Contemporary Improvements in Postoperative Mortality After Major Cancer Surgery are Associated with Weakening of the Volume-Outcome Association.

Abstract

Regionalization of complex visceral surgery across the United States has followed identification of a volume-outcome association. However, improvements in postoperative mortality overall during the last decade may have weakened the strength of this association. The National Cancer Database was used to identify patients undergoing colon, esophageal, liver, and pancreatic surgery from 2003 to 2011. Hospitals were divided into low-volume (< 33rd %tile), medium-volume (34-66th %tile), and high-volume (> 67th %tile) groups. Annual cancer-specific adjusted observed versus expected (O/E) ratios for 30- and 90-day mortality for each volume strata were calculated and plotted over time. In the year 2003, the O/E ratios decreased from low- to medium- to high-volume hospitals for all cancer surgeries for both 30- and 90-day mortality, indicating a strong volume-outcome relationship. For all volume strata, the O/E ratios trended downward from 2003 to 2011 for both 30- and 90-day mortality for all cancer surgeries. This trend was more pronounced for low- and medium-volume than for high-volume hospitals. Consequently, by 2011 the confidence intervals of the O/E ratios for the low-volume groups, and particularly for the medium-volume groups, overlapped those for the high-volume groups for most of the cancer surgeries studied. The volume-outcome association for major cancer surgery is dynamic and has attenuated over time primarily due to improvements in postoperative mortality at low- and medium-volume hospitals.