The volume-outcomes effect in hepato-pancreato-biliary surgery: hospital versus surgeon contributions and specificity of the relationship

Abstract

Background: Although the relationship between hepato-pancreato-biliary (HPB) procedure volume and outcomes is established, the relative importance of hospital and surgeon effects and the specificity of the volume-outcomes effect remain ill-defined. We sought to comprehensively characterize the hospital and surgeon volume-outcomes relationships in high-risk HPB surgery.

Study design: The 1998 to 2005 State Inpatient Databases for Florida, Maryland, and New York were used to identify patients undergoing complex HPB surgery and to quantify hospital and surgeon procedure volumes. The effects of hospital and surgeon procedure volumes on casemix-adjusted inpatient mortality were analyzed using multilevel logistic regression models.

Results: For hepatic resection, hospital procedure volume predicted mortality (high versus low volume, odds ratio [OR] 0.48, p=0.04), but surgeon volume did not (p=0.42). For pancreatic resection, in contrast, both hospital (OR 0.32, p < 0.001) and surgeon (OR 0.30, p < 0.001) procedure volume predicted mortality. The hospital volume effect for pancreatic resection was largely explained by surgeon volume. In both procedure groups, volume-outcomes effects were very specific. Only volumes of the primary procedure were predictive of mortality; volumes of related HPB procedures and overall HPB volume demonstrated no independent effect on mortality.

Conclusions: In HPB surgery, the relative contributions of hospital versus surgeon volume vary according to the specific procedure in question. In addition, the association between hospital or surgeon volume and in-hospital mortality is very specific to the procedure in question. High-volume expertise in one area of HPB surgery does not translate into improved outcomes for related procedures. These data may have implications for quality assessment and improvement, patient referral, and HPB surgical training.