ABSTRACT
Our study investigated the impact of packed red blood cell (pRBC) transfusion on the occurrence of bloodstream infections (BSIs) in patients admitted in a multidisciplinary intensive care unit (ICU), further assessing potential associations with particular BSI types. A nested matched (1:1) case-control design was implemented. Sex, age, admission category, Acute Physiology and Chronic Health Evaluation score II (plus Injury Severity Score in trauma patients) were used for matching. Controls were selected to have an ICU length of stay at least equal to the time to first BSI episode of the corresponding cases. Propensity scores for receiving pRBC transfusion were calculated in the entire prospective cohort. Of 582consecutive ICU patients, 165 matched case-control pairs were formed. In multivariable analysis, pRBC transfusion was independently associated with 2-fold probability for BSI (adjusting for matching variables and propensity score). There was a significant dose-dependent association of BSI risk with regard to the number of pRBC units transfused (odds ratios [OR], 1.73, 2.09, 2.34 for 1-3, 4-6, and more than 6 pRBC units transfused, respectively, compared with nontransfused patients, P values .116, .018, and .015, respectively). In subgroup analysis, catheter-related BSIs displayed the strongest association with pRBC transfusion (OR = 5.01, P = .014).
