients also found no statistically significant difference. There was no evidence of publication bias from the funnel plot, Begg's test (P=0.62) or Egger's test (P=0.69).
No statistically significant difference was observed between CAS and CEA for the 1-month stroke rate (RR 1.3, 95% CI: 0.4, 3.6, P=0.4) or disabling stroke (RR 0.9, 95% CI: 0.2, 3.5, P=0.9), based on 831 participants in 3 RCTs.
CAS was associated with a statistically significant reduction in 1-month MI rates (RR 0.3, 95% CI: 0.1, 0.8, P=0.02; 814 patients in 2 RCTs) and cranial nerve injury (RR 0.05, 95% CI: 0.01, 0.3, P<0.001; 918 patients in 3 RCTs) compared with CEA.
No statistically significant difference was observed between CAS and CEA for major bleeding (RR 0.6, 95% CI: 0.2, 2.1; 918 patients in 3 RCTs), 1-year or more rates of ipsilateral stroke (RR 0.8, 95% CI: 0.5, 1.2, P=0.2; 814 patients in 2 RCTs) or for revasculisation (RR 0.9, 95% CI: 0.1, 19.4, P=0.9; 657 patients in 2 RCTs).