Background: Length of postoperative stay (LOS) has gained increasing attention as a potential indicator of surgical efficiency. Our objective was to examine the feasibility of assessing LOS at 182 hospitals to identify institutions with outlying performance.
Methods: Patients were identified who underwent colorectal surgery at 182 hospitals participating in the American College of Surgeon's National Surgical Quality Improvement Program (ACS NSQIP) from 2006 to 2007. Regression models for extended LOS (greater than the 75th percentile) were developed to identify hospitals whose ratios of observed to expected events (O/E) were significantly better (low outlier) or worse (high outlier) than expected after adjustment for case mix. To evaluate strategies for evaluating LOS that would be minimally influenced by the occurrence of complications, separate models were developed for patients categorized either by (1) the nonoccurrence or occurrence of any postoperative complication or (2) tercile of preoperative morbidity risk.
Results: The 23,098 patients selected for this study were partitioned into groups without complications (0% complications), with complications (100%) or into terciles of preoperative morbidity risk (with 22.4%, 38.7%, and 60.0% of patients having complications, respectively). In general, the greater the complication rate the longer the LOS and the fewer the number of statistical outliers that were identified.
Conclusions: ACS NSQIP data can provide individual hospitals with riskadjusted LOS measures that can be used to identify outlying performance and motivate quality improvement efforts.