Hypothesis: Decreased length of stay (LOS) after pancreatoduodenectomy is due to multiple factors, including a lower complication rate and more efficient perioperative care for all patients, with and without complications.
Design: A retrospective review, validation cohort.
Setting: A single university hospital referral center.
Patients: A consecutive sample of patients undergoing pancreatoduodenectomy from January 9, 1986, to December 21, 1992 (group 1 [n = 104]) and from February 16, 1993, to November 9, 1998 (group 2 [n = 111]).
Intervention: Mann-Whitney test and linear [correction of logistic] regression analysis applied to clinical variables and LOS.
Main outcome measures: Difference in median LOS between early and late groups and identification of factors predictive of decreased LOS.
Results: Total LOS decreased between the 2 groups (26 days [range, 13-117 days] vs 15 days [range, 5-61 days]; P<.001), with a decrease in preoperative (4 days [range, 0-28 days] vs 2 days [range, 0-36 days]; P<.001) and postoperative (19 days [range, 11-95 days] vs 12 days [range, 4-58 days]; P<.001) LOS (data given for group 1 vs group 2). Major complications decreased from 49% in group 1 to 25% in group 2 (P<.001). Postoperative LOS decreased for patients with (25 days [range, 15-95 days] vs 20 days [range, 8-58 days]; P = .05) and without (15 days [range, 11-47 days] vs 11 days [range, 4-55 days]; P<.001) major complications (data given for group 1 vs group 2). Multivariate analysis identified age (P = .01), pancreatic fistula (P<.001), delayed gastric emptying (P<.001), biliary complications (P<.001), operative time (P<.005), extra-abdominal infection (P<.005), use of a percutaneous stent (P = .04), and year of operation (P<.001) as independent predictors of total LOS.
Conclusion: A reduction in complications in combination with factors leading to a streamlining of perioperative care has contributed to the decreased LOS after pancreatoduodenectomy.