Quantitative Assessment of Ventriculostomy-Related Hemorrhage: A Volume-Based Classification System to Predict New Neurological Symptoms.

Abstract

Hemorrhage is one of the most feared complications following ventriculostomy placement. Current studies have assessed factors associated with increased risk of ventriculostomy-related hemorrhage (VRH). However, the clinical significance of VRH has not been determined. To correlate quantitative volumetric measurements of VRH with new neurological symptoms. A retrospective review of our institutional database of ventriculostomy patients during the last decade was performed. Patients' demographics and procedural details such as indication, number of passes and position of the catheter were recorded. VRH volume was quantified on noncontrast head computed tomography using the Picture Archiving Communication System (Carestream Vue®, Rochester, New York) semi-automated livewire segmentation tool. Patients with new neurological symptoms within 48 h of VRH were considered symptomatic. Several clinical confounders were ruled out. Logistic regression analyses were performed. The best volumetric cut-offs in predicting symptomatic VRH were determined through receiver operating characteristic (ROC) curve analysis. A total of 3090 patients underwent ventriculostomy procedures and 179 (∼6%) developed VRH. A total of 41 (1.06%) patients with VRH developed new neurological symptoms. Only 12 (0.39%) were attributable to a new VRH. Multivariable logistic regression showed that volume of the hemorrhage (OR 1.17, P=.006) is the only significant predictor of symptomatic VRH. ROC curve analysis demonstrated that VRH volume <1.10 cc has 91.7% sensitivity to rule out symptomatic VRH, whereas a volume >7.59 cc has 95.5% specificity to predict symptomatic VRH. Approximately 6% of patients developed postprocedural VRH, but only 0.4% were symptomatic. VRH volumes <1 cc are extremely unlikely to become symptomatic, whereas volumes >7.5 cc may predict development of new neurological deficits.