High Volume Hospitals Are Not Synonymous with High Quality: A National Analysis of Aortic and Mitral Valve Procedures

S. Sakowitz¹, S. Bakhtiyar², N. Chervu¹, A. Vadlakonda³, K. Ali⁴, P. Benharash^5
1UCLA David Geffen School of Medicine, Los Angeles, California ²University of California Los Angeles UCLA, Los Angeles, California ³David Geffen School of Medicine at UCLA, Saratoga, California ⁴UCLA, Los Angeles, California ⁵UCLA Division of Cardiac Surgery, Los Angeles, California

Purpose: The volume-outcome relationship has been well-established across surgical fields.[1-2] Yet, with improvements in benchmarking, the association of center-level operative volume and clinical quality in cardiac surgery has been questioned. In the current era of quality-based reimbursements, we sought to reassess whether hospital volume accurately reflects quality in valve operations.

Methods: Patients >= 18 years receiving surgical aortic valve replacement (SAVR), mitral valve repair (MVV), or mitral valve replacement (MVR) with/without coronary artery bypass grafting (CABG) were identified in the 2016-2020 Nationwide Readmissions Database.
Center-level annual volume was calculated for each procedure. Hierarchical mixed-effects models were developed to determine risk-adjusted rates of in-hospital mortality attributable to center-level effects (PROM), with hospital identifiers as the random effect. Bayesian methodology was applied to predict intercepts and hospitals were ranked by increasing center-specific PROM. Spearman rank correlation coefficients with Bonferroni’s adjustment were utilized to assess correlations between annual hospital volume and PROM.
Hospitals were categorized into terciles based on procedure-specific annual volume (top tertile: High Volume Hospital [HVH]) as well as PROM. We then assessed the extent of center misclassification by hospital volume and PROM, with misclassification considered the proportion of HVH in the highest PROM tertile.

Results: In total, an estimated 169,760 SAVR, 57,213 MVR, and 57,575 MVV procedures were performed at a mean of 752 hospitals between 2016-2020. Median annual volumes were 106 for SAVR (interquartile range [IQR] 54-192), 53 (IQR 28-94) for SAVR-CABG, 30 for MVR (IQR 14-53), and 42 (IQR 18-90) for MVV.
Weak inverse relationships were identified between associated risk-adjusted mortality and isolated SAVR volume (r -0.21, P< 0.001), SAVR-CABG volume (r =-0.18, P< 0.001), and MVR volume (r -0.23, P< 0.001). A strong relationship was found between MVV volume and PROM (r -0.52, P< 0.001).
We subsequently evaluated mismatch between high-volume centers and PROM for each procedure. Of SAVR HVH, 41% were found to be in the highest PROM tercile. Similarly, of SAVR-CABG HVH, 42% were in the highest PROM group. Evaluating MVR and MVV HVH, 38% and 59% were in the highest PROM terciles, respectively. Altogether, if volume alone was utilized to classify centers, 33% of all hospitals would be misclassified when considering performance based on PROM.

Conclusion: Hospital SAVR and MVR volumes are not accurate surrogates for high-quality, such that many centers would be misclassified in their performance. Therefore, centralization of care to high-volume centers may limit access without ensuring high-quality. Rather than volume alone, comprehensive, risk-adjusted measures of hospital outcomes that incorporate other metrics are warranted.[3-4]

Identify the source of the funding for this research project: None