Objective: The aim of the study was to evaluate the relationship between hospital volume and outcome after pancreaticoduodenectomy (PD).
Summary background data: Previous reviews for the hospital volume-outcome relationship after pancreatic resection were limited owing to clinical or methodological heterogeneity, resulting from differences in surgical procedures and high-volume hospital (HVH) definitions across studies.
Methods: We conducted a rigorous meta-analysis on the influence of hospital volume on various outcomes after PD using strict inclusion criteria and single cutoff values for HVHs.
Results: Thirteen studies based on nationwide databases from 11 countries, and including 58,023 patients in total, were included in this study. The overall pooled odds ratio (OR) for mortality favoring the HVH group was 2.37 [95% confidence interval (CI): 1.95-2.88] with high heterogeneity (I = 63%). We therefore classified all included studies into categories according to the cutoff values for HVH as defined in each individual study. The pooled OR for each category of 1 to 19, 20 to 29, and ≥30 PDs per year was 1.94, 2.34, and 4.05, respectively. There were significant differences among these categories (I = 58.9%, P = 0.09). The 2 former categories showed no statistical interstudy heterogeneities. The data did not suggest publication bias. These trends persisted in all subgroup analyses. Postoperative length of stay in the HVH group was significantly shorter with mild interstudy heterogeneity.
Conclusions: This meta-analysis included studies from different countries with disparate health care systems and provided strong evidence for an inverse association between higher hospital volume and lower mortality after PD. Variations in HVH cutoff values across studies majorly influenced the overall heterogeneity.