115 Variation in Payments for Spine Surgery Episodes of Care

Abstract

Abstract INTRODUCTION Spine surgery is expensive and marked by high variation across regions and providers. Bundled payments have potential to reduce unwarranted spending associated with spine surgery. We sought to quantify variation in spine surgery payments, document sources of variation, and determine the influence of patient-level, surgeon-level, and hospital-level factors on payment variation. METHODS Cross-sectional analysis of commercial and Medicare claims data from January 2012 through March 2015 in the state of Michigan. Participants were adult patients who underwent spine-based surgical procedures. We used hierarchical regression models to analyze contributions of patient-level covariates and influence of individual surgeons and hospitals. Our primary outcome was price-standardized 90-day episode payments. We used intraclass correlation coefficients measures of variability accounted for by each level of a hierarchical model to quantify sources of spending variation. RESULTS >We analyzed 17,436 spine surgery episodes performed by 195 surgeons at 50 hospitals. Mean payment in highest cost quintile of episodes ($52,767; 95% CI, $52,133 to $53,401) significantly exceeded mean episode payment in lowest cost quintile ($9814; 95% CI, $9757 to $9872; P<.001). Facility payments for index admission and post-discharge payments were the greatest contributors to overall variation: 39.4% and 32.5%, respectively. After accounting for patient-level covariates, the remaining hospital-level and surgeon-level effects accounted for 2.0% (95% CI, 1.1% to 3.8%) and 4.0% (95% CI, 2.9% to 5.6%) of total variation, respectively. CONCLUSION Significant variation exists in total episode payments for spine surgery, driven mostly by variation in post-discharge and facility payments. Hospital and surgeon effects account for relatively little of observed variation. Our results suggest that alternative payment models applied to spine surgery including bundled payment will not be successful if their design assumes that surgeon and hospital preferences are the primary source of variation in costs.