Restrictive vs Liberal Transfusion Strategy in Patients With Acute Brain Injury: The TRAIN Randomized Clinical Trial

Abstract

Importance: Blood transfusions are commonly administered to patients with acute brain injury. The optimal hemoglobin transfusion threshold is uncertain in this patient population.

Objective: To assess the impact on neurological outcome of 2 different hemoglobin thresholds to guide red blood cell transfusions in patients with acute brain injury.

Design, setting, and participants: Multicenter, phase 3, parallel-group, investigator-initiated, pragmatic, open-label randomized clinical trial conducted in 72 intensive care units across 22 countries. Eligible patients had traumatic brain injury, aneurysmal subarachnoid hemorrhage, or intracerebral hemorrhage; hemoglobin values below 9 g/dL within the first 10 days after injury; and an expected intensive care unit stay of at least 72 hours. Enrollment occurred between September 1, 2017, and December 31, 2022. The last day of follow-up was June 30, 2023.

Interventions: Eight hundred fifty patients were randomly assigned to undergo a liberal (transfusion triggered by hemoglobin <9 g/dL; n = 408) or a restrictive (transfusion triggered by hemoglobin <7 g/dL; n = 442) transfusion strategy over a 28-day period.

Main outcomes and measures: The primary outcome was occurrence of an unfavorable neurological outcome, defined as a Glasgow Outcome Scale Extended score between 1 and 5, at 180 days following randomization. There were 14 prespecified serious adverse events, including occurrence of cerebral ischemia after randomization.

Results: Among 820 patients who completed the trial (mean age, 51 years; 376 [45.9%] women), 806 had available data on the primary outcome, 393 in the liberal strategy group and 413 in the restrictive strategy group. The liberal strategy group received a median of 2 (IQR, 1-3) units of blood, and the restrictive strategy group received a median of 0 (IQR, 0-1) units of blood, with an absolute mean difference of 1.0 unit (95% CI, 0.87-1.12 units). At 180 days after randomization, 246 patients (62.6%) in the liberal strategy group had an unfavorable neurological outcome compared with 300 patients (72.6%) in the restrictive strategy group (absolute difference, -10.0% [95% CI, -16.5% to -3.6%]; adjusted relative risk, 0.86 [95% CI, 0.79-0.94]; P = .002). The effect of the transfusion thresholds on neurological outcome at 180 days was consistent across prespecified subgroups. In the liberal strategy group, 35 (8.8%) of 397 patients had at least 1 cerebral ischemic event compared with 57 (13.5%) of 423 in the restrictive strategy group (relative risk, 0.65 [95% CI, 0.44-0.97]).

Conclusions and relevance: Patients with acute brain injury and anemia randomized to a liberal transfusion strategy were less likely to have an unfavorable neurological outcome than those randomized to a restrictive strategy.

Trial registration: ClinicalTrials.gov Identifier: NCT02968654.

Figure 1.. Flow of Participants in the TRAIN Trial

GCS indicates Glasgow Coma Scale; ICH, intracerebral hemorrhage; and ICU, intensive care unit. ^aThe GCS is a neurological assessment tool used by medical and nursing staff to monitor the clinical progression of a patient’s consciousness after acute brain injury. The scale is based on 3 types of responses to stimuli (eye, verbal, and motor); the overall score is the sum of the individual assessments for each function. The maximum score is 15, indicating a normal state of consciousness, while the minimum score is 3, indicating a deep state of unconsciousness. ^bPatients were randomly allocated in a 1:1 ratio to a restrictive strategy (hemoglobin concentration <7 g/dL) or a liberal strategy (hemoglobin concentration <9 g/dL) to determine when red blood cell transfusion should be given. Stratification was performed based on center, type of brain injury (traumatic brain injury, subarachnoid hemorrhage, or intracerebral hemorrhage), and GCS score at the time of randomization (3-5, 6-9, or 10-13). ^cAmong the 30 patients excluded after randomization because consent was withdrawn, no further data were registered, including outcome data.

Figure 2.. Median Daily Lowest Hemoglobin Concentration at Baseline and After Randomization

Baseline values were the last blood hemoglobin level measured before randomization. Day 0 was the day of randomization. Boxes indicate IQRs; bars inside the boxes, medians; whiskers, highest and lowest values within 1.5 times the IQRs; and circles outside the boxes, outlying data. See eFigure 1 in Supplement 1 for presentation showing medians (95% CIs) and eFigure 2 in Supplement 1 for median daily highest hemoglobin concentrations.

Figure 3.. Relative Risk of the Primary Outcome of Unfavorable Neurological Outcome at 180 Days Among All Patients and in Prespecified Subgroups

ICP indicates intracranial pressure and SOFA, Sequential Organ Failure Assessment. Neurological outcome 180 days after randomization was assessed using the Glasgow Outcome Scale Extended (GOS-E), which was dichotomized as unfavorable (GOS-E score of 1-5) or favorable (GOS-E score of 6-8); the scale ranges from 1 to 8, with death being included in the scale (GOS-E score of 1) and higher scores indicating better outcome. The size of each box is proportional to the size of the corresponding subgroup. ^aThe SOFA score is calculated from 6 baseline variables; scores range from 0 to 24, with higher scores indicating greater severity of disease.

Figure 4.. Distribution of Glasgow Outcome Scale Extended (GOS-E) Scores 180 Days After Randomization (Secondary Outcome)

Each cell corresponds to a score on the scale; the width of each cell represents the percentage of patients with that score. The vertical dashed line indicates the GOS-E score used for dichotomization.