Independent Predictors for Operative Treatment of Chronic Subdural Hematoma

Objectives Treatment of chronic subdural hematoma (CSDH) can be conservative or operative. Since clinical guidelines are lacking, a large variation in daily management of CSDH exists. We aimed to quantify the number of CSDH patients that are treated operatively and conservatively and to develop a prediction model that provides insight into the arguments behind the choice for operative treatment. Methods All consecutive CSDH patients ≥ 18 years that were treated in three neurosurgical centers in the Netherlands between 2003 and 2019 were retrospectively included. Our outcome measure was the decision for operative treatment after initial diagnosis. Combining conventional multivariable logistic regression analysis and machine learning, four prediction models were developed and compared. The prediction model with the best performance was used for further analyses. Results Of 684 included patients (29% female, median age 75 years), 428 (63%) were treated operatively. The multivariable logistic regression model performed best with an area under the curve of 0.90, a sensitivity of 0.87, and a specificity of 0.78. Independent predictors for operative treatment were treating facility, younger age, premorbid mRS ≤ 1, focal neurological deficit, cognitive impairment, and MGS score on admission, larger hematoma volume, and ≥ 10 mm midline shift on baseline CT. The highest predictive value for operative treatment was found for treating facility, hematoma volume, and ≥ 10 mm midline shift. Conclusion While most CSDH patients are treated operatively, over a third are treated conservatively. Decisions for operative treatment are influenced by premorbid level of functioning, symptoms on admission, CT parameters, and, notably, interhospital variation, highlighting the need for standardized treatment guidelines.