Outcome prediction in intracranial tumor surgery: the National Surgical Quality Improvement Program 2005-2010 JOURNAL OF NEURO-ONCOLOGY Bekelis, K., Bakhoum, S. F., Desai, A., MacKenzie, T. A., Roberts, D. W. 2013; 113 (1): 57-64


Accurate knowledge of individualized risks is crucial for decision-making in the surgical management of patients with brain tumors. Precise delineation of those risks remains a topic of debate. We attempted to create a predictive model of outcomes in patients undergoing craniotomies for tumor resection (CTR). We performed a retrospective cohort study involving patients who underwent CTR from 2005 to 2010 and were registered in the American College of Surgeons National Quality Improvement Project database. A model for outcome prediction based on individual patient characteristics was developed. Of the 1,834 patients, 457 had meningiomas (24.9 %) and 1377 had non-meningioma tumors (75.1 %). The respective 30-day postoperative risks were 2.1 % for stroke, 1.3 % for MI, 2.7 % for death, 2.4 % for deep surgical site infection, and 6.6 % for return to the OR. Multivariate analysis demonstrated that pre-operative tumor-related neurologic deficit, stroke, altered mental status, and weight loss, were independently associated with most outcomes, including post-operative MI, death, and deep surgical site infection. An additive effect of the variables on the risk of all outcomes was observed. A validated model for outcome prediction based on individual patient characteristics was developed. The accuracy of the model was estimated by the area under the receiver operating characteristic curve, which was 0.687, 0.929, 0.749, 0.746, and 0.679 for postoperative risk of stroke, MI, death, infection, and return to the OR, respectively. Our model can provide individualized estimates of the risks of post-operative complications based on pre-operative conditions, and can potentially be utilized as an adjunct in the decision-making for surgical intervention in brain tumor patients.

View details for DOI 10.1007/s11060-013-1089-3

View details for Web of Science ID 000318300700007

View details for PubMedID 23436132