Background: Interhospital differences in blood transfusion practice during coronary artery bypass graft (CABG) surgery have been noted, but the underlying issues have not been identified.
Study design and methods: Records of 3217 consecutive CABG cases in five university teaching hospitals in 1992 and 1993 were stratified by hospital, type of revascularization conduit, patients' sex, and other factors. Statistical methods were used to compare patient characteristics, transfusion outcomes, and hospital outcomes.
Results: Forward two-step logistic regression using patient likelihood of red cell transfusion factors in the first step and the specific hospital in the second step revealed a significant effect of hospital on the delta odds ratios for red cell transfusion. This finding was confirmed by analyses of a highly stratified subset of cases, males in diagnosis-related group 107 (primary cases of coronary bypass without coronary catheterization) who underwent revascularization with venous and internal mammary artery grafts, revealing variations among hospitals from 109 to 457 units of red cells transfused per hundred cases. Corresponding variations in transfusions of all blood components were from 324 to 1019 units by hospital. Variation in red cell transfusion practice among surgeons in the same hospital was not responsible for these interhospital differences.
Conclusion: The effect of the specific hospital on transfusion practice is attributed to institutional differences that, through reasons of training or hierarchy, become ingrained in hospitals.