Development of a composite endpoint for randomized controlled trials in pancreaticoduodenectomy.

ABSTRACT

BACKGROUND: Few randomized controlled trials (RCTs) have been performed in patients undergoing pancreaticoduodenectomy (PD). An important factor contributing to this is the large number of patients needed to adequately power RCTs for relevant clinical single endpoints. A PD-specific composite endpoint (CEP) could solve this problem. The aim of the present study was to develop a PD-specific CEP, consisting of complications related to PD, allowing reduction in sample sizes and improving the ability to compare outcomes. METHODS: PD-specific CEP components were selected after a systematic review of the literature and consensus between 25 international pancreatic surgeons. Ultimately, prospective cohorts of patients who underwent PD in two high-volume HPB centers (London, UK, and Maastricht, NL) were used to assess the event rate and effect of implementing a PD-specific CEP. RESULTS: From a total of 18 single-component endpoints, intra-abdominal abscess, sepsis, post-PD hemorrhage, bile leakage, gastrojejunostomy leakage, leakage of the pancreatic anastomosis, delayed gastric emptying, and operative mortality within 90 days were selected to be included the PD-specific CEP. All eight components had consensus definitions and a Dindo-Clavien classification of 3 or more. The incidence of the PD-specific CEP was 24.7 % in the Maastricht cohort and 23.3 % in the London cohort. These incidence rates led to a twofold reduction in the theoretical calculated sample size for an adequately powered RCT on PD using this CEP as a primary endpoint. CONCLUSIONS: The proposed PD-specific CEP enables clinical investigators to adequately power RCTs on PD and increases the feasibility, comparability, and utility in meta-analysis.