Risk factors for complications in patients undergoing pancreaticoduodenectomy: A NSQIP analysis with propensity score matching

Background Reports on the safety of minimally invasive pancreaticoduodenectomy compared to open pancreaticoduodenectomy (OPD) have demonstrated mixed results. One study comparing robotic pancreaticoduodenectomy (RPD) vs OPD demonstrated decreased complications associated with RPD. Objectives To evaluate the morbidity of RPD vs OPD using a national data set. Methods This is a retrospective cohort study from 2014 to 2017. Factors associated with complications in patients undergoing pancreaticoduodenectomy were evaluated using multivariate logistic regression (MVA) and propensity score matching (PSM). Results Of 13 110 PDs performed over the study period, 12 612 (96.2%) were OPD and 498 (3.8%) were RPD. Patients who underwent RPD vs OPD were less likely to have any complications (46.8% vs 53.3%; P = .004), surgical complications (42.6% vs 48.6%; P = .008), wound complications (6.2% vs 9.1%; P = .029), clinically relevant postoperative pancreatic fistulas (11.9% vs 15.6%; P = .026), sepsis (6.2% vs 9.3%; P = .019), and pneumonia (1.6% vs 3.8%; P = .012). On MVA, OPD was associated with increased complications compared with RPD. On PSM analysis, OPD remained a significant predictor for any (OR, 1.29; 95% CI, 1.03-1.61; P = .029) and surgical (OR, 1.26; 95% CI, 1.00-1.58; P = .048) complications. Conclusions This is the largest multicenter study to evaluate the impact of RPD on morbidity and suggests RPD is associated with decreased morbidity.