A Novel Risk Calculator for Suboccipital Decompression for Adult Chiari Malformation. (16th November 2020)

• Record Type: Journal Article
• Authors: Feghali, James; Marinaro, Elizabeth; Lubelski, Daniel; Luciano, Mark G; Huang, Judy
• Is Part Of: Neurosurgery. Volume 67(2010)Supplement 1
• Publication Date: 2020-11-16
• Subjects: Nervous system -- Surgery -- Periodicals; 617.48005
• DOI: 10.1093/neuros/nyaa447_785
• Languages: English
• ISSNs: 0148-396X

Abstract

Abstract: INTRODUCTION: Patient counselling and selection for surgical therapy in adult Chiari malformation type I (CM-1) remains debatable. METHODS: The database from years 2011 through 2017 was queried to identify the subset of CM-1 patients undergoing suboccipital decompression. Univariable analysis was conducted to identify baseline factors associated with non-home discharge and 30-day re-operation following the initial decompression procedure. Logistic regression and the Akaike Information Criterion were used to identify the optimal models predictive of both outcomes. Performance was assessed using receiver operating curves and validated with bootstrapping. Calibration was assessed by verifying a non-significant P -value for the Hosmer-Lemeshow test. RESULTS: The sample consisted of 706 CM-1 patients of mostly females (80%) with an ASA classification of II (52%). The rate of non-home discharge was 5.2%, and the re-operation rate was 6.6% with most re-operation diagnoses being CSF-related and requiring CSF flow diversion or CSF leak repair. The optimal model predictive of non-home discharge consisted of age (odds ratio [OR] = 1.05, p = .001), diabetes (OR = 2.44, p = .080), and ASA class (OR = 1.94, p = .082) with an area under the curve (AUC) of 0.720. The bias-corrected AUC obtained with bootstrapping was 0.707. The model showed good calibration (Hosmer-Lemshow P -value = 0.430). The optimal model predictive of re-operation consisted of female sex (OR = 0.48, p = .031), BMI (OR = 1.05, p = .002), and ASA class (OR = 3.44, p = .001) with an AUC of 0.726. The bias-corrected AUC obtained through bootstrapping was 0.718. The model showed good calibration (Hosmer-Lemshow P -value = 0.953). A calculator for both outcomes was deployed under the following URL: https://jhuspine3.shinyapps.io/Discharge_Reop_Calculator/. CONCLUSION: We have used a large international database to develop a simple risk calculator based on readily available pre-operative variables. Following subsequent validation, this tool can help optimize patient counselling and decision-making in adult CM-1. In particular, a special focus on modifiable risk factors such as BMI can improve the pre-operative counselling process.