Does prior transfusion worsen outcomes from infection in surgical patients?

Background: Controversy continues to exist regarding the immunomodulatory effects of cellular blood transfusions in the fields of oncology, transplantation, and infectious diseases. Numerous studies have correlated transfusion with hospital-acquired infection, but the impact of transfusion on infection-related mortality has not been addressed. The objective of this study was to determine the effect of transfusion on outcomes among infected surgical patients. Methods: Data on all hospital-acquired infectious episodes among surgical intensive care unit and ward patients were collected prospectively over 39 months at a single university hospital. The relationships between prior transfusion (defined as the receipt of allogeneic red blood cells or platelets during the index hospitalization but prior to the development of infection) and over 100 variables were examined using univariate and multiple logistic regression analysis, with inclusion of a propensity score for prior transfusion to attempt to account for treatment selection bias. Results: During the study period, 1,228 infectious episodes occurred; 641 were associated with the transfusion of packed red blood cells or platelets. Univariate analysis revealed that patients with those infectious episodes following transfusion had higher Acute Physiology and Chronic Health Evaluation II (APACHE II) scores (18.3 ± 0.3 vs. 10.9 ± 0.3, p < 0.0001) and higher mortality (26.0% vs. 8.9%, p < 0.0001). Those patients who died received a greater number of transfusions prior to infection than those who lived (7.7 ± 0.8 vs. 4.5 ± 0.3, p = 0.0002). Multiple logistic regression analysis revealed that the propensity score for prior transfusion independently predicted mortality (odds ratio 9.45, 95% confidence interval 3.71-24.09, p < 0.001), but that neither the presence or absence of transfusion (OR 0.89, 95% CI 0.54-1.46, P = 0.6) nor the absolute number of units of red blood cells or platelets transfused (OR for red blood cells 1.00, 95% CI 0.98-1.02, p = 0.7; OR for platelets 0.97, 95% CI 0.90-1.05, p = 0.5) independently predicted mortality. Conclusions: The transfusion of packed red blood cells or platelets prior to infection is associated with more severe disease among surgical patients, but once corrected for treatment selection bias does not appear to worsen outcomes from infection.