The relationship between hospital volume and post-operative mortality rates for upper gastrointestinal cancer resections: Scotland 1982–2003

Abstract Background Centralisation of surgical treatment of cancer has resulted in improved outcomes. We aimed to determine evidence of benefit for specialised management of upper gastrointestinal cancer in high-volume centres in Scotland. Methods Discharge records of patients undergoing oesophagectomy, gastrectomy, hepatectomy or pancreatectomy between 1982 and 2003 were identified. Hospital data were analysed on a year-by-year basis to derive ‘hospital-years’. Hospital-years were divided into quartiles by volume, and were analysed with regard to in-hospital mortality during the operative admission [Chi-square test ( χ2 ) and Chi-square test for trend ( χ2trend )]. Results 10,625 patients and 982 in-hospital deaths were included. In-hospital mortality rates declined during the study period: oesophagectomy 11.7–7.9%; gastrectomy 11.2–7.2%; hepatectomy 11.1–3.0%; and pancreatectomy 8.3–4.9%. For all resections except gastrectomy, mortality decreased as quartile of hospital-year volume increased (oesophagectomy: χ2 p = 0.006, χ2trend p = 0.001; hepatectomy: χ2 p = 0.004, χ2trend p = 0.003; pancreatectomy: χ2 p = 0.002, χ2trend p = 0.001). ORs of death were lower for oesophagectomy (OR = 0.58; 95%CI = 0.39, 0.88; p = 0.009) and pancreatectomy (OR = 0.35; 95%CI = 0.19, 0.64; p < 0.001) in hospital-years within highest-volume quartiles compared with lowest. Scattergraphs of all resection types demonstrated inverse power relationships between number of resections per hospital-year and mortality. Conclusion Concentration of cancer care has had major effects on health service delivery in Scotland. Centralisation should be supported in surgical management of upper gastrointestinal cancer.