Objective: To analyse the volume-outcome effect of pancreatic surgery by means of a systematic review, and to determine the effect of the ongoing plea for centralisation of pylorus-preserving pancreaticoduodenectomy in the Netherlands.
Design: Systematic review and retrospective evaluation.
Method: A systematic search for studies comparing hospital mortality rates after pancreatic resection in high- and low-volume hospitals was conducted. The studies were independently assessed regarding design, inclusion criteria, threshold value for high and low volume and primary hospital mortality outcome. Data were obtained from the Dutch nation-wide registry on the mortality outcome of pancreaticoduodenectomy in 1994-2003. Hospitals were divided into 4 categories based on the number of pancreaticoduodenectomies performed. The effect of the ongoing plea for centralisation was analysed.
Results: Twelve observational studies comprising a total of 19,688 patients were included. Because the studies were too heterogeneous to allow a meta-analysis, a qualitative analysis was performed. The relative risk of dying in a high-volume hospital compared with a low-volume hospital was between 0.07 and 0.76 and was inversely proportional to the arbitrarily defined volume cut-off values. Various analyses conducted over a to-year period in the Netherlands reported mortality rates of 14-17% in hospitals that performed fewer than 5 pancreaticoduodenectomies per year, compared with rates of 0.0-3.50 degrees h in hospitals that performed more than 24 pancreaticoduodenectomies per year. The percentage of patients undergoing surgery in hospitals with a volume less than ro pancreaticoduodenectomies per year was 57% in 2000-2003 (454/792), compared with 65% (280/428) in 1994-1995.
Conclusion: This systematic review provided evidence of an inverse relationship between hospital volume and mortality after pancreaticoduodenectomy and confirmed the value of centralisation of this procedure in high-volume hospitals. The 10-year-long plea of the Dutch surgical community for quality assessment and, if necessary, centralisation has not resulted in a reduction in mortality rates after pancreatic resection or a change in referral patterns in The Netherlands.