1094. Factors Associated with Increased Length-of-Stay Following Aneurysmal Subarachnoid Hemorrhage
Authors: Campbell Liles; Jonathan Dallas, BS; Stephen Gannon, CCRP; Chevis Shannon, MBA, MPH, DrPH; Matthew Fusco, MD; Rohan Chitale, MD (Nashville, TN)
Introduction: Most subarachnoid hemorrhage (SAH) length of stay (LOS) models use the Nationwide Inpatient Sample (NIS) database, which has limited real world applicability due to inadequate assessment of SAH severity, comorbidities, and complications. This study applies a previously validated synthetic Hunt & Hess equivalent, the NIS-SAH Severity Scale (NIS-SSS), the Neurovascular Comorbidity Index (NCI), and in-hospital complications to improve aneurysmal SAH LOS modelling accuracy and relevance. Methods: SAH patients from 2012-2015 were identified by applying ICD-9 diagnostic/procedural codes for SAH (430), aneurysm coiling (397.2, 397.9, 395.2) and aneurysm clipping (395.1) to the unweighted NIS database. Patients with in-hospital mortality, head trauma, AVM/AVF, and LOS less than one day were excluded. The cohort underwent univariate and multiple linear regressions (at P<0.05) with independent factors pertaining to SAH severity (NIS-SSS), neurovascular comorbidities (NCI), complications, hospital type, and patient sociodemographics. Missing data was imputed via the “multivariate imputation by chained equations” method. Results: 4708 patients met inclusion criteria. Following multiple linear regression, variables predictive of increased LOS include infectious complications (ß-coefficient: 5.06, P<.001), medical complications (ß-coefficient: 3.89, P<.001), surgical complications (ß-coefficient: 2.97, P<.001), vasospasm (ß-coefficient: 1.80, P<.001), severity (NIS-SSS) (ß-coefficient: .61, P<.001), comorbidities (NCI) (ß-coefficient: .69, P<.001), medium and large bed hospitals (ß-coefficient: 1.95, P=.028 and ß-coefficient:2.47, P=.002 respectively), non-Medicare insurance (notably Medicaid (ß-coefficient: 5.45, P<.001), Self-Pay (ß-coefficient: 2.16, P=.001), and private (ß-coefficient: 1.86, P<.001)), age (ß-coefficient: .033, P=.016), and African-American race (ß-coefficient: .99, P=.03). Shorter LOS was associated with coiling (ß-coefficient: -1.04, P=.002) and private, non-profit hospitals (ß-coefficient: -1.43, P<.001). Conclusion: Using improved NIS severity and comorbidity scales, our analysis finds that SAH LOS is significantly affected by severity, comorbidities, complications, insurance status, hospital type, and race. In conjunction with improved mortality modelling, this predictive model can help clinicians better manage patient recovery by addressing modifiable risk factors affecting LOS.