Impact of nationwide centralization of pancreaticoduodenectomy on hospital mortality

Abstract

Abstract Background The impact of nationwide centralization of pancreaticoduodenectomy (PD) on mortality is largely unknown. The aim of this study was to analyse changes in hospital volumes and in-hospital mortality after PD in the Netherlands between 2004 and 2009. Methods Nationwide data on International Classification of Diseases, ninth revision (ICD-9) code 5-526 (PD, including Whipple), patient age, sex and mortality were retrieved from the independent nationwide KiwaPrismant registry. Based on established cut-off points of annually performed PDs, hospitals were categorized as very low (fewer than 5), low (5–10), medium (11–19) or high (at least 20) volume. A subgroup analysis based on a cut-off age of 70 years was also performed. Results Some 2155 PDs were included. The number of hospitals performing PD decreased from 48 in 2004 to 30 in 2009 (P = 0·011). In these specific years, the proportion of patients undergoing PD in a medium- or high-volume centre increased from 52·9 to 91·2 per cent (P < 0·001). Nationwide mortality rates after PD decreased from 9·8 to 5·1 per cent (P = 0·044). The mortality rate during the 6-year period was 14·7, 9·8, 6·3 and 3·3 per cent in very low-, low-, medium- and high-volume hospitals respectively (P < 0·001). The difference in mortality between medium- and high-volume centres was statistically significant (P = 0·004). The volume–outcome relationship was not influenced by age (P = 0·467). The mortality rate after PD in patients aged at least 70 years was 10·4 per cent compared with 4·4 per cent in younger patients (P < 0·001). Conclusion With nationwide centralization of PD, the in-hospital mortality rate after this procedure decreased. Further centralization of PD is likely to decrease mortality further, especially in the elderly.