Analysis of clinical use Supplementary Spetzler-Martin grading system for brain arteriovenous malforma-tions / 中国神经精神疾病杂志

ABSTRACT

Objective We evaluate if supplementary grading system can refine patient selection and accurately predict neurological outcome in BAVM. Methods We retrospectively study 221 BAVM patients who were treated micro?surgically by our hospital. The score of pre and post operation mRS and relative clinical, radiology data were collected. Two different logistic models (Spetzler-Martin, Supplementary Spetzler-Martin grading model) were constructed to com?pare the area under ROC. Results Some factors are significant different between worse outcome patients and good out?come patientsNon-hemorrhagic presentations prior surgery, AVM bigger than 3cm, diffuse shape of AVM and the elder patients. Predictive accuracy was higher for the supplementary model (ROC area, 0.91), than the Spetzler-Martin model (ROC area, 0.774). So the predictive accuracy of supplementary model was significantly better than that of the Spet?zler-Martin model (P=0.0362). Conclusions Supplementary Spetzler-Martin model can improve preoperative risk pre?diction and subgroup the patients more efficiently. When the score less than 5(including 5) in supplementary Spet?zler-Martin patients seem to have lower risk relative to surgery.