Differences in complication rates among the centres in the SPACE study.
INTRODUCTION: Despite the high grade of standardisation of study protocols, there is still room for variability among the centres in specific treatment aspects. We evaluated the treatment risk in stent-protected angioplasty of the carotid versus endarterectomy (SPACE) associated with the specific patient enrolment rates of the centres. MATERIALS AND METHODS: The analysed endpoints were ipsilateral stroke or death [primary outcome event (pOE)] and any stroke or death [secondary outcome event (sOE)] until 30 days after treatment. A binary logistic regression analysis with random effects was performed separately for each treatment arm. The centres were secondarily categorised in three classes: I)>/=25 patients enrolled, II) ten to 24 patients and III)<10 patients and a hierarchic log linear model was fitted to test the three-way interaction of treatment, number of patients per class and outcome. RESULTS: The random effects logistic regression analysis in the carotid artery stenting (CAS) arm proved a significant increase in pOE with decreasing number of patients enrolled (-0.0190 +/- 0.0085, p = 0.025, deviance 35.7 with 32 df), whereas no such effect was found in the carotid endartectomy (CEA) arm (-0.010 +/- 0.008, p = 0.24, deviance 39.78 with 32 df). In the log linear model, there was a significant interaction between treatment, number of patients per centre and sOE (p = 0.023). The odds ratios for sOE in the enrolment classes (CAS vs. CEA) were 0.98 (95% CI 0.50-1.94, p = 0.95) for class I, 1.13 (95% CI 0.47-2.77, p = 0.77) for class II and 11.56 (95% CI 1.40-253.45, p = 0.01) for class III centres. CONCLUSION: Despite rigorous standardisation and quality requirements for operator qualification, there seemed to be a decrease in complication rate withincreasing patient enrolment numbers in the CAS arm while this signal could not be detected in the CEA arm of SPACE.