In-hospital survival after pancreatoduodenectomy is greater in high-volume hospitals versus lower-volume hospitals: a meta-analysis.

Abstract

Variation in cut-off values for what is considered a high volume (HV) hospital has made assessments of volume-outcome relationships for pancreaticoduodenectomy (PD) challenging. Accordingly, we performed a systematic review and meta-analysis comparing in-hospital mortality after PD in hospitals above and below HV thresholds of various cut-off values. PubMed/MEDLINE, Embase and Cochrane Library were searched to 4 January 2021 for studies comparing in-hospital mortality after PD in hospitals above and below defined HV thresholds. After data extraction, risk of bias was assessed using the Downs and Black checklist. A random-effects model was used for meta-analysis, including meta-regressions. Registration: PROSPERO, CRD42021224432. From 1855 records, 17 observational studies of moderate quality were included. Median HV cut-off was 25 PDs/year (IQR: 20-32). Overall relative risk of in-hospital mortality was 0.37 (95% CI: 0.30, 0.45), that is, 63% less in HV hospitals. All subgroup analyses found an in-hospital survival benefit in performing PDs at HV hospitals. Meta-regressions from included studies found no statistically significant associations between relative risk of in-hospital mortality and region (USA vs. non-USA; p=0.396); or 25th percentile (p=0.231), median (p=0.822) or 75th percentile (p=0.469) HV cut-off values. Significant inverse relationships were found between PD hospital volume and other outcomes. In-hospital survival was significantly greater for patients undergoing PDs at HV hospitals, regardless of HV cut-off value or region. Future research is required to investigate regions where low-volume centres have specialized PD infrastructure and the potential impact on mortality.