Variation in pancreatoduodenectomy as delivered in two national audits.

BACKGROUND: Nationwide audits facilitate quality and outcome assessment of pancreatoduodenectomy. Differences may exist between countries but studies comparing nationwide outcomes of pancreatoduodenectomy based on audits are lacking. This study aimed to compare the German and Dutch audits for external data validation. METHODS: Anonymized data from patients undergoing pancreatoduodenectomy between 2014 and 2016 were extracted from the German Society for General and Visceral Surgery StuDoQ|Pancreas and Dutch Pancreatic Cancer Audit, and compared using descriptive statistics. Univariable and multivariable risk analyses were undertaken. RESULTS: Overall, 4495 patients were included, 2489 in Germany and 2006 in the Netherlands. Adenocarcinoma was a more frequent indication for pancreatoduodenectomy in the Netherlands. German patients had worse ASA fitness grades, but Dutch patients had more pulmonary co-morbidity. Dutch patients underwent more minimally invasive surgery and venous resections, but fewer multivisceral resections. No difference was found in rates of grade B/C postoperative pancreatic fistula, grade C postpancreatectomy haemorrhage and in-hospital mortality. There was more centralization in the Netherlands (1·3 versus 13·3 per cent of pancreatoduodenectomies in very low-volume centres; P < 0·001). In multivariable analysis, both hospital stay (difference 2·49 (95 per cent c.i. 1·18 to 3·80) days) and risk of reoperation (odds ratio (OR) 1·55, 95 per cent c.i. 1·22 to 1·97) were higher in the German audit, whereas risk of postoperative pneumonia (OR 0·57, 0·37 to 0·88) and readmission (OR 0·38, 0·30 to 0·49) were lower. Several baseline and surgical characteristics, including hospital volume, but not country, predicted mortality. CONCLUSION: This comparison of the German and Dutch audits showed variation in case mix, surgical technique and centralization for pancreatoduodenectomy, but no difference in mortality and pancreas-specific complications.