FEAST: Mortality after Fluid Bolus in African Children with Severe Infection

FEAST Trial Group, 2011, NEJM;364(26)2483-95

Clinical Question

  • In paediatric patients with severe febrile illnesses, in a resource limited setting, do fluid boluses with albumin vs. saline vs. no fluid boluses affect mortality?


  • Randomised controlled trial
  • Use of opaque, sealed envelopes
  • Permuted blocks of random sizes
  • Stratified according to clinical centre
  • Non-blinded
  • Sample size calculation
    • With 3600 patients 80% power to detect a 33% relative reduction in mortality with saline vs. control and a 40% reduction with albumin vs. saline
    • Assuming risk of death of 11% in control
  • Intention to treat analysis
  • Patients lost to follow up/withdrawal of consent assumed to be alive at end of study (n=17 for primary endpoint)
  • Stopped early on recommendation of safety monitoring committee


  • 6 centres in Africa
  • January 2009 – January 2011


  •  Inclusion criteria:
    • Children aged 60 days to 12 years
    • Presence of all 3:
      1. Severe febrile illness
      2. Impaired consciousness (prostration or coma) and/or respiratory distress (increased work of breathing)
      3. Impaired perfusion, as evidenced by at least 1 of the following:
        • CRF >=3 seconds
        • Lower-limb temperature gradient
        • Weak radial-pulse volume
        • HR >180 if age <1 year, >160 if 1-5 years, or >140 if age >5 years
  • Exclusion criteria:
    • Severe malnutrition
    • Gastroenteritis
    • Non-infectious cause of shock (e.g. trauma, surgery, burn)
    • Volume expansion contraindicated
  • 3170 patients randomised
  • Baseline characteristics
    • In patients without severe hypotension similar across groups
      • Median age 24 months
      • 62% had prostation
      • 15% comatose
      • 83% had respiratory distress
      • 37% had convulsions during this illness
      • 39% had Lactate >5mmol/l
      • Mean Hb 7.1g/dl
      • 33% had Hb<5g/dl
      • 57% had malaria


  • In patients without severe hypotension (n=3141) randomised to
    • 20ml/kg normal saline 
    • 20ml/kg 5% albumin
    • No bolus
  • In children with severe hypotension (n=29)
    • (BP<50 if <12 months, BP<60 if age 1-5, BP<70 in children >5 years
    • Randomised to
      • 40ml/kg normal saline
      • 40ml/kg 5% albumin
  • After a protocol amendment in June 2010 initial bolus increased to 40ml/kg in patients without severe hypotension and 60ml/kg in patients with severe hypotension
  • In both groups, 
    • Patients in bolus groups received additional 20ml/kg at 1 hour if impaired perfusion
    • If severe hypotension developed then treated with 40ml/kg fluid (saline in control group)
  • All children 
    • treated on general paediatric wards; assisted ventilation other than short-term bag-and-mask support unavailable
    • Transfused if Hb <5g/dl
      • 45% in albumin-bolus group received transfusion vs. 47% in saline-bolus vs. 43% in control group
  • Median volume of all fluids given, including blood, for albumin-bolus group vs. saline bolus group vs. control
    • In 1st hour: 20ml/kg vs. 20ml/kg vs. 1.2ml/kg
    • In 2nd hour: 4.5ml/kg vs. 5ml/kg vs. 2.9ml/kg
    • In 1st 8 hours: 40ml/kg vs. 40ml/kg vs. 10.1ml/kg


  • Primary outcome: 
    • In patients without severe hypotension mortality at 48 hours was significantly worse in bolus groups vs. control group. There was no significant difference between albumin vs. saline
      • 10.6% in albumin bolus group vs. 10.5% in saline-bolus group vs. 7.3% in control group
      • RR of death with saline vs. control 1.44 (95% C.I. 1.09-1.9, P=0.01)
      • RR of death with albumin vs. saline 1 (95% C.I. 0.78-1.29, P=0.96)
  • Secondary outcomes: 
    • Pulmonary oedema
      • 1.3% vs. 0.6% vs. 0.6%
    • Mortality at 4 weeks significantly worse in bolus groups
      • 12.2% vs. 12.0% vs. 8.7%
  • In patients with severe hypotension (n=29)
    • No significant difference in mortality comparing albumin vs. saline bolus
      • 69% vs. 56% (RR 1.23, 95% C.I. 0.7-2.16, P=0.45)
  • Sub-group analysis
    • Fluid bolus resulted in either no benefit or harm in all subgroups

Authors’ Conclusions

  • In a resource limited setting in Africa, children with severe febrile illnesses and impaired perfusion, who were treated with fluid boluses had a significantly increased mortality compared with controls


  • Randomised controlled trial
  • Allocation concealment maintained
  • Minimal loss to follow up
  • Multi-centre


  • Inclusion criteria non-specific for diagnosis of shock. This may have led to a number of patients without shock receiving large fluid boluses. 
  • Change of study protocol halfway through study
  • Concerns over external validity (applying this trial to the population that I see)
    • Using a transfusion threshold of 5 meant that a number of patients were significant anaemia were given large volume of fluids leading to haemodilution. We are not told if patients had their haemoglobin re-measured post fluid bolus. 
    • Large number of patients had malaria, that is characterised by obstruction of the microcirculation with infected erythrocytes. Fluid boluses have little effect and can increase the risk of complications (see the review in the external links for more info)

The Bottom Line

  • In an African setting, febrile children with impaired perfusion, had an increased mortality if they were treated with a fluid bolus compared with no fluid bolus. The population studied had a high incidence of malaria and severe anaemia. They were managed with a low transfusion threshold and without critical care facilities. This is very different to the population that I manage. Therefore I do not think that these results can be applied to my patients. 

External Links


Summary author: @davidslessor
Summary date: 21st June 2015
Peer-review editor: @stevemathieu75


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.