Learning Curve of Mechanical Thrombectomy: Procedural Efficiency and Clinical Outcomes in Early Operator Experience.

PURPOSE: The learning curve for interventional neuroradiologists starting independent mechanical thrombectomy (MT) remains insufficiently understood. This study investigated associations between operator experience, procedural performance, and clinical outcomes to estimate the number of procedures for expert-level performance.

METHODS

We retrospectively analyzed MT procedures from our stroke registry (2020-2024). Operators with low-to-intermediate prior neurovascular experience were followed during their first 20-50 independent cases. Procedural outcomes included procedure duration, number of passes, door-to-groin-puncture time, first-pass success, reperfusion success (modified treatment in cerebral infarction (mTICI) ≥ 2b-3), and complications. Learning curves were benchmarked against expert operator performance. Associations between operator experience and outcomes were evaluated with mixed-effects models. Clinical outcomes were assessed using changes in the National Institutes of Health Stroke Scale (NIHSS) and modified Rankin Scale (mRS) from admission to discharge and at 3‑month follow-up.

RESULTS

295 MT procedures were performed by seven operators (median 50 each; interquartile range 25-50). Procedure duration decreased significantly with experience (β = -0.7 min per procedure; 95% CI -1.2 to -0.1; p = 0.01). Linear extrapolation suggested that procedural efficiency comparable to the local expert benchmark may be reached after approximately 188 procedures, although this estimate should be interpreted cautiously. Other procedural metrics showed non-significant improvements. NIHSS at discharge and mRS at 3‑month follow-up did not differ by experience level.

CONCLUSION

Increasing operator experience was associated with shorter procedure duration, whereas no significant associations were observed for other procedural metrics. No differences in procedural complications or clinical outcomes were detected between operator experience groups.