Background: This study was undertaken to identify predictors of hospital length of stay (LOS) for heart failure (HF) patients using clinical variables available at the time of admission and hospital characteristics.
Methods and results: A cohort of 70,094 HF patients discharged to home from 246 hospitals participating in the Get With the Guidelines-Heart Failure was analyzed for admission predictors for LOS. The analysis incorporated patient characteristics (PC) first, then added hospital characteristics (HC) followed by standard laboratory evaluations (SL), including troponin and brain natriuretic peptide (BNP). There were 31,995 patients (45.6%) with LOS < 4 days, 26,750 (38.2%) with LOS 4 to 7 days, and 11,349 (16.2%) with LOS > 7 days. Patients with longer LOS had more comorbidities and a higher severity of disease on admission. Overall models explained a modest amount of LOS variation, with an r(2) of 4.8%, with PC responsible for 1.3% of variation and together with SL explained 2.2% of variation. HC did not change the variation.
Conclusions: Based on admission vital signs and BNP levels, patients with longer LOS have more comorbidities and a higher disease severity. The ability to risk stratify for LOS based on patient admission and hospital characteristics is limited.
Copyright © 2011 Elsevier Inc. All rights reserved.