Hospital teaching status and volume related to mortality after pancreatic cancer surgery in a national cohort

Abstract

Abstract Background The association between hospital teaching status and mortality after pancreatic resection is not well explored. Although hospital volume is related to short-term mortality, the effect on long-term survival needs investigation, taking into account hospital teaching status and selective referral patterns. Methods This was a nationwide retrospective register-based cohort study of patients undergoing pancreatic resection between 1990 and 2010. Follow-up for survival was carried out until 31 December 2011. The associations between hospital teaching status and annual hospital volume and short-, intermediate- and long-term mortality were determined by use of multivariable Cox regression models, which provided hazard ratios (HRs) with 95 per cent c.i. The analyses were mutually adjusted for hospital teaching status and volume, as well as for patients' sex, age, education, co-morbidity, type of resection, tumour site and histology, time interval, referral and hospital clustering. Results A total of 3298 patients were identified during the study interval. Hospital teaching status was associated with a decrease in overall mortality during the latest interval (years 2005–2010) (university versus non-university hospitals: HR 0·72, 95 per cent c.i. 0·56 to 0·91; P = 0·007). During all time periods, hospital teaching status was associated with decreased mortality more than 2 years after surgery (university versus non-university hospitals: HR 0·86, 0·75 to 0·98; P = 0·026). Lower annual hospital volume increased the risk of short-term mortality (HR for 3 or fewer compared with 4–6 pancreatic cancer resections annually: 1·60, 1·04 to 2·48; P = 0·034), but not long-term mortality. Sensitivity analyses with adjustment for tumour stage did not change the results. Conclusion Hospital teaching status was strongly related to decreased mortality in both the short and long term. This may relate to processes of care rather than volume per se. Very low-volume hospitals had the highest short-term mortality risk.