Is the transfusion requirement predictable in critically ill patients after admission to the intensive care unit?

Abstract

Objective: In intensive care medicine the clinical decision to order and transfuse red blood cells (RBC) is usually based on hematocrit or hemoglobin levels. The intention of this study was to investigate whether clinical or laboratory variables, taken after the admission of patients to the intensive care unit (ICU), are able to predict the transfusion requirement of the following 72 h.

Design: The values of initially measured systolic blood pressure, hematocrit level, and the values of 2 scores of severity of disease (Acute Physiology And Chronic Health Evaluation [APACHE-II], Mortality Prediction Model [MPM]) were calculated after the admission of patients to the ICU. The decision for transfusion was based on specific criteria. The median values of the scores, those of the variables, and the median number of transfused RBC units of the surviving group were compared to the values of the group of patients who died during hospital stay. The quantity of RBC transfusions was compared to the variables and score values by linear regression analysis. Additionally, the values of the patients who did not receive blood transfusion were compared to those of patients who required RBC. Furthermore, the patient group with neurosurgical diseases was compared to the group without neurosurgical diseases.

Patients: 117 patients were prospectively and consecutively investigated in an 8-bed ICU of a university hospital.

Results: Nonsurvivors required significantly more units of RBC during the first 72 h (p < 0.05). Patients who did not require transfusion had a higher hematocrit and a lower APACHE-II value at admission (p < 0.001). In the MPM values no differences were found. Patients with neurosurgical diseases had a higher initial hematocrit value, and they required less units of RBC in comparison to patients without neurosurgical diseases. In the analysis of linear regression neither in the initially measured systolic blood pressure nor in the APACHE-II and MPM we found a strong linear correlation to the quantity of blood transfusion.

Conclusions: A hematocrit value < or = 20% and a APACHE-II score > or = 20 at the time of admission to the ICU referred to a demand for blood transfusion. We believe that these parameters are useful as predictive instruments. The initially measured systolic blood pressure had no prognostic capacity. In the individual patient a number of factors should be taken into account to decide whether to transfuse or not.