Premium
A clinical prediction tool to estimate the number of units of red blood cells needed in primary elective coronary artery bypass surgery
Author(s) -
Welsby Ian,
Crow Jennifer,
Bandarenko Nicholas,
Lappas George,
PhillipsBute Barbara,
StaffordSmith Mark
Publication year - 2010
Publication title -
transfusion
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.045
H-Index - 132
eISSN - 1537-2995
pISSN - 0041-1132
DOI - 10.1111/j.1537-2995.2010.02711.x
Subject(s) - medicine , logistic regression , bypass grafting , blood transfusion , blood product , cardiac surgery , coronary artery bypass surgery , ordered logit , predictive value , risk factor , blood bank , surgery , artery , emergency medicine , statistics , mathematics
BACKGROUND: Red blood cell (RBC) transfusion is common during cardiac surgical procedures. Empiric crossmatching, without attempting to estimate individual transfusion requirements is typical. We hypothesized that a clinical prediction tool could be developed to estimate the number of units of RBCs needed for coronary artery bypass grafting (CABG) surgery. STUDY DESIGN AND METHODS: With institutional review board approval, detailed demographic, risk factor, and transfusion data of primary elective CABG procedures (n = 5887) from September 1, 1993, to June 20, 2002, were studied and the data set was divided into development and validation subgroups. Multivariable ordinal logistic regression was used to develop and validate transfusion risk factors, assign them a relative weight, and create a model to stratify patients into groups depending on predicted need for 0, 2, 4, or more than 4 RBC units. The model was compared with current standard practice of crossmatching 4 RBC units in terms of observed blood product usage over the study period. RESULTS: Demographic and transfusion risk factor variables in the development (n = 3876) and validation (n = 2011) data sets were similar. The predictive value of the model was good for the development and validation groups, with a c‐index of 0.79 and 0.78, respectively. Applying the predictive model reduced the number of crossmatches by 30% without underproviding RBC units and increased the percentage of patients crossmatched exactly for the required number of units from 11% to 21%. CONCLUSIONS: Predictive factors for RBC transfusion were identified and used to construct a clinical tool to conserve blood bank resources without increasing patient risk.