Abstract P290: Predicting Readmission and Mortality after Ischemic Stroke

Background: Stroke patients have high rates of all-cause rehospitalization at 30 days (7-24%) and within one year (30-62%). Limited information is available on how to predict these outcomes. We assessed two existing predictive models and whether adding NIHSS or modified Rankin scale (mRS) score improved their predictive ability for 30-day non-elective rehospitalization and/or mortality post-stroke. Methods: Using a cohort of ischemic stroke patients in a large multi-ethnic integrated healthcare system from June 2010 to December 2017, we tested 2 existing predictive models for a composite outcome (30-day non-elective hospitalization or death). The models were based on administrative data (LACE) as well as a comprehensive model (Transition Support Level, TSL) that included multiple clinical data elements from the electronic health record, including admission acute physiology and discharge care directives. LACE score included information on length of stay, acuity, comorbidities, and emergency department visits. The models, NIHSS and mRS scores were tested independently, and in combination and combined with age and sex. We assessed model performance using the area under the receiver operator characteristic (c statistic), Nagelkerke pseudo-R 2 , and Brier score. Results: The study cohort included 4843 patients with 5,014 stroke hospitalizations. Average age was 71.9 ± 14 years with 50.6% female and 52.1% White. Median initial NIHSS was 4. Median length of stay was 4 days, and 55.6% were discharged home. There were 527 non-elective readmissions within 30 days. The logistic models revealed that the best performing models were TSL (c-statistic = 0.685) and TSL plus mRS score at discharge (c-statistic = 0.694) [Table]. Conclusions: We found that neither the initial NIHSS nor the mRS score at discharge enhanced the predictive ability of the LACE or TSL models. Future efforts at prediction of short-term stroke outcomes will need to incorporate new data elements.