Abstract 202: Predictors of Percutaneous Endoscopic Gastrostomy in Intracerebral Hemorrhage

Abstract

Introduction: Intracerebral hemorrhage (ICH) patients often require percutaneous endoscopic gastrostomy (PEG) placement. Previous studies identifying predictors of PEG placement in ICH have been small and have not accounted for specific clinical findings on validated stroke severity scores. We sought to determine early predictors of PEG placement in a large cohort of ICH patients. Methods: We retrospectively identified all primary ICH cases by ICD-9 code at our two academic centers from 2009-2011. All cases were physician-reviewed. Patients who were made comfort care only, transferred to hospice, or died within 72 hours of presentation were excluded. Demographic data, stroke characteristics including clinical severity scores, processes of care, and outcomes were ascertained. We also constructed an abbreviated four-item NIHSS composite score, which included level of consciousness (item 1a), facial palsy (4), best language (9), and dysarthria (10). Variables with p-values <0.25 in the univariate were considered for regression analysis. Linearity of the logit was assessed for continuous variables. Results: Of the 610 subjects with primary ICH, 519 met inclusion criteria and 89 (17.1%) underwent PEG placement. Median time from presentation to PEG placement was 11.8 days. Univariate predictors of PEG placement are shown in the Table. In the best multivariate logistic regression model we found younger age (OR 0.99 (0.97, 1.00), p=0.04), higher four-item NIHSS composite score (OR 1.15 (1.05, 1.26), p=0.003), lower initial GCS (OR 0.91 (0.85, 0.97), p=0.003), and greater ICH volume (OR 2.97 (1.11, 7.97) for highest quintile vs. lowest, p=0.02) were independently associated with PEG placement. This model achieved an area under the curve of 0.78 with a good fit. Conclusions: We identified several independent early predictors of PEG placement in ICH. Early identification of ICH patients who will require PEG placement could result in reduced resource utilization, quicker initiation of rehabilitation, shorter hospital length of stay, and reduction of preventable complications. Prospective studies are required to determine optimal patient selection and timing for PEG placement.