Background: Reducing length of stay (LOS) has been a priority for hospitals and health care systems. However, there is concern that this reduction may result in increased hospital readmissions.
Objective: To determine trends in hospital LOS and 30-day readmission rates for all medical diagnoses combined and 5 specific common diagnoses in the Veterans Health Administration.
Design: Observational study from 1997 to 2010.
Setting: All 129 acute care Veterans Affairs hospitals in the United States.
Patients: 4,124,907 medical admissions with subsamples of 2 chronic diagnoses (heart failure and chronic obstructive pulmonary disease) and 3 acute diagnoses (acute myocardial infarction, community-acquired pneumonia, and gastrointestinal hemorrhage).
Measurements: Unadjusted LOS and 30-day readmission rates with multivariable regression analyses to adjust for patient demographic characteristics, comorbid conditions, and admitting hospitals.
Results: For all medical diagnoses combined, risk-adjusted mean hospital LOS decreased by 1.46 days from 5.44 to 3.98 days, or 2% annually (P < 0.001). Reductions in LOS were also observed for the 5 specific common diagnoses, with greatest reductions for acute myocardial infarction (2.85 days) and community-acquired pneumonia (2.22 days). Over the 14 years, risk-adjusted 30-day readmission rates for all medical diagnoses combined decreased from 16.5% to 13.8% (P < 0.001). Reductions in readmissions were also observed for the 5 specific common diagnoses, with greatest reductions for acute myocardial infarction (22.6% to 19.8%) and chronic obstructive pulmonary disease (17.9% to 14.6%). All-cause mortality 90 days after admission was reduced by 3% annually. Of note, hospitals with mean risk-adjusted LOS that was lower than expected had a higher readmission rate, suggesting a modest tradeoff between hospital LOS and readmission (6% increase for each day lower than expected).
Limitations: This study is limited to the Veterans Health Administration system; non-Veterans Affairs admissions were not available. No measure of readmission preventability was used.
Conclusion: Veterans Affairs hospitals demonstrated simultaneous improvements in hospital LOS and readmissions over 14 years, suggesting that as LOS improved, hospital readmission did not increase. This is important because hospital readmission is being used as a quality indicator and may result in payment incentives. Future work should explore these relationships to see whether a tipping point exists for LOS reduction and hospital readmission.
Primary funding source: Office of Rural Health and the Health Services Research & Development Service, Veterans Health Administration, U.S. Department of Veterans Affairs.