Thais Reif, MD, Deborah S Keller, MS, MD, Guanying Yu, MD, Haiqing Zhang, MD, Laura Hyde, MD, Ahmed Al-Mazrou, MD, Ravi P Kiran, MD. Columbia University Medical Center
Background: New methods of improving surgical quality are needed. A key quality outcome measure is lengthof stay(LOS), as it’s linked to complications and costs. A novel method to reduce LOS is to apply Statistical Process Control(SPC) to identify process measures- specific institutional practices that lead to an outcome measure- related to outliers from the mean. SPC uses statistical methods to monitor and control a process, reducing variation and improving quality. Our goal was to determine if SPC can be used to identify outliers, patterns for outliers, and and their impact on quality outcome measures in elective colorectal surgery.
Methods: Review of a prospective divisional database was performedfor elective colorectal procedures using an abdominal approach from 1/1/13-5/1/2018. Univariate analysis was performed to identify outliers, and compare the demographic, operative, and outcome variables acrossoutliers and non-outliers. Control charts were used to display the data graphically and analyze the process performance of LOS over time. The upper and lower control limits were setat +/-1 standard deviation(SD) from the mean.
Results: 1,097 consecutive patients were evaluated. The overall mean LOS was 6.6 days(SD 4.4), giving control limits of 11 and 2. There were 115 LOS outliers. The mean outlier LOS was 21.4(SD 12.6). The process of LOS was unstable, with large deviations from the mean. However, the rate of outliers annually and their spread around the mean was consistent across years. Outliers had comparable age, gender, BMI, albumin, primary diagnosis, and procedure performed to non-outliers, but significantly greater co-morbidities and higher rates of previous abdominal surgery(both p<0.05). LOS outliers had significantly higher rates of intraoperative complications(23.5%), post-operative complications(81.7%), post-discharge complications(19.1%), unplanned reoperations(20.8%), readmissions(26.0%), and use of post-discharge nursing facilities(21.7%) than non-outliers(all p<0.05).
Conclusions: Using SPC, outliers for LOS and the trends over time were easily identified. The number of outliers and spread of variation above the mean LOS remained stable over time. The outliers had a significant impact on quality outcome measures, with higher rates of complications, readmission, reoperation, and healthcare utilization. Being an outlier may be a marker of a sicker patient. Regardless, there are consistent patterns for these LOS outliers, and the current “one-size fits all approach” may not be the best fit for these complex patients. With this, further work can focus on developing models to identify the process measures predictive of a prolonged LOS outcome measure, and proactively developing cost-effective care practices
Presented at the SAGES 2017 Annual Meeting in Houston, TX.
Abstract ID: 94997
Program Number: P338
Presentation Session: Poster Session (Non CME)
Presentation Type: Poster