Procedural risks of carotid intervention in 19,000 patients

Background Randomized controlled trials (RCTs) show that carotid endarterectomy (CEA) and carotid stenting (CAS) reduce long-term stroke risk in symptomatic and asymptomatic patients with carotid artery stenosis. Historical RCTs may not represent contemporary practice and administrative datasets may estimate procedural risks more reliably. We studied procedural risks after carotid intervention in a novel, international administrative data set of 18,997 patients admitted to 28 hospitals across 7 countries. Methods Symptomatic and asymptomatic patients undergoing CEA (n = 16,220) and CAS (n = 2,777) between 2011 and 2015 were studied retrospectively. The primary outcome was in-hospital death within seven days. The secondary outcome was the proportion of patients whose length of hospital stay (LOS) exceeded 2 days. We also describe the rate of computerized tomography brain imaging within 2 days of CEA and CAS (proxy for stroke) as procedural strokes were not reliably recorded. Results In symptomatic patients after CEA, mortality was 0.2% [5/2,118] (95% confidence interval: 0.1–0.5), and 57.0% [628/1,101] (54.1–60.0) had prolonged LOS. In asymptomatic patients after CEA, mortality was 0.1% [21/14,102] (0.1–0.2), and 28.5% [2,864/10,039] (27.7–29.4) had prolonged LOS. In symptomatic patients after CAS, mortality was 3.3% [10/307] (1.3–5.2), and 64.3% [144/224] (58.0–70.5) had prolonged LOS. In asymptomatic patients after CAS, mortality was 0.7% [18/2,470] (0.4–1.1), and 27.5% [601/2,187] (25.6–29.4) had prolonged LOS. After CEA, 8.1% [89/1,101] (6.5–9.7) symptomatic patients and 2.1% [207/10,039] (1.8–2.3) asymptomatic patients underwent brain imaging. After CAS, 7.1% [16/224] (4.0–10.7) symptomatic patients and 3.2% [71/2,187] (2.5–4.0) asymptomatic patients underwent brain imaging. Conclusions Death and LOS after CEA and CAS were higher in symptomatic than asymptomatic patients. Symptomatic patients undergoing CAS had particularly increased risk of death. This may be partly explained by case selection, with more comorbid patients preferentially undergoing CAS. While RCTs effectively compare long-term efficacy of CEA versus CAS, administrative datasets can provide reliable estimates of contemporary procedural risks.