Abstract 5918: Length of Stay for Acute Heart Failure: Effects of Cohorting Patients on Variability

Abstract

Background: Quality metrics for heart failure (HF) are increasingly utilized as an indicator of quality care. Mean length of stay (LOS) is a commonly utilized metric that is made available to the public but variability in LOS is not reported. Efforts to cohort HF pts in the hospital improve quality of care but the effect on LOS is less well defined. Methods: As part of an ongoing performance improvement project at ChristianaCare, LOS was evaluated using standard descriptive statistics. A LOS distribution was also generated monthly to visualize LOS variability. Comparisons were made for HF pts cohorted on a dedicated HF unit (HFU) vs. all other units (Other) and by inclusion/exclusion of HF pts with prolonged LOS > 10 days to reduce the effects of outliers. Results: 71 % of all HF patients were discharged in < 5 days, 82 % in < 10 days. Mean LOS (March 07–March 08) for HF was 6.63 days overall (6.1 HFU vs. 7.2 other; p = 0.05). After exclusion of long LOS HF pts, mean LOS fell to 4.4 days (4.3 HFU vs. 4.5 Other; p = 0.06). Median LOS for all HF patients was 5.0 days regardless of location. The standard deviation of LOS was significantly lower for HFU patients (4.5 vs. 11.5 days, p < 0.02) The LOS graph (Graph 1) highlights the wide variability in LOS and how mean statistics may be misleading. Conclusions: Mean LOS can be misleading as a metric for quality of HF care provided as outliers significantly impact this value. Other statistics and graphical displays can more accurately reflect quality of care. In addition, cohorting of HF patients significantly reduces overall variability in LOS and suggests a more consistent approach to acute HF care.