PURPOSE: Data suggests transfusion increases nosocomial infections and negatively impacts outcomes in the critically ill. We assessed the impact of transfusion on risk and microbiolgy of infections in ICU patients and its effect on length of stay (LOS) and mortality.
METHODS: We performed a retrospective study of all patients admitted to a 24-bed adult Medical-Surgical ICU between July 2003 and September 2006 and entered in the Project IMPACT database. We examined the association between transfusion and development of nosocomial infections, causative organisms, ICU and hospital LOS, and hospital mortality. Categorical data was compared using the Chi-Square test and continuous variables were analyzed by the Mann-Whitney U-test.
RESULTS: 2436 patients, 1823 not transfused (Group 1) and 613 transfused (Group 2)were evaluated. Median ICU LOS was 2.4d [1.6-4.3] in Group 1 vs. 4d [2.2-10] in Group 2 and median hospital LOS was 10d [7-18] in Group 1 vs. 17d [10-32] Group 2 (p<0.001 for both). Mortality was 8.7 % in Group 1 vs. 13.1% in Group 2 (OR 1.5 [1.2-2] p=0.003). 224 infections were documented in 158 patients: 58 pneumonia, 40 bacteremias, 8 fungemias, 27 central-line associated infections, 15 clinical diagnoses of sepsis with negative cultures, and 76 pyelonephritis. The odds-ratio of having at least 1 nosocomial infection for a transfused patient was 1.9 [1.4-2.7] (p<0.001). Adjusting for acuity and other confounders using a mixed regression model, transfusion was an independent predictor of nosocomial infections, prolonged ICU and hospital LOS and hospital mortality. Median time from first transfusion to first infection was 6d [4-12]. All Gram-positive organisms, MRSA, VRE, and all Gram-negatives and Acinetobacter species were significantly more frequent in Group 2 (p<0.05).
CONCLUSION: Critically-ill patients receiving transfusion have a higher mortality, longer length of stay and more nosocomial infections.
CLINICAL IMPLICATIONS: The decision to transfuse critically ill patients should take into account the increased risk of infection, higher mortality and longer LOS in these patients. In addition, the threshold for consideration of the diagnosis of infection should be lower in transfused patients.
DISCLOSURE: Ralph Daher, No Financial Disclosure Information; No Product/Research Disclosure Information