Impact of hospital volume on outcomes in acute pancreatitis: a study using a nationwide administrative database

Background Although several population-based studies have shown higher hospital volume (HV) to be associated with better outcomes in acute pancreatitis, they failed to adjust for disease severity and did not take into account the potentially non-linear relationship between HV and outcomes. Using a Japanese nationwide administrative database, this study aimed to evaluate the volume-outcome relationship in acute pancreatitis by means of statistical methods that permitted such considerations. Methods In-hospital mortality, length of stay, and total costs for patients with acute pancreatitis were analyzed using multivariate regression models fitted with generalized estimating equations. Adjustment for severity was based on the Japanese Severity Scoring System and other patient characteristics. We used restricted cubic spline functions to examine the potential non-linear relationships between HV and outcomes. Results In all, 17,415 eligible patients with acute pancreatitis were identified from 1,032 hospitals between 1 July 2010 and 30 September 2011. The in-hospital mortality rate was 2.6 %, and the median total costs were US $7,740 (interquartile range, 5,150-11,920). The overall and non-linear volume-outcome relationships were not significant either for in-hospital mortality or total costs. The median length of stay was 14 days (interquartile range, 10-22), and high HV was positively associated with shorter hospitalization (overall, P < 0.001; non-linear, P = 0.194). Conclusions Despite the shorter hospitalization with higher HV, no inverse volume-outcome relationship was evident for acute pancreatitis. Further evidence is required to justify the volume-based selective referral of acute pancreatitis patients.