Neurophysiologic Effects of Red Blood Cell Transfusion in Severe TBI: A Substudy of a Randomized Controlled Trial

Neurophysiologic Effects of Red Blood Cell Transfusion in Severe TBI: A Substudy of a Randomized Controlled Trial


A key paradigm in the critical care of patients with severe traumatic brain injury (TBI) is avoidance of physiologic derangements that may contribute to “secondary” brain injury. There is a significant association between reductions in cerebral oxygen delivery and poor neurologic outcomes [1-2]. Because oxygen delivery is highly dependent on the concentration of circulating hemoglobin (Hb), anemia is a potential cause of secondary injury, and in turn worsened neurologic outcomes [3-4]. Clinicians commonly transfuse red blood cell cells (RBCs) at a higher Hb threshold in neurocritical care patients than in other populations [5].

The neurophysiologic effects of RBC transfusion in patients with TBI are incompletely understood. Several studies suggest that transfusion increases brain tissue oxygen tension (PbtO2), although this increment appears to be of relatively small magnitude and transient duration [6-10]. It is possible that the impact of transfusion on PbtO2 may be influenced by the duration that RBCs have been stored prior to transfusion, but results of published studies have been conflicting [9, 11- 12]. Differing findings have also been reported in studies assessing the effects of transfusion on regional oxygen saturation (rSO2) using near infrared-spectroscopy [13-14]. Transfusion does not appear to have a clear effect on cerebral lactate, pyruvate and glucose concentrations, assessed using microdialysis [8, 15]. Although RBC transfusion probably improves oxygen delivery to the brain, it may have other deleterious neurophysiologic effects, including impairment of cerebral blood flow autoregulation, measured using the pressure reactivity index (PRx) [16]. Because anemia induces cerebral vasodilatation, RBC transfusion may also have an impact on intracranial pressure (ICP), although published studies have not reported such an effect [8, 10].


To prospectively determine the impact of RBC transfusion on ICP, cerebral perfusion pressure (CPP), rSO2, PbtO2, and PRx. We will assess each of the following: (1) The effect of differing transfusion thresholds (liberal vs. restrictive) on neurologic monitoring parameters over time; (2) The effect of individual units of RBCs on monitoring parameters, irrespective of the group to which a patient was randomly allocated; and (3) Whether the duration of RBC storage prior to transfusion modifies the effect of transfusion on neurologic monitoring parameters.


The HEMOglobin transfusion threshold in Traumatic brain Injury OptimizatioN (HEMOTION) study is a multi-center international randomized controlled trial comparing a liberal transfusion strategy (triggered by Hb < 100 g/L) with a restrictive transfusion strategy (triggered by Hb < 70 g/L). The study is expected to enroll 712 patients with moderate-severe TBI (admission GCS < 12) and Hb < 100 g/L, and the primary outcome is 6-month neurologic outcome.

As a substudy of HEMOTION, we will assess the impact of differing transfusion thresholds on neurophysiologic endpoints among patients undergoing neurologic monitoring. Patients considered for this substudy will have an arterial catheter and an ICP monitor in situ as part of routine care. The effects of transfusion on PbtO2, rSO2, and/or PRx will also be assessed among included patients in whom these parameters are either being monitored as part of routine care or are readily available to be measured at the time of randomization.

ICP, CPP, rSO2, and PbtO2 will be recorded hourly in all patients. During the two hours before and after transfusion, these parameters will be recorded every 15 minutes. Determination of PRx requires use of specialized software (ICM+; Cambridge, U.K.), such that the impact of transfusion on this parameter will only be assessed at centers where it is available. Average PRx will be determined for the two hours before and after transfusion.

Generalized estimating equation models with be created to compare physiologic endpoints in patients managed with a liberal vs. restrictive transfusion strategy. Similar analyses will be used to assess the impact of transfusion strategy on the proportion of time with “critical” values (ICP > 25, rSO2 < 55%, PbtO2 < 20, or PRx > 0.2). We will also generate models for “before and after” comparisons to determine the effects of transfusion in the two hours before and after each transfusion, and whether observations are modified by the duration of RBC storage.