Background: Because of the immunomodulatory effects of transfusion, we attempted to identify factors associated with blood product use and determine the association of transfusion quantity with postoperative infection.
Study design: We studied total perioperative transfusion of blood products for 15,592 cardiovascular operations performed from July 1998 to May 2003. Infection end points were septicemia/bacteremia (n=351, 2.2%) and superficial (n=353, 2.3%) and deep (n=212, 1.4%) sternal wound infections. Factors associated with blood product administration were used to form balancing scores to adjust for differences in patient characteristics among those receiving and not receiving blood products.
Results: Fifty-five percent of patients received packed red blood cells (RBC), 21% received platelets, 13% got fresh frozen plasma (FFP), and 3% got cryoprecipitate. Factors associated with RBC use included older age, female gender, higher New York Heart Association class, lower hematocrit, reoperation, and longer cardiopulmonary bypass time--all indicative of higher-risk patients. The more RBC units transfused, the higher was the occurrence of septicemia/bacteremia (p < 0.0001) and superficial (p=0.0007) and deep (p < 0.0001) sternal wound infection. Use of FFP (septicemia/bacteremia) and platelets (septicemia/bacteremia and deep sternal wound infection) mitigated against this association only slightly.
Conclusions: Blood products tended to be used in the sickest patients. But after accounting for this, risk of infection increased incrementally with each unit of blood transfused. Although cause and effect cannot be established, results suggested that blood product transfusion is an independent risk factor for postoperative infection in cardiac surgical patients, blood products are more likely to be used in the sickest patients, and no amount of blood loss treated by transfusion is innocuous.