S. P. Shubeck1,2,3, U. Nuliyalu3, J. B. Dimick1,3, H. Nathan1,3 1University Of Michigan,Department Of Surgery,Ann Arbor, MI, USA 2University Of Michigan,National Clinician Scholars Program,Ann Arbor, MI, USA 3University Of Michigan,Center For Healthcare Outcomes & Policy,Ann Arbor, MI, USA
Introduction: The Centers for Medicare and Medicaid Services have implemented many bundled payment programs focused on reducing the costs of specific surgical procedures or service lines. These bundled payment models hold hospitals accountable for the costs of entire episodes of surgical care, thereby encouraging efficiency and cost containment. However, little is known about whether hospital costs across different procedures are correlated. If so, hospital-wide efforts to improve efficiency might be useful; if not, these efforts would need to be targeted and service-specific. We therefore sought to determine the degree of cost correlation for surgical procedures within hospitals.
Methods: Using 100% Medicare claims data for 2010-2013, we identified patients aged 65-99 years undergoing elective surgical procedures including: colectomy, proctectomy, coronary artery bypass grafting (CABG), total hip replacement (THR), total knee replacement (TKR), esophagectomy, pancreatectomy, and abdominal aortic aneurysm repair (AAA). We calculated price-standardized, risk-adjusted Medicare payments for the entire “surgical episode” from the index admission through 30 days after discharge. The average cost for each procedure at each hospital was then calculated. We quantified cost associations between procedures for hospitals in the highest quintile of spending using Kappa statistics, ranging from 0-1, that take into account the possibility of agreement by chance alone.
Results: This study included 3530 hospitals performing colectomy, 2399 performing proctectomy, 1158 performing CABG, 3176 performing THR, 3390 performing TKR, 792 performing esophagectomy, 875 performing pancreatectomy, and 1645 performing AAA. Hospitals in the highest quintile of costs in one procedure were seldom high-cost in others. As expected, clinically unrelated procedures had weakly related costs (pancreatectomy and esophagectomy, K=0.12; proctectomy and TKR, K=0.06; esophagectomy and AAA, K=0.01, Figure 1A). Surprisingly, even some clinically similar procedures demonstrated only moderate cost relationship, such as colectomy and proctectomy (K=0.16, Figure 1B). The strongest relationship was found for THR and TKR, K=0.5 (Figure 1C).
Conclusion: We found that almost all procedures included in this study had weak cost relationships with other common procedures. The main exception to this finding was THR and TKR. Our findings suggest that broader inferences about hospital efficiency cannot be based on the institution’s performance in a single surgical procedure or service line. Additionally, initiatives to reduced surgical spending that are targeted at certain procedures are unlikely to have spillover effects on unrelated procedures in the same hospital.
