Discriminant analysis prediction of the need for ventriculoperitoneal shunt after subarachnoid hemorrhage

Shunt operations have conventionally been performed to deal with normal-pressure hydrocephalus after subarachnoid hemorrhage. The indication and timing of shunt operations are often based on clinical symptoms and head computed tomography findings, and the early identification of the need for such surgery would be advantageous. The present study examined whether this need could be predicted solely on the basis of data collected on patient admission. A total of 120 consecutive patients with subarachnoid hemorrhage who underwent radical surgery for aneurysm were analyzed for potential risk factors for the onset of hydrocephalus that could be investigated on admission. Statistically significant differences between those patients who required a shunt operation and those who did not were found in terms of age, Hunt and Kosnik grade on first visit, Glasgow Coma Scale score on first visit, Fisher group, presence/absence of hydrocephalus, presence/absence of intraventricular hemorrhage, and transverse dimension of the third ventricle and distance between lateral ventricles measured by head computed tomography scan on first visit. Discriminant analysis performed on these 8 variables yielded a single discriminant function with a high sensitivity of 85.3% and a high specificity of 87.2%. Our findings indicate that the discriminant function is capable of predicting the need for the shunt operation soon after patient admission and can shorten the waiting time for the operation, and hence can be expected to contribute to decreasing the length of hospital stay in these patients.