Complications of Spinal Fusion: Comparing Medicare Transforming Episode Accountability (TEAM) Hospitals to Control Hospitals.

Abstract

Medicare's Transforming Episode Accountability Model (TEAM) holds hospitals accountable for 30-day post-operative complication across 18 categories of spinal fusion bundles, yet baseline differences in complication rates between participant and non-participant hospitals may confound policy evaluation and performance under this model. We compared complication rates for 18 TEAM-defined categories of fusion between participant and non-participant hospitals, and characterized patient- and hospital-level factors contributing to variation. Retrospective cohort study of Medicare beneficiaries undergoing inpatient or hospital outpatient lumbar and cervical fusion. Fee-for-service Medicare beneficiaries undergoing spinal fusion (2016-2021) were included. Excluded were Medicare Advantage, United Mine Workers, Maryland hospitals, and patients undergoing complex fusions (8+ vertebral levels, or fusion for primary diagnosis of spinal curvature, malignancy, infection). THIRTY DAY POST-DISCHARGE: all-cause readmission and post-operative complications (cardiac, vascular, infection, thromboembolic, cerebrovascular and device-related). We linked hospital TEAM participants to fee-for-service Medicare fusion claims and calculated 30-day complications by fusion category. Complication rates were estimated using multilevel logistic regression with a hospital-level random effects, adjusting for TEAM-specified covariates: fusion type, stratification, year, age, dual eligibility, social disability insurance, grouped total and select Hierarchical Conditions Classification comorbidity, bundled payment participation history, and the Census Division. An expanded model added patient- and hospital-level factors. Variation was quantified using hospital variance, intraclass correlation coefficients (ICC), and Coefficient of Variation (CV). TEAM participant and non-participant hospitals had similar 30-day readmission rates for inpatient (8.9%; OR 1.01; 95%CI 0.97-1.06; p=0.517) and outpatient (5.7%; OR 0.87; 95%CI 0.75-1.00; p=0.049) fusions, with comparable inpatient (28.2%; OR 0.97; 95%CI 0.92-1.03; p=0.367) and outpatient (10.4%; OR 0.97; 95%CI 0.86-1.10; p=0.645) composite complication rates. For readmission the hospital-level ICC was 1.06%, with TEAM adjusters explaining 61.6% of hospital variance, and reducing the CV from 27.6% to 17.1%. The composite complications, the ICC was 3.61%, with TEAM adjusters explaining 36.6% of variance, reducing the CV from 31.6% to 25.2%. Expanded covariates provided minimal incremental variance explanation. Postoperative complication rates were similar between TEAM participant and non-participant hospitals, and additional adjustments for surgical invasiveness, socioeconomic, patient, and hospital factors did not alter this finding. Between-hospital variance was small relative to case-level variance (∼94%), indicating complications are driven more by patient-level than institutional factors. Although TEAM adjusters substantially reduced the hospital variance, persistent wide variation in readmission and complication rates suggest a considerable quality gap remains.