Hospital Volume Thresholds in Reverse Shoulder Arthroplasty: A Nationwide Analysis
Kevin Xavier Farley, BS1, Andrew Schwartz, MD2, Charles A Daly, MD1, Michael Brandon Gottschalk, MD1 and Eric R Wagner, MD, MS2, (1)Emory University, Atlanta, GA, (2)Emory University, Department of Orthopaedics, Atlanta, GA
INTRODUCTION: With increasing utilization of the reverse total shoulder arthroplasty (RSA), it is becoming increasingly critical to assess the association of procedure-specific hospital volume and outcomes after RSA. Our objective was to use stratum-specific likelihood ratio (SSLR) analysis to delineate concrete definitions of volume.
METHODS: The Nationwide Readmission Database was queried from the years 2011 through 2015 for patients undergoing elective RSA. Annual hospital volume and outcome data was collected, including 90-day readmission, 90-day revision, 90-day complications, length of stay (LOS,>2 days considered “extended”), median cost ($18,452), and discharge disposition. SSLR analysis was performed to determine hospital volume cut-offs associated with increased risks for each outcome. Cut-offs generated through SSLR analysis were confirmed via binomial logistic regression. These cut-offs were then compared to traditional quartile analysis (≤14, 15-31, 32-57, ≥58 procedures/year).
RESULTS: SSLR analysis produced hospital volume cut-offs for each outcome studied, with higher volume centers having improved postoperative results. The suggested cut-offs for optimal 90-day outcomes ranged from 54 to 70 cases per year, while the cut-offs for optimal cost and resource utilization were slightly higher. SSLR analysis for 90-day readmission produced three hospital volume categories, each statistically different from each other (≤16, 17-69, ≥70). These were similar to the strata for 90-day revision (≤16, 17-53, ≥54) and 90-day complications (≤9, 10-68, ≥69). SSLR analysis produced 6 hospital volume categories for cost of care over the median value (≤5, 6-25, 26-47, 48-71, 72-105, ≥106), 5 categories for an extended LOS (≤10, 11-25, 26-59, 60-105, ≥106), and 4 categories for non-home discharge (≤31, 32-80, 81-105, ≥106). SSLR analysis resulted in a better model fit for all outcomes assessed (7.2% to 18.6% improvement in AIC values compared to quartile analysis). Proportions for an adverse event for each volume category are displayed in the figure. The proportion of patients receiving care at centers that performed <20 procedures/year decreased from 2011 (46.4%) to 2015 (26.8%), while the proportion of patients that received care at a center that performed >100 procedures/year increased from 2011 (3.3%) to 2015 (13.3%).
DISCUSSION: As the incidence of RSA in the United States rises, the need for evidence driven volume cut-offs is critical to promote cost-effective and high-quality care. We have defined concrete hospital surgical volumes (~60-70 cases/year) that maximize outcomes after RSA. This information may be used in future policy and administrative decisions to consolidate complex procedures, such as RSA, at high-volume centers due to their superior outcomes.
Back to 2020 Abstracts