RESUMEN
Objectives: To develop a scoring system based on independent predictors of the need for ventriculoperitoneal (VP) shunt after brain tumor resection in pediatric patients. Methods: A total of 416 pediatric patients (≤ 14 years old) with brain tumors who underwent surgery were randomly assigned to the training (n = 333) and validation cohorts (n = 83). Based on the implementation of VP shunt, the training cohort was divided into the VP shunt group (n = 35) and the non-VP shunt group (n = 298). Univariate and multivariate logistic analyses were performed. A scoring system was developed based on clinical characteristics and operative data, and scores and corresponding risks were calculated. Results: Age < 3 (p = 0.010, odds ratio [OR] = 3.162), blood loss (BL) (p = 0.005, OR = 1.300), midline tumor location (p < 0.001, OR = 5.750), preoperative hydrocephalus (p = 0.001, OR = 7.044), and total resection (p = 0.025, OR = 0.284) were identified as independent predictors. The area under the curve (AUC) of the scoring system was higher than those of age < 3, BL, midline tumor location, preoperative hydrocephalus, and total resection (0.859 vs. 0.598, 0.717, 0.725, 0.705, and 0.555, respectively; p < 0.001). Furthermore, the scoring system showed good performance in the validation cohort (AUC = 0.971). The cutoff value for predictive scores was 5.5 points, which categorized patients into low risk (0-5 points) and high risk (6-14 points) groups. Conclusions: Our scoring system, integrating age < 3, BL, midline tumor location, preoperative hydrocephalus, and total resection, provides a practical evaluation. Scores ranging from 6 to 14 points indicate high risk.
RESUMEN
BACKGROUND: To identify the risk factors for postoperative hydrocephalus and the need for ventriculoperitoneal (VP) shunt after posterior fossa tumor (PFT) resection in pediatric patients and establish a predictive model. METHODS: A total of 217 pediatric patients (≤14 years old) with PFTs who underwent tumor resection from November 2010 to December 2020 were divided into a VP shunt group (n = 29) and non-VP shunt group (n = 188). Univariate and multivariate logistic regression were performed. A predictive model was established based on the independent predictors. Receiver operating characteristic curves were generated to determine the cutoff values and areas under the curve (AUCs). The Delong test was performed to compare the AUCs. RESULTS: Age less than three years (P = 0.015, odds ratio [OR] = 3.760), blood loss (BL) (P = 0.002, OR = 1.601), and locations at fourth ventricle (P < 0.001, OR = 7.697) were the independent predictors. The predictive model was as follows: total score = age (<3; yes = 2, no = 0) + BL + tumor locations (fourth ventricle; yes = 5, no = 0). The AUC of our model was higher than those of age less than three years, BL, locations at the fourth ventricle, and compound factors (age <3 + locations) (0.842 vs 0.609, 0.734, 0.732, and 0.788, respectively). The cutoff values of the model and BL were 7.5 points and 2.75 U, respectively. CONCLUSIONS: BL, age less than three years, and tumors at the fourth ventricle were independent predictors. Model scores over 7.5 points predict a high risk.
