Importance Several neurosurgical pathologies, ranging from glioblastoma to hemorrhagic stroke, use volume thresholds to guide treatment decisions. For chronic subdural hematoma (cSDH), with a risk of retreatment of 10-30%, the relationship between pre- and post-operative cSDH volume and retreatment is not well understood.

Objective Investigate the potential link between pre- and post- operative cSDH volumes and risk of retreatment.

Design Retrospective chart review

Setting Three level one trauma centers, February 2009 - August 2021

Participants Patients with unilateral cSDH

Methods We used a 3D deep learning, automated segmentation pipeline to calculate pre- and post-operative cSDH volumes. To identify volume thresholds, we constructed a receiver operating curve (ROC) using both pre- and post-operative volumes in to predict cSDH retreatment and selected the threshold with the highest Youden’s index. Then, we developed a light gradient boosting machine to predict risk of cSDH recurrence using cSDH volumes and clinical features.

Main Outcomes Surgical retreatment of cSDH

Results We identified 538 patients with unilateral cSDH, of whom 62 (12%) underwent surgical retreatment within six months of the index surgery. cSDH retreatment was associated with higher pre- (122 vs. 103 mL; p<0.001) and post-operative (62 vs. 35 mL; p<0.001) volumes. Patients with >140 mL pre-operative cSDH volume has nearly triple the risk of cSDH recurrence compared to those below 140 mL; while a post-operative volume >46 mL led to an increased risk for cSDH retreatment (22% versus 6%; p<0.001). On multivariate modeling, our model had an area under the curve of 0.76 (95% confidence interval: 0.60 - 0.93) for predicting cSDH retreatment. The most important features were pre- and post-operative volume, platelet count, and age.

Conclusions/Relevance Larger pre- and post-operative cSDH volumes increase the risk of cSDH retreatment. Volume thresholds may allow identification of patients at high risk of cSDH retreatment who would benefit from adjunct treatments. Machine learning algorithm can quickly provide accurate estimates of pre and post operative volumes.

Disclosures J. Vargas: 2; C; Viz.AI, Synchron, Borvo, Imperative Care. 4; C; Viz.AI, Imperative Care, Cerenovus, Q’APel. M. Pease: None. M. Snyder: None. J. Blalock: None. S. Wu: None. E. Nwachuku: None. A. Mital: None. D. Okonkwo: None. R. Kellogg: 2; C; VizAI, Cerenovus, Imperative Care. 4; C; VizAI.