Abstract 12833: Hospital Variation in Bleeding Complications Following Percutaneous Coronary Intervention (PCI): Results from the National Cardiovascular Data Registry (NCDR)
Background: There is increasing interest in benchmarking hospital PCI outcomes. We sought to determine: a) site-level variation in post-PCI bleeding rates; and b) the degree to which hospital bleeding rates and ranks were altered by patient case-mix adjustment.
Methods: We examined bleeding complications rates among 1,036 NCDR participating hospitals performing >50 PCI cases from 2/08 to 10/10 (n= 729,164 PCI procedures). Post-PCI bleeding was defined as any of the following: intracranial hemorrhage, cardiac tamponade, non-CABG related blood transfusion (if pre-procedure hemoglobin was ≥ 8 g/dl), or an absolute decrease in hemoglobin of ≥ 3g/dl (if pre-procedure hemoglobin ≤ 16 g/dl). Hospital bleeding rates and ranks were compared with and without adjustment for 19 clinical factors using the NCDR validated bleeding model (C-index 0.77). High- or low-hospital outliers were identified using unadjusted and risk-adjusted bleeding using hierarchical models.
Results: Overall, unadjusted hospital post-PCI bleeding rates varied from 2.9%, 4.1%, 5.9%, 7.8%, 10.0% (for the 10th, 25th,50th,75th, and 90th percentiles). Center-level variation in bleeding persisted after case-mix adjustment (from 2.97% to 8.82% for the 10th to 90th percentiles). While hospital observed and adjusted bleeding ranks were correlated (Figure), individual ranks changed significantly with adjustment: median (25th, 75th) rank change = 68 (23, 136) places. High- and low-hospital ‘outliers’ (based on 95% CI) also shifted with risk-adjustment; with 27% of high- and 33% of low-outliers changing designation after adjustment.
Conclusion: There is wide variation in post-PCI bleeding rates among hospitals that are only partially accounted for by patient case-mix. Providing sites with bleeding outcomes feedback could support quality improvement efforts, yet such metrics will need to be risk-adjusted for meaningful comparison across centers.
- © 2011 by American Heart Association, Inc.