Rates and patterns of death after surgery in the United States, 1996 and 2006

Abstract

Background: Nationwide rates and patterns of death after surgery are unknown.

Methods: Using the Nationwide Inpatient Sample, we compared deaths within 30 days of admission for patients undergoing surgery in 1996 and 2006. International Classification of Diseases codes were used to identify 2,520 procedures for analysis. We examined the inpatient 30-day death rate for all procedures, procedures with the most deaths, high-risk cardiovascular and cancer procedures, and patients who suffered a recorded complication. We used logistic regression modeling to adjust 1996 mortality rates to the age and gender distributions for patients undergoing surgery in 2006.

Results: In 1996, there were 12,573,331 admissions with a surgical procedure (95% confidence interval [CI], 12,560,171-12,586,491) and 224,111 inpatient deaths within 30 days of admission (95% CI, 221,912-226,310). In 2006, there were 14,333,993 admissions with a surgical procedure (95% CI, 14,320,983-14,347,002) and 189,690 deaths (95% CI, 187,802-191,578). Inpatient 30-day mortality declined from 1.68% in 1996 to 1.32% in 2006 (P < .001). Of the 21 procedures with the most deaths in 1996, 15 had significant declines in adjusted mortality in 2006. Among these 15 procedures, 8 had significant declines in operative volume. The inpatient 30-day mortality rate for patients who suffered a complication decreased from 12.10% to 9.84% (P < .001).

Conclusion: Nationwide reporting on surgical mortality suggests that the number of inpatient deaths within 30 days of surgery has declined. Additional research to determine the underlying causes for decreased mortality is warranted.