S. Mohanty1,2, J. Paruch1,3, K. Y. Bilimoria1,4, M. Cohen1, V. E. Strong5, S. M. Weber6 1American College Of Surgeons,Division Of Research And Optimal Patient Care,Chicago, IL, USA 2Henry Ford Hospital,Department Of Surgery,Detroit, MI, USA 3University Of Chicago Pritzker School Of Medicine,Department Of Surgery,Chicago, IL, USA 4Northwestern University Feinberg School Of Medicine,Department Of Surgery, Surgical Outcomes And Improvement Center,Chicago, IL, USA 5Memorial Sloan-Kettering Cancer Center,Department Of Surgery,New York, NY, USA 6University Of Wisconsin School Of Medicine And Public Health,Department Of Surgery,Madison, WI, USA
Introduction:
To allow fair comparisons of hospital quality, most risk adjustment approaches adjust for patient comorbidities and the primary procedure. However, secondary procedures done at the same time as the index case may increase operative risk and merit inclusion in adjustment models. Including such information could also improve individual patient risk prediction. Our objectives were to evaluate the impact of complexity adjustment on (1)postoperative outcomes, (2)model performance and (3)hospital rankings in gastric cancer surgery.
Methods:
Using 2007-2012 American College of Surgeons National Surgical Quality Improvement Program data, patients who underwent surgery for gastric adenocarcinoma were identified. Procedure complexity was characterized using secondary procedure CPT© codes and total work relative value units (RVUs). Regression models were developed to evaluate the association between complexity variables and outcomes. The influence of procedure complexity on model performance and hospital comparisons was examined.
Results:
Among 3,467 patients who underwent gastrectomy for adenocarcinoma, a secondary procedure was reported for 81.9% of total gastrectomies and 69.6% of partial gastrectomies. The presence of secondary procedures was associated with greater odds for adverse outcomes. For example, patients who underwent a synchronous bowel resection had a higher risk of mortality (OR=2.14, 95%CI: 1.07-4.29) and reoperation (OR=2.09, 95%CI: 1.26-3.47) (Table 1). Model performance was slightly better for nearly all outcomes with complexity adjustment (morbidity c-statistics: standard model, 0.690; RVU model, 0.694; secondary procedure model, 0.701). Hospital ranking did not change significantly after complexity adjustment (mortality, weighted κ 0.84).
Conclusion:
Surgical complexity adjustment improved individual risk prediction but did not appreciably affect hospital rankings. Inclusion of complexity variables into risk prediction tools, such as the ACS NSQIP Risk Calculator, should be considered.
