Practice pattern variability for myocardial revascularization: Impact on resource use across 24 centers

Objective: To determine the predictors of hospital length of stay (LOS) after elective uncomplicated coronary artery bypass graft surgery. Design: Retrospective analysis of the EPI-1 database, 1991-1993. Setting: Multicenter; 24 academic, private, federal, and health maintenance institutions. Participants: Patients undergoing elective CABG surgery (n = 2,417). Measurements and Main Results: Using a systematic sampling scheme at each site, each patient was evaluated to identify markers of chronic disease, perioperative test data, treatments, adverse outcomes, and LOS. Institutional differences in the care of patients free of complications were assessed using a multivariate model. LOS was the outcome variable selected to estimate cost. A total of 861 patients (37%) were free of any complication. The mean site LOS ranged from 5.4 to 9.0 days, with half of the 24 centers reporting a hospital LOS routinely >7.1 days. The predominant factor associated with a complication-free LOS was site per se, accounting for 32% of the variability in hospital LOS that could not be explained by any site characteristic (eg, size, geographic location, academic affiliation). Multivariable analysis identified 3 demographic predictors - age >75 years (increasing LOS by 1.3 days), admission from the emergency department (increasing LOS by 0.7 days), and uninsured or Medicaid-insured (increasing LOS by 0.4 days); 2 historical predictors - New York Heart Association class III or IV congestive heart failure (increasing LOS by 0.5 days) and history of arrhythmia (increasing LOS by 0.7 days); and 2 practice patterns - transfusion of blood products (increasing LOS by 0.3 days) and delayed extubation (increasing LOS by 0.5 days). Previous myocardial infarction, diabetes, chronic obstructive pulmonary disease, neurologic disease, and other historical factors were not associated with LOS in patients without a complication. Conclusion: A substantial variability in LOS after complication-free coronary artery bypass graft surgery was determined predominantly by site per se, even after adjustment for disease severity, site type or location, and surgical and anesthetic practices. The variability in LOS was likely due to practice style influences and represents an opportunity to decrease waste in the provision of a common and expensive procedure.