Hospital-level variation in cardiac rehabilitation metrics

Background: To inform the delivery of cardiac rehabilitation (CR) care nationwide at the hospital level, we described hospital-level variation in CR metrics, overall and stratified by the hospital's tier of cardiac care provided.

Methods: This retrospective cohort analysis used Medicare fee-for-service (FFS) data (2018-2020), Parts A and B, and American Hospital Association (AHA) data (2018). We included beneficiaries with an acute myocardial infarction (AMI), percutaneous coronary intervention (PCI), or coronary artery bypass graft (CABG) in 2018, aged ≥65 years, and continuously enrolled in a FFS plan. We calculated hospital-level metrics for hospitals with ≥20 CR-qualifying events, which were identified using diagnostic/procedure codes. Claims for CR were identified by Healthcare Common Procedure Coding System (HCPCS) codes. We used multi-level models to examine patient- and hospital-level factors associated with CR metrics. Hospitals were stratified by tier of cardiac care provided (comprehensive, AMI/PCI, AMI-only care).

Results: Across the US, 2,212 hospitals treated individuals aged ≥65 years with a CR-qualifying event in 2018. By tier of cardiac care, 44.4% of hospitals provided comprehensive care, 31.2% provided AMI/PCI care, and 24.4% provided AMI-only care. Across all hospitals, there was substantial variation in CR enrollment (median 19.6%, interquartile range [IQR]=7.0%, 32.8%). Among hospitals with enrollment (n=1,866), median time to enrollment was 55.0 days (IQR=41.0, 71.0), median number of CR sessions was 26.0 (IQR=23.0, 29.0), and median percent completion was 26.0% (IQR=10.5%, 41.2%). There was also substantial variation in CR performance metrics among hospitals within each tier of cardiac care (e.g., median percent CR enrollment was 30.7% [IQR=20.7%-41.3%] among comprehensive care hospitals, 18.6% [IQR=9.5%-27.7%] among AMI/PCI hospitals, and 0.0% [IQR=0.0%-7.7%] among AMI-only hospitals). In adjusted analyses, characteristics associated with lower odds of CR enrollment included patient-level factors (older age, female sex, non-White race or ethnicity), and hospital-level factors (for-profit ownership, regions other than the Midwest, rural location, medium/large hospital size).

Conclusions: This is the first national, hospital-level analysis of CR metrics among Medicare beneficiaries. Substantial variation across hospitals, including peer hospitals within the same tier of cardiac care, indicates opportunities for hospital-level quality improvement strategies to improve CR referral and participation metrics.