Dupuis C, Garrouste-Orgeas M, Bailly S, et al. Effect of transfusion on mortality and other adverse events among critically ill septic patients: An observational study using a Marginal Structural Cox Model. Crit Care Med 2017; 45: 1972-80 PMID: 28050898

From Wiki Journal Club
Jump to: navigation, search
Dupois C, et al. "Effect of Transfusion on Mortality and Other Adverse Events Among Critically Ill Septic Patients: An Observational Study Using a Marginal Structural Cox Model". Critical Care Medicine. 2017. 45(12):1972-1980.
PubMedFull text

Clinical Question

RBC Transfusion (RBCT) is often required in patients with sepsis. One RCT and several cohort studies and analyses have led to conflicting results. 1-4 Does RBC Transfusion in critically-ill septic patients have adverse effects on mortality?

Bottom Line

In this retrospective study, RBC Transfusion was shown not to affect overall mortality in critically ill patients with sepsis; in fact, there was a protective effect in patients with low hematocrits. However, there was an increased rate of nosocomial infections and severe hypoxemia in those transfused. RBCT needs to be individualized for this patient subgroup. This study may be more reflective of actual clinical practice than other similar studies.

Major Points

Following the results of the landmark TRICC trial, 5 the trigger for RBC transfusion (RBCT) in most critically-ill patients is generally accepted as 70 g/L. Anemia is common in critically-ill patients with sepsis and is associated with morbidity and mortality. However, numerous trials have indicated that transfusion itself might be harmful. Thus, the optimal transfusion thresholds in these patients is controversial.

Thus far, only the TRISS trial (Transfusion Requirements in Septic Shock) has actually compared two RBCT strategies among septic shock patients. 1 This trial excluded patients who were already transfused, or those with acute coronary syndromes. Despite this trial demonstrating similar outcomes with restrictive (Hb < 70) as opposed to liberal transfusion (Hb > 90) thresholds, there remain doubts about the external validity of this RCT. 4

This study aimed to estimate the average causal effect of a daily leucoreduced RBCT among critically-ill patients with sepsis using a marginal structural cox model (MSCM). A retrospective analysis of the French prospectively-collected OutcomeRea database on all patients admitted over a 16-year period was conducted.

The primary objective was to determine the effect of one or more RBCT within 1 day on the 30-day ICU mortality. Other outcomes assessed were the effects of RBCT on the occurrence of nosocomial infections and severe hypoxemia (defined as a P/F ratio of < 200). Patients were divided into subgroups based on their hematocrit during ICU stay.

Of the group of adult patients with severe sepsis or septic shock, 36.6% received 1 or more RBCTs. In general, patients who were transfused were more severely ill, were more likely to be mechanically ventilated or on renal replacement therapy, had a higher incidence of immunosuppression and had more bleeding events, a worse ICU LOS and mortality. There were more ischemic events only in the subgroup with the lowest hematocrit level (median Hematocrit 26, IQR 24-28). In this subgroup, RBCT had a protective effect on mortality.

Transfusion was not associated with an increased risk of mortality across all subgroups; however, this was associated with an increased risk of nosocomial infections and with severe hypoxemia. Transfusion of at least one RBC concentrate was associated with a lower risk of death for the low hematocrit subgroup. This effect was not observed for groups with higher hematocrits. Transfusion thresholds need to be individualized for critically-ill patients with sepsis or septic shock and may be of benefit for patients with lower hematocrits.


No guidelines for this specific subset of ICU patients.


  • Retrospective analysis of a French multicentre prospective observational cohort
  • N = 5296
  • No. who received transfusions = 1940
  • Setting: 23 French ICUs
  • Enrollment: 1998-2014
  • Follow-up: 30 days
  • Analysis: Marginal Structural Cox Models that were survival causal models to estimate the real causal effect of the treatment on outcome.
  • Primary outcome: The effect of RBCT on 30-day ICU mortality.


Inclusion Criteria

All patients admitted for the first time to participating ICUs with severe sepsis or septic shock during the study period ICU stay for over 3 days

Exclusion Criteria

Bleeding at the time of admission

Baseline Characteristics

  • Median age 65, predominantly male

For the group that received RBCT:

  • Higher surgical admissions
  • Higher incidence of immunosuppression
  • Higher Severity of illness (SAPS II) and organ dysfunction (SOFA) scores
  • Higher incidence of septic shock at presentation
  • More volume of fluid received
  • Higher incidence of bleeding events and nosocomial infections
  • Lower incidence of ischemic events (except in the subgroup with the lowest hematocrit)
  • Increased ICU LOS and mortality




Primary Outcomes

Transfusion of at least one RBC a day was not associated with mortality
(HR 1.07; 95% CI 0.88-1.30; P=0.52).

Secondary Outcomes

RBCT significantly associated with severe hypoxemia
(HR 1.29; 95% CI 1.14-1.47; P < 0.01)
RBCT associated with 2-fold increased risk of nosocomial infection
(HR 2.77; 95% CI 2.33-3.28; P<0.01)

Subgroup Analysis

More ischaemic events in the subgroup with very low hematocrit (Median Hct 26). Improved survival demonstrated only in this particular subgroup of patients who received blood transfusions (HR 0.72; 95% CI 0.55-0.95; P=0.02).


  • Retrospective observational study.
  • Marked baseline differences between the two groups with the group that received transfusion being noticeably sicker.
  • Inability to adjust for unmeasured confounders.
  • No mechanism for specifically looking at transfusion-associated circulatory overload (TACO) vs. transfusion-associated acute lung injury (TRALI) as cause for hypoxemia following blood transfusions.


Publicly funded

Further Reading

1. Holst LB, Haase N, Wetterslev J, et al. Lower versus higher hemoglobin threshold for transfusion in septic shock. N Engl J Med 2014; 371:1381-91 (https://www.ncbi.nlm.nih.gov/pubmed/25270275) 2. Park DW, Chun B-C, Kwon S-S et al. Red blood cell transfusions are associated with lower mortality in patients with severe sepsis and septic shock: A propensity-matched analysis. Crit Care Med 2012; 40:3140-45 (https://www.ncbi.nlm.nih.gov/pubmed/22975891) 3. Dupuis C, Sonneville R, Adrie C, et al. Impact of transfusion on patients with sepsis admitted in intensive care units: A systematic review and meta-analysis. Ann Intensive Care 2017; 7: 5 (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5209327/) 4. Kramer AH: Is a hemoglobin concentration as low as 70 g/dL adequate for all critically ill patients with sepsis? Legitimate doubts remain! Crit Care Med 2017; 45:2101-02 (https://www.ncbi.nlm.nih.gov/pubmed/29148987) 5. Hebert PC, Wells G, Blajchman MA, et al. A multicenter, randomized controlled clinical trial of transfusion requirements in critical care. N Engl J Med 1999; 340:409-17 (http://www.nejm.org/doi/full/10.1056/NEJM199902113400601)