Relation between hospital surgical volume and outcome for pancreatic resection for neoplasm in a publicly funded health care system.

BACKGROUND Recent studies from the United States have shown that institutions with higher numbers of pancreatic resection procedures for neoplasm have lower mortality rates associated with this procedure. However, minimal work has been done to assess whether the results of similar volume-outcome studies within a publicly financed health care system would differ from those obtained in a mixed public-private health care system. METHODS A population-based retrospective analysis was used to examine pancreatic resection for neoplasm in Ontario for the period 1988/89 to 1994/95. Outcomes examined included in-hospital case fatality rate and mean length of stay in hospital. For each hospital, total procedure volume for the study period was defined as low (fewer than 22), medium (22-42) or high (more than 42). Regression models were used to measure volume-outcome relations. RESULTS The likelihood of postoperative death was higher in low-volume and medium-volume centres than in high-volume centres (odds ratio 5.1 and 4.5 respectively; p < 0.01 for both). Mean length of stay was greater in low- and medium-volume centres than in high-volume centres (by 7.7 and 9.2 days respectively, p < 0.01 for both). INTERPRETATION This study adds to growing evidence that, for pancreatic resection for neoplasm, patients may have better outcomes if they are treated in high-volume hospitals rather than low-volume hospitals.