Novel Comorbidity Index for Pediatric Neurosurgical Procedures

Introduction: Pediatric neurosurgical procedures are complex and high-risk. Comorbidity indexes are important tools for assessing disease burden and predicting outcomes, but existing tools like the Charlson and Elixhauser indexes are designed for adults, limiting their use in pediatric patients. This study aimed to create a comorbidity index specifically for pediatric neurosurgery patients, using data from the Healthcare Cost and Utilization Projects’ Kids Inpatient Database (HCUP-KID) with in-hospital mortality as the primary outcome.

Methods: Pediatric patients undergoing neurosurgical procedures were extracted from the HCUP-KID, using the 2016 (n = 22,343) as the training data and 2019 (n = 23,044) as the validation set. The Clinical Classification Software clustered individual ICD-10 codes. From these groupings, 167 were chosen based on a plausible relationship to mortality, pre-admission existence, and independence from the principal reason for hospitalization. These along with age (years), neonatal status (<=28 days old), and gender, comprised the final feature set. A Logistic Regression was trained and tested. The new model's performance was compared to the Elixhauser Van Walraven and Charlson Swiss scores, using the area under the receiver operator characteristic (AUROC).

Results: On the 2019 HCUP-KID data, the Elixhauser and Charlson scores had AUROCs of 0.7167 and 0.4958, respectively. The logistic regression model AUROC was 0.8122. Significant comorbidities included pituitary disorders (OR: 13.2,95%CI: 11.5-15.3), and coagulation and hemorrhagic disorders (OR: 7.6,95%CI:6.5-8.8). Other comorbidities such as obesity (OR: 0.43,95%CI:0.28-0.67) and asthma (OR: 0.48,95%CI:0.36-0.64) were not significant after Holm-Bonferroni correction.

Conclusion : This study presents novel comorbidity scoring systems for pediatric neurosurgery that outperform adult indexes. Pediatric neurosurgery patients have unique comorbidities affecting surgical risk. In our model, coagulopathies have a higher odds ratio than liver disease, opposite to the Elixhauser index. This new pediatric comorbidity index can be used to guide surgical decision-making and risk-stratify patients in outcomes research.