Liberal Versus Restrictive Transfusion Strategy in Critically Ill Oncological Patients: The Transfusion Requirements in Critically Ill Oncologic Patients Randomized Controlled Trial

Fabricio. 2017. Critical Care Medicine 2017;45(5)766-776. doi:10.1097/CCM.0000000000002283

Clinical Question

  • In cancer patients with septic shock does a restrictive vs. a liberal transfusion threshold reduce 28 day mortality?


  • A number of papers have investigated what transfusion threshold we should use in critically ill patients. The TRICC study included critically ill patients with acute anaemia, and the TRISS study included patients with septic shock. Both studies reported that a restrictive strategy was safe and reduced blood transfusions. Where as the TRISOP study reported that the composite outcome of mortality and major complications was reduced with a liberal threshold in surgical oncology patients. There is the additional concern of the association of blood transfusion with cancer progression. This study was therefore performed to try and determine what transfusion threshold we should use in oncology patients with septic shock.


  • Randomised controlled trial
  • Allocation through internet based system
  • Patients and outcome assessors blinded
  • Treating clinicians non blinded
  • Intention to treat analysis
  • Sample size calculation
    • 300 patients required to give 80% power to detect a reduction in mortality from 50% in liberal to 34% in restrictive with false positive rate of 5%


  • Single tertiary centre, Brazil
  • Data collected June 2012 – May 2014


  • Inclusion criteria – Oncological patients with septic shock
    • Adult patients
    • Diagnosis of solid cancer
    • Septic shock within 1st 6 hours of ICU admission
      • Suspected or verified infection with mean arterial pressure <65mmHg despite fluid therapy requiring vasopressors
  • Exclusion criteria:
    • Inability to receive blood transfusions
    • Too high expected mortality or transfusion rate
    • Haematological malignancy
    • End-stage renal failure
  • 1,658 patients screened of whom 300 enrolled
  • Comparing liberal vs. restrictive groups
    • Age: 62 vs. 61
    • Co-morbidities
      • Diabetes: 20% vs. 19%
      • COPD 9% vs 6%
      • Congestive heart failure: 7% vs. 5%
    • Type of tumour
      • Gastro-intestinal tract: 52% vs. 59%
      • Lung: 15% vs. 13%
    • Chemotherapy 4 week before ICU admission: 26% vs. 22%
    • Radiotherapy 4 week before ICU admission: 5% vs. 4%
    • Site of infection
      • Lung: 69% vs. 61%
    • Positive cultures
      • 28% vs. 34%
    • SAPS III score: 55 vs. 59
    • SOFA score: 6 vs. 7


  • Restrictive strategy
    • Patients received 1 unit of RBCs each time Hb <7g/dL
    • 41% of patients had RBC transfusion
    • Units of RBC transfused, median (IQR): 0 (0-2)


  • Liberal transfusion
    • Patients received 1 unit of red blood cells each time Hb <9g/dL
    • 61% of patients had RBC transfusion
    • Units of RBC transfused, median (IQR): 1(0-3)

For both intervention and control groups

  • Haemoglobin levels checked on ICU admission, twice daily, and following transfusion
  • RBCs were leukodepleted
  • Intervention only performed during ICU stay. Transfusion threshold post ICU stay determined by treating clinician
  • Blood transfusion allowed outside study protocol for life threatening situations


  • Primary outcome: liberal vs. restrictive
    • All-cause mortality at 28 days – no significant difference
      • 45.0% vs. 55.6% (Hazard ration 0.74, 95% C.I. 0.53-1.04, p=0.08)
  • Secondary outcomes
    • No significant difference in:
      • Need for mechanical ventilation
        • 30.9% vs. 38.7%, HR 0.71 (95% C.I. 0.44-1.15), p=0.16
      • Renal replacement therapy
        • 8.7% vs. 12.0%, HR 0.70 (95% C.I. 0.33-1.49), p=0.35
      • Need for inotropic support
        • 16.2% vs. 22.1%, HR 0.68 (95% C.I. 0.38-1.22), p=0.19
      • Acute myocardial infarction
        • 2.7% vs. 2.7%, HR 1.00 (95% C.I. 0.25-4), p=1
      • 60 day all cause mortality
        • 56.4% vs. 65.5%, HR 0.77 (95% C.I. 0.555-1.07), p=0.12
      • ICU length of stay: 7 vs. 7 days, p=0.73
      • Hospital length of stay: 15 vs. 13 days, p=0.23
    • Significantly increase 90 day mortality in restrictive group
      • 59.1% vs. 70.2%, HR 0.72 (95% C.I. 0.53-0.97), p=0.03
      • Number needed to treat = 9 patients
      • Fragility index = 0 patients

Authors’ Conclusions

  • A survival trend was found favouring the liberal transfusion strategy


  • Randomised controlled trial
  • Blinding of outcome assessors
  • Registered on


  • Single centre
  • The study was powered for a 16% reduction in mortality. However the difference reported between the 2 treatments groups was 11.6%. This is still clinically significant and would have required a larger study to be powered to detect this difference.
  • Some baseline differences with higher SAPS and SOFA scores in the restrictive group
  • Patients recruited despite having Hb >9. In the liberal group only 61% of patients received a blood transfusion. Any differences reported for patients who had Hb >9 throughout the trial would have not been due to the difference in transfusion thresholds.

The Bottom Line

  • This single centre randomised controlled trial reported that in oncological patients with septic shock a restrictive vs. liberal transfusion threshold resulted in no significant difference in 28 day mortality.
  • The study reported that one of their secondary outcomes, 90 day mortality, was significantly reduced at 90 days. I would be cautious about putting too much emphasis on this result as the fragility index was 0; there were no differences in 28 and 60 day mortality as well as a number of other secondary outcomes, and with there being only 1 unit difference in median units of RBC transfused between the two treatment groups, I would be surprised if this caused a 11% difference in mortality. Larger studies would be needed to confirm or refute this finding.

External Links


Summary author: @davidslessor
Summary date: 5th June 2017
Peer-review editor: @avkwong

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.