When Are Packed Red Blood Cell Transfusions During ECMO Beneficial to Survival?

Packed red blood cell (PRBC) transfusion during venovenous (VV) extracorporeal membrane oxygenation (ECMO) is associated with a decreased mortality risk only when the patient’s hemoglobin concentration is less than 7 g/dL, according to study findings published in The Lancet Respiratory Medicine.

Hemoglobin thresholds for PRBC transfusion are usually higher in patients receiving VV ECMO than in other patients who are critically ill. Researchers for the prospective, observational PROTECMO study (ClinicalTrials.gov Identifier NCT03815773) sought to determine how mortality rates in patients receiving PBRC transfusion during VV ECMO were affected by the patient’s hemoglobin concentration at the time of transfusion.

The PROTECMO study was conducted in 41 ECMO centers in 19 countries in Europe, North America, Asia, and Australia. Participants were adult patients at least 18 years of age with acute respiratory distress syndrome (ARDS) who were receiving VV ECMO.

The trial enrolled 604 patients (431 [71%] men; mean [SD] age, 50 [13.6] years) from December 1, 2018, to February 22, 2021. The mean hemoglobin concentration at cannulation was 10.9 (2.4) g/dL and was significantly lower in high-volume centers.

Longitudinal data were obtained for 7944 days on ECMO. The mean hemoglobin concentration during ECMO was 9.1 (1.2) g/dL. Increased levels were observed in the first 7 days (mean hemoglobin, 9.3 [1.5] g/dL), but hemoglobin levels progressively decreased despite the different quartiles of baseline hemoglobin levels.

PRBC transfusions were performed on 2432 of the 7944 study days. Patients received a median of 115 mL (interquartile range [IQR], 48-223) of PRBC per day during ECMO. The median amount of PRBC that was transfused on the days of a transfusion was 425 mL (IQR, 350-556), and 504 (83%) of 604 patients received at least 1 unit of PRBC and were transfused on a median of 31% of days (IQR, 15%-37%). In addition, 100 (17%) patients completed ECMO without any PRBC transfusion, with a median ECMO duration of 5.7 days (IQR, 3.4-9.8), significantly shorter than patients who received transfusions.

The overall mean pretransfusion hemoglobin concentration was 8.1 g/dL (SD 1.1), although it varied based on the clinical rationale for transfusion.

An increased hemoglobin concentration was associated with a lower probability of death in the intensive care unit during the 28-day follow-up (hazard ratio [HR] 0.87 [95% CI, 0.78-0.98], according to a time-dependent Cox model. However, when assessing hemoglobin concentration by strata, a daily hemoglobin concentration of less than 7 g/dL was the only consistent threshold as a risk factor for death (2.99 [1.95-4.60]).

PRBC transfusions were associated with a decreased risk of death only when they were performed when the hemoglobin concentration was less than 7 g/dL (HR 0.15 [95% CI, 0.03-0.74]; P = .019). The greatest hemoglobin concentration used as a trigger for PRBC transfusion that was associated with a decreased mortality risk was 7.2 g/dL, according to the time-dependent Cox model (HR 0.38 [0.17-0.83]) and the marginal structural Cox proportional hazards model (HR, 0.23 [0.06-0.87]).

Limitations of this study include the unaccounted for effect of unknown confounding variables on clinical outcomes; limited longitudinal data; and the investigators’ lack of direct access to clinical data.

“During VV ECMO for ARDS there was no universally accepted trigger for transfusion, although the threshold appears to be lower than in previous recommendations,” the investigators commented. “Transfusion of PRBC was consistently associated with lower mortality when done when hemoglobin concentrations were less than 7 g/dL. However, residual confounding exists, and these associations should be confirmed in a prospective interventional trial.”

Disclosure: Some of study authors declared affiliations with biotech, pharmaceutical, and/or device companies. Please see the original reference for a full list of authors’ disclosures.