National Surgical Quality Improvement Program and Trauma Quality Improvement Program have shown variations in risk-adjusted outcomes across hospitals. Our study hypothesis was that there would be similar variation in risk-adjusted outcomes of emergency general surgery (EGS) patients.METHODS
We undertook a retrospective analysis of the National Inpatient Sample database for 2010. Patients with EGS diseases were identified using American Association for the Surgery of Trauma definitions. A hierarchical logistic regression model was used to model in-hospital mortality, accounting for patient characteristics, including age, sex, race, ethnicity, insurance type, and comorbidities. Predicted-to-expected mortality ratios with 90% confidence intervals were used to identify hospitals as low mortality (ratio significantly lower than 1), high mortality (ratio significantly higher than 1), or average mortality (ratio overlapping 1).RESULTS
Nationwide, 2,640,725 patients with EGS diseases were treated at 943 hospitals in 2010. About one-sixth of the hospitals (139, 15%) were low mortality, a quarter were high mortality (221, 23%), and the rest were average mortality. Mortality ratio at low mortality hospitals was almost four times lower than that of high mortality hospitals (0.57 vs. 2.03, p < 0.0001). If high and average mortality hospitals performed at the same level as low mortality hospitals, we estimate 16,812 (55%) more deaths than expected.CONCLUSION
There are significant variations in risk-adjusted outcomes of EGS patients across hospitals, with several thousand higher than expected number of deaths nationwide. Based on the success of National Surgical Quality Improvement Program and Trauma Quality Improvement Program, we recommend establishing EGS quality improvement program for risk-adjusted benchmarking of hospitals for EGS patients.LEVEL OF EVIDENCE
Care management, level III.