A proposal for an individual patient data based meta-analysis of randomized controlled trials of allogeneic transfusion and postoperative bacterial infection.

Abstract

LLOGENEIC blood transfusions have been reported to cause immunomodulatory effects, which purportedly enhance the survival of renal allografts in recipients, ~ increase the recurrence rate of resected malignancies, 2-5 increase the frequency of postoperative bacterial infections, 3,5,6-8 reduce the recurrence rate of Crohn's disease? and/or accelerate the progression of human immunodeficiency virus infection.l~ The specific constituents of donor blood that mediate these immunomodulatory effects remain unknown, but both animal and human data suggest that these effects are mediated by allogeneic leukocytes, n Plausible mechanisms for these effects have been advanced by several authors, 12-16 but experiments in laboratory animals may not directly extrapolate to humans. a7 Moreover, different pathophysiologic mechanisms may be involved in each of the reported immnnosuppressive effects associated with allogeneic blood transfusions. Finally, future research may corroborate some of the effects and fail to confirm others. The association of perioperative allogeneic transfusion with postoperative bacterial infection has been studied extensively, with over 30 observational studies reported. 3,6-8 Although some of these studies detected no adverse transfusion effect, ~8-21 in most allogeneic blood transfusions emerge as the single, best predictor of postoperative infection. In one report, allogeneic blood transfusions accounted for a 10-fold increase in the risk of infection, in both univariate and multivariate analyses) Many observational studies compared the frequency of postoperative infection in transfused and untransfused patients who differed with regard to baseline characteristics relating to both the need for