Hospital Procedural Volume Predicts Short-term Outcomes for Patients Undergoing Procedures Requiring Large-bore Access.

Abstract

Large-bore access procedures (LBP), such as transcatheter aortic valve replacement (TAVR), endovascular aneurysm repair (EVAR), and implantation of mechanical circulatory support (MCS), have been increasing in practice.1 While hospitals with a higher volume of TAVR procedures showed reduced in-hospital mortality and related complications,2 the relationship between procedural volume and outcomes for LBPs, collectively, has not been established, which is the goal of this research letter. This is a retrospective cohort study whereby all hospitalizations with patients undergoing LBP from 1 January 2016 to 31 December 2018, in the United States, were identified using the Nationwide Readmissions Database (NRD).3 LBPs included TAVR, EVAR, balloon aortic valvuloplasty (BAV), and implantation of MCS. Outcomes of interest were in-hospital mortality, vascular complications, major bleeding, and 30-day readmissions. Major bleeding was defined as bleeding requiring blood transfusion. The annualized hospital volume of LBP was analysed as both continuous and categorical variables (in quartiles) for the mentioned outcomes. Quartiles were chosen to ensure an equal number of hospitalizations in each volume category—quartile 1 (Q1), quartile (Q2), quartile 3 (Q3), and quartile 4 (Q4). Q1 was defined as a low-volume hospital and Q4 as a high-volume hospital. Outcomes were compared against Q1 by controlling for age, sex, comorbidities, and the Elixhauser Comorbidity Index, and the related odds ratio (OR) was estimated. Restricted cubic splines with hierarchical models were created to incorporate hospital characteristics as random effects in the model to help assess any nonlinear relationship between the hospital’s volume and outcomes. We also divided LBPs into TAVR and non-TAVR populations to determine if such outcomes also applied to the non-TAVR LBP. The present study did not require the institutional review board approval per institutional policy for de-identified publicly available dataset.