Pancreaticobiliary endoscopic ultrasound in England 2007 to 2016: Changing practice and outcomes

Abstract

Background and study aims Population-level data on the outcomes of pancreaticobiliary endoscopic ultrasound (PB-EUS) are limited. We examined national PB-EUS and fine-needle aspiration (FNA) activity, its relation to pancreatic cancer therapy, associated mortality and adverse events. Patients and methods Adults undergoing PB-EUS in England from 2007–2016 were identified in Hospital Episode Statistics. A pancreatic cancer cohort diagnosed within 6 months of PB-EUS were studied separately. Multivariable logistic regression models examined associations with 30-day mortality and therapies for pancreatic cancer. Results 79,269 PB-EUS in 68,908 subjects were identified. Annual numbers increased from 2,874 (28 % FNA) to 12,752 (35 % FNA) from 2007 to 2016. 8,840 subjects (13 %) were diagnosed with pancreatic cancer. Sedation related adverse events were coded in 0.5 % and emergency admission with acute pancreatitis in 0.2 % within 48 hours of PB-EUS. 1.5 % of subjects died within 30 days of PB-EUS. Factors associated with 30-day mortality included increasing age (odds ratio 1.03 [95 % CI 1.03–1.04]); male sex (1.38 [1.24–1.56]); increasing comorbidity (1.49 [1.27–1.74]); EUS-FNA (2.26 [1.98–2.57]); pancreatic cancer (1.39 [1.19–1.62]); increasing deprivation (least deprived quintile 0.76 [0.62–0.93]) and lower provider PB-EUS volume (2.83 [2.15–3.73]). Factors associated with surgical resection in the pancreatic cancer cohort included lower provider PB-EUS volume (0.44 [0.26–0.74]) and the least deprived subjects (1.33 [1.12–1.57]). 33 % of pancreatic cancer subjects who underwent EUS, did not subsequently receive active cancer treatment. Conclusions Lower provider PB-EUS volume was associated with higher 30-day mortality and reduced rates of both pancreatic cancer surgery and chemotherapy. These results suggest potential issues with case selection in lower-volume EUS providers.