C. Nguyen1, L. Moradi1, L. Wood1, A. Liwo1, D. Chu1, G. Kennedy1, M. Morris1 1University Of Alabama at Birmingham,Birmingham, Alabama, USA
Introduction: Predicted length of stay (pLOS) is a measurement used by patients, physicians, and hospitals to assess quality of care and efficiency. The American College of Surgeons’ National Surgical Quality Improvement Program (ACS NSQIP) and Vizient both predict a patient’s length of hospital stay; however, the ACS NSQIP Risk Calculator uses a model developed from patient health information gathered by trained NSQIP nurses whereas Vizient uses billing and record data to predict length of stay. In this study, we aim to compare the pLOS from the ACS NSQIP risk calculator and Vizient database to the actual length of stay (aLOS) of colorectal patients at a single institution.
Methods: We queried our institutional NSQIP database for all patients undergoing colorectal surgery between Jan 2015 and Feb 2019. Patients were stratified into those with and without postoperative, pre-discharge NSQIP measured complications. Predicted length of stay was calculated using the ACS NSQIP Risk Calculator and Vizient’s database. Wilcoxon tests were used to compare median pLOS and aLOS. All tests were conducted at an alpha level of 0.05.
Results: Of the 1,267 colorectal patients reviewed, 906 had colectomies and 361 had proctectomies. The cohort had a median age of 59 and 52.1% was female. Significant differences were found between the median pLOS for NSQIP and Vizient for colorectal surgeries overall (NSQIP 6.5 vs. Vizient 5.7 days; p<0.001), for colectomies (NSQIP 6 vs. Vizient 5.6 days; p=0.02), and for proctectomies (NSQIP 7 vs. Vizient 5.9 days; p<0.001). Patients without in-hospital postoperative complications had an aLOS that was significantly different from the pLOS for both NSQIP and Vizient (Table 1). However, aLOS and pLOS were similar for patients experiencing pre-discharge NSQIP measured complications (Table 1).
Conclusion: Overall, NSQIP and Vizient were able to predict LOS in patients experiencing a pre-discharge NSQIP measured complication. However, they both overestimated LOS in patients who did not experience a complication. Future work should investigate this relationship further.
