Should Transfusion Trigger Thresholds Differ for Critical Care Versus Perioperative Patients? A Meta-Analysis of Randomized Trials

Abstract

Objective: To address the significant uncertainty as to whether transfusion thresholds for critical care versus surgical patients should differ.

Design: Meta-analysis of randomized controlled trials.

Setting: Medline, EMBASE, and Cochrane Library searches were performed up to 15 June 2016.

Patients: Trials had to enroll adult surgical or critically ill patients for inclusion.

Interventions: Studies had to compare a liberal versus restrictive threshold for the transfusion of allogeneic packed RBCs.

Measurements and main results: The primary outcome was 30-day all-cause mortality, sub-grouped by surgical and critical care patients. Secondary outcomes included myocardial infarction, stroke, renal failure, allogeneic blood exposure, and length of stay. Odds ratios and weighted mean differences were calculated using random effects meta-analysis. To assess whether subgroups were significantly different, tests for subgroup interaction were used. Subgroup analysis by trials enrolling critically ill versus surgical patients was performed. Twenty-seven randomized controlled trials (10,797 patients) were included. In critical care patients, restrictive transfusion resulted in significantly reduced 30-day mortality compared with liberal transfusion (odds ratio, 0.82; 95% CI, 0.70-0.97). In surgical patients, a restrictive transfusion strategy led to the opposite direction of effect for mortality (odds ratio, 1.31; 95% CI, 0.94-1.82). The subgroup interaction test was significant (p = 0.04), suggesting that the effect of restrictive transfusion on mortality is statistically different for critical care (decreased risk) versus surgical patients (potentially increased risk or no difference). Regarding secondary outcomes, for critically ill patients, a restrictive strategy resulted in reduced risk of stroke/transient ischemic attack, packed RBC exposure, transfusion reactions, and hospital length of stay. In surgical patients, restrictive transfusion resulted in reduced packed RBC exposure.

Conclusions: The safety of restrictive transfusion strategies likely differs for critically ill patients versus perioperative patients. Further trials investigating transfusion strategies in the perioperative setting are necessary.

Figure 1.

Forest plot showing the odds ratio (OR) and 95% CIs for the risk of 30-day mortality of patients receiving the liberal versus restrictive transfusion triggers, by critical care and perioperative subgroups. Thirty-day mortality is decreased with a restrictive strategy for critical care patients, but not perioperative patients.

Figure 2.

Forest plot showing the odds ratio (OR) and 95% CIs for the risk of 30-day mortality of patients receiving the liberal versus restrictive transfusion triggers, by restrictiveness of transfusion trigger for the critical care and perioperative subgroups. Thirty-day mortality is decreased with the most restrictive transfusion triggers for critical care patients, but there is an opposite direction of effect for the perioperative patients.

Figure 3.

Forest plot showing the odds ratio (OR) and 95% CIs for the risk of myocardial infarction (MI) of patients receiving the liberal versus restrictive transfusion triggers, by critical care and perioperative subgroups. There is a trend toward increased MI in perioperative patients receiving the restrictive strategy.