Effect of centralization of pancreaticoduodenectomy on nationwide hospital mortality and length of stay.

Background: Despite the persistence of large differences in operative mortality rates between centres, the value of centralization of pancreaticoduodenectomy (PD) remains under debate. This cohort study analysed the effect of centralization of PD on nationwide hospital mortality and length of hospital stay in Belgium. Methods: Data on in-hospital mortality and duration of hospital stay after PD from 2000 to 2004 were obtained from the Belgian national registry database. Analysis of mortality and hospital stay was based on 1842 PDs from all 126 hospitals. Logistic regression analysis was used to assess the effect of patient referral on the national mortality rate. Results: The national mortality rate was 8·4 per cent and the median duration of hospital stay after operation was 21·6 (range 3–117) days. There was a significant relationship between the annual number of PDs per hospital and both mortality rate (P = 0·005) and hospital stay (P = 0·027). Application of a cut-off volume of ten PDs per year per centre would necessitate 56·8 per cent of all patients being referred, resulting in an expected national mortality rate of 6·0 per cent. Conclusion: Referral of patients to more experienced centres for PD is expected to result in a significant reduction in hospital mortality rate and duration of hospital stay, regardless of the experience of the referring centre. Action towards centralization should be undertaken nationwide. Copyright © 2007 British Journal of Surgery Society Ltd. Published by John Wiley & Sons, Ltd.