A Predictive Model for Chronic Hydrocephalus After Clipping Aneurysmal Subarachnoid Hemorrhage

Abstract

Chronic hydrocephalus after clipping aneurysmal subarachnoid hemorrhage (aSAH) often results in poor outcomes. This study was to establish and validate model to predict chronic hydrocephalus after aSAH by least absolute shrinkage and selection operator logistic regression. The model was constructed from a retrospectively analyzed. Two hundred forty-eight patients of aSAH were analyzed retrospectively in our hospital from January 2019 to December 2021, and the patients were divided into chronic hydrocephalus (CH) group (n=55) and non-CH group (n=193) according to whether occurred CH within 3 months. In summary, 16 candidate risk factors related to chronic hydrocephalus after aSAH were analyzed. Univariate analysis was performed to judging the risk factors for CH. The least absolute shrinkage and selection operator regression was used to filter risk factors. Subsequently, the nomogram was designed by the above variables. And area under the curve and calibration chart were used to detect the discrimination and goodness of fit of the nomogram, respectively. Finally, decision curve analysis was constructed to assess the practicability of the risk of chronic hydrocephalus by calculating the net benefits. Univariate analysis showed that age (60 y or older), aneurysm location, modified Fisher grade, Hunt-Hess grade, and the method for cerebrospinal fluid drainage, intracranial infections, and decompressive craniectomy were significantly related to CH ( P <0.05). Whereas 5 variables [age (60 y or older), posterior aneurysm, modified Fisher grade, Hunt-Hess grade, decompression craniectomy] from 16 candidate factors were filtered by LASSO logistic regression for further research. Area under the curve of this model was 0.892 (95% confidence interval: 0.799-0.981), indicating a good discrimination ability. Meanwhile, the result of calibration indicated a good fitting between the prediction probability and the actual probability. Finally, decision curve analysis showed a good clinical efficacy. In summary, this model could conveniently predict the occurrence of chronic hydrocephalus after aSAH. Meanwhile, it could help physicians to develop personalized treatment and close follow-up for these patients.