588

Abstract

Introduction: Acuity scoring systems were implemented in the pediatric intensive care unit (PICU) to provide an assessment of severity of illness and outcome. The Pediatric Risk of Mortality (PRISM), Pediatric Incidence of Mortality (PIM), and Pediatric Logistic Organ Dysfunction (PELOD) scoring systems are the most commonly used predictors of mortality. Packed red blood cell (PRBC) transfusions are common practice and associated with adverse outcomes. Although some factors used in the clinical decision to transfuse are also variables in acuity scoring, therapeutic interventions are not included in calculations of acuity. We hypothesize that transfusion in the PICU results in an increase in mortality that is independent of acuity. Methods: We conducted a retrospective review of all patients admitted to a tertiary care PICU between 2008 and 2012. T-test and Chi-square were utilized to compare groups. Multiple regression and logistic regression analysis were used to examine outcome. Results: Of 5,178 PICU admissions, 480 (9.3%) received PRBC transfusions. Utilizing multiple regression analysis, transfusion, age at admission, gender, PRISM, and PIM were all significantly predictive of average PELOD score. Using regression analysis, transfusion, PIM, and age at admission were significantly predictive of length of stay (LOS) with transfusion adding 7.0 days (P<.001), an increase in PIM by 10 adding 7.5 days (P<.001), and a 10 month increase in age decreasing LOS by 0.06 days (P<.001). Assessing transfusion, PRISM, and PIM, only transfusion was a significant predictor of mechanical ventilator days adding 5.9 vent days (P< 0.001). Mortality occurred during 51(10.6%) admissions among transfused and 43 (0.9%) of the 4,698 not transfused (P<.000). Logistic regression analysis showed transfusion, PRISM, and PIM were significantly predictive of mortality with transfusion independently increasing mortality by 2.9 times (95%CI: 1.6-5.5), an increase of 10 in PRISM leading to a 4.9 fold increase in risk of death (95%CI: 3.4-7.2), and an increase in PIM by 5 leading to a 21.9 fold increase in risk of death (95%CI: 8.6-55.4). Conclusions: Transfusion was predictive of higher PELOD score, LOS, and ventilator days. When controlled for severity of illness, transfusion was independently associated with an increase in mortality.