Impact of Operative Time on Outcomes after Pancreatic Resection: A Risk-Adjusted Analysis Using the American College of Surgeons NSQIP Database

BACKGROUND: Longer operative time (OT) has been associated with negative outcomes in various surgical procedures, but its role in pancreatic resection, a complex, high-acuity endeavor, is not yet well defined. The aim of this study was to analyze the relationship between OT and pancreatectomy outcomes in a risk-adjusted fashion. STUDY DESIGN: This retrospective cohort study analyzed patients undergoing pancreaticoduodenectomy (PD) or distal pancreatectomy (DP) between 2014 and 2015 using the procedure-targeted pancreatectomy database of the American College of Surgeons NSQIP. Univariable analyses and multiple backward stepwise conditional logistic regression models were used to assess the impact of OT on postoperative occurrences. RESULTS: Among 10,157 patients, 6,844 PDs and 3,313 DPs were performed. Median operative time was 358 minutes (interquartile range 282 to 444 minutes) for PD and 213 minutes (interquartile range 157 to 285 minutes) for DP. Male sex, younger age, obesity, neoadjuvant treatment, minimally invasive approaches, and vascular/concurrent organ resections were associated with longer OT for both procedures. Morbidity increased in a stepwise manner with increasing OT. After risk adjustment, increasing OT was negatively associated with overall morbidity, major complications, pancreatectomy-specific complications, infectious complications, and prolonged hospital stay. These associations were independent from patients' preoperative characteristics, operative approach, vascular or concurrent organ resection, and postoperative diagnosis. These findings held true for both PD and DP. Conversely, the association between OT and mortality was mainly driven by the excessive operative durations for PDs, and was not significant for DPs. CONCLUSIONS: Longer OT is independently associated with worse perioperative outcomes after pancreatic resection, and should be considered a relevant parameter in risk-adjustment processes for outcomes evaluation. These findings suggest possible areas of quality improvement through individual and system-level initiatives.