Real-world efficacy of transfusion with liberal or restrictive strategy in traumatic brain injury.

Objective: The short-term efficacy of red blood cell (RBC) transfusion among general traumatic brain injury (TBI) patients is unclear.

Methods: We used the MIMIC database to compare the efficacy of liberal (10 g/dL) versus conservative (7 g/dL) transfusion strategy in TBI patients. The outcomes were neurological progression (decrease of Glasgow coma scale (GCS) of at least 2 points) and death within 28 days of ICU admission. Each eligible individual was cloned and assigned each of the replicates to one of the treatment arm. The imbalance induced by informative censoring was adjusted by inverse probability weighting. The standardized, weighted pooled logistic regression with 500 bootstrap resampling was used to estimate the cumulative risk difference and 95% confidence interval (CI).

Results: Of the 1141 eligible individuals, 29.0% received RBC transfusion. Compared with the restrictive group, the liberal strategy reduced early death (3 days: 5%, 95% CI: 2%-7%; 7 days: 6%, 95% CI: 3%-11%); however, no significant difference of mortality risk at 28-day or neurological progression risk at any time points was observed. The risk of coagulopathy at 3 days was increased by 7% (95% CI: 1%-19%) in the liberal group. The subgroup analysis indicated a beneficial effect of liberal transfusion on mortality in hemodynamically unstable patients.

Interpretation: Compared with the restrictive strategy, the liberal strategy does not improve the short-term neurological prognosis and death among patients with TBI in a real-world situation. The liberal strategy may be beneficial to survival at very early stage or in hemodynamically unstable subgroup.