Patterns of red blood cell transfusion use and outcomes in patients undergoing percutaneous coronary intervention in contemporary clinical practice: Insights from the NCDR ® Matthew W. Sherwood, Yongfei Wang, Jeptha P. Curtis, Eric D. Peterson, Sunil V. Rao
Disclosures Matthew W. Sherwood – None Yongfei Wang – None Jeptha P. Curtis – None Eric D. Peterson – Research Support >10K : Eli Lilly, Janssen Pharm., PI of Data Analytic Center for ACC Sunil V. Rao – Research grants - Ikaria, sanofi-aventis; Consultant/honoraria - The Medicines Co, Terumo Medical, ZOLL, Astra Zeneca, Daiichi Sankyo Lilly, Janssen
Funding Support and Disclaimer This research was supported by the American College of Cardiology Foundation’s National Cardiovascular Data Registry (NCDR). The views expressed in this presentation represent those of the author(s), and do not necessarily represent the official views of the NCDR or its associated professional societies identified at
Background Prior studies have shown that there is marked variation in the use of red blood cell transfusion (RBCT) among patients with acute coronary syndromes Contemporary post-procedure RBCT patterns in patients undergoing PCI are unclear Documenting variation in RBCT practice is important since RBCT has been independently associated with morbidity and mortality in patients with ischemic heart disease
Objectives To determine the variability in use of RBCT in hospitals across the United States To determine patient factors associated with RBCT To determine whether RBCT has an independent association with patient outcomes – Is an association of transfusion with outcomes independent of bleeding
Methods Database – NCDR ® Cath-PCI ® database Patients – 1,323,965 patients undergoing PCI at 1282 hospitals between 7/2009-9/2011 Exclusions – patients who underwent in-hospital CABG – More then 1 PCI during hospital stay – Missing data on bleeding events, procedural complications, d/c status
Outcomes and Definitions Primary – Transfusion rates Secondary – Clinical Outcomes – MI – Stroke – In-hospital Death Definition – Bleeding Events – Hemoglobin drop of ≥3 g/dL – Transfusion of whole blood or packed red blood cells – Procedural intervention/surgery at the bleeding site to reverse/stop or correct the bleeding
Analyses Rates of transfusion by site were determined and then risk adjusted rates were calculated Patient clinical characteristics and in-hospital outcomes were compared between patients who did and did not receive RBCT Logistic regression was used to determine the adjusted association between RBCT and in-hospital death, MI, or stroke – Secondary analyses performed to determine whether any adverse effect of transfusion was independent of bleeding events
Patient Characteristics (%) Without RBCT N= With RBCT N=29255 Age (mean, SD)64.5 (12.1)70.5 (12.1) Gender (% Female) HTN Diabetes ESRD on dialysis Prior MI Prior CHF P values for all comparisons <0.001
Transfusion Pattern by Hgb
Transfusion Rates by hospital site
Adjusted transfusion rates Number of hospitals Risk adjusted for all variables in the established NCDR mortality and bleeding models
Outcomes by transfusion status Patient Outcomes (%) Without RBCT N= With RBCT N=29255 MI Stroke CHF Cardiogenic Shock In-hospital Death Bleeding Events Access Site bleeding Non-Access Site bleeding P values for all comparisons <0.001
Independent assoc. of RBCT with outcomes Patient OutcomesOdds Ratios MI, Stroke, In-hospital Death 2.18 ( ) MI1.96 ( ) Stroke3.92 ( ) In-hospital Death2.02 ( ) Model includes all variable in the established NCDR mortality model; Reference is no transfusion All patients
Patient OutcomesOdds Ratios MI, Stroke, In-hospital Death 1.95 ( ) MI1.71 ( ) Stroke4.07 ( ) In-hospital Death1.73 ( ) Model includes all variable in the established NCDR mortality model; Reference is no transfusion Independent assoc. of RBCT with outcomes Patients without bleeding
Limitations Data is observational thus events are reported, not adjudicated Temporal relationship between Hct, transfusion, and events is uncertain Cannot infer causality
Conclusions Considerable variation in transfusion practices exists across the U.S., and persists after adjustment for patient differences Transfusion patterns by Hgb level are different in patient with bleeding vs. without bleeding RBCT is independently associated with adverse cardiac events in patients undergoing PCI – This association still holds in patients without bleeding events
Clinical Implications Our results are consistent with prior reports demonstrating the potential hazard associated with RBCT among ACS patients Randomized trials of transfusion strategies are needed in patients undergoing PCI to guide clinical practice Until these data are available, operators should continue to adopt practices that reduce the risk for bleeding and transfusion