Skip to main content

Foundational Trials

FEAST

Image: Shutterstock / Riccard Mayer

Ampoule of adrenaline being lifted from a drug tray

Publication

  • Title: Mortality after Fluid Bolus in African Children with Severe Infection
  • Acronym: FEAST
  • Year: 2011
  • Journal published in: The New England Journal of Medicine
  • Citation: Maitland K, Kiguli S, Opoka RO, et al. Mortality after Fluid Bolus in African Children with Severe Infection. N Engl J Med. 2011;364(26):2483–2495

Context & Rationale

  • Background
    Rapid fluid bolus therapy has long been embedded in paediatric shock algorithms, largely extrapolated from physiology and higher-resource critical care practice. In African district hospital settings—where malaria, severe anaemia and limited access to intensive care are common—there was genuine equipoise regarding whether boluses improved outcomes or precipitated harm.
  • Research Question/Hypothesis
    Does fluid bolus therapy (albumin or saline) reduce mortality compared with no bolus in children with severe febrile illness and impaired perfusion? Secondarily, is albumin superior to saline?

Design & Methods

  • Research Question: In African children with severe febrile illness and impaired perfusion, does fluid bolus therapy (albumin or saline) compared with no bolus change all-cause mortality at 48 hours?
  • Study Type: Randomised, multicentre, controlled, open-label trial.
  • Population:
    • Age: Children aged >60 days and <12 years admitted with history of fever and/or abnormal temperature (pyrexia ≥37.5°C or hypothermia <36°C).
    • Core entry requirement: Severe illness AND impaired perfusion.
    • Severe illness (≥1): impaired consciousness (prostration or coma) and/or respiratory distress (deep breathing or increased work of breathing).
    • Impaired perfusion (≥1): capillary refill >2 seconds; lower limb temperature gradient; weak radial pulse volume; or severe tachycardia (HR >180 bpm if <12 months; >160 bpm if 12 months–5 years; >140 bpm if >5 years).
    • Strata:
      • FEAST A: impaired perfusion without severe hypotension at presentation (3-arm: albumin vs saline vs no bolus control).
      • FEAST B: hypotensive shock at presentation (2-arm: albumin vs saline; no control arm).
    • Severe hypotension (systolic BP): <50 mmHg (<12 months), <60 mmHg (12 months–5 years), <70 mmHg (>5 years).
    • Hypotensive shock definition: severe hypotension PLUS ≥1 sign of impaired perfusion.
  • Key Exclusions:
    • Severe acute malnutrition: visible severe wasting and/or kwashiorkor.
    • Gastroenteritis: >3 watery stools in the previous 24 hours.
    • Contraindications to intravascular volume expansion: chronic renal failure; pulmonary oedema (SpO₂ <90% plus bilateral basal crepitations).
    • Non-infectious causes of severe illness: trauma; burns; intoxication.
    • Prior isotonic volume expansion: already received an isotonic volume expander during the current illness.
  • Intervention:
    • Fluids tested: 5% human albumin solution vs 0.9% saline.
    • Bolus infusion time: all boluses were given over 1 hour.
    • FEAST A (3-arm):
      • Initial bolus: 20 mL/kg of allocated fluid (albumin or saline).
      • At 1 hour reassessment (based on clinical signs):
        • If impaired perfusion persisted: additional 20 mL/kg of allocated fluid.
        • If hypotensive shock developed: 40 mL/kg of allocated fluid.
        • If no impaired perfusion: maintenance fluids only.
    • FEAST B (hypotensive shock; 2-arm):
      • Initial bolus: 40 mL/kg of allocated fluid (albumin or saline).
      • At 1 hour reassessment (based on clinical signs):
        • If impaired perfusion persisted: additional 20 mL/kg of allocated fluid.
        • If hypotensive shock persisted: additional 40 mL/kg of allocated fluid.
        • If no impaired perfusion: maintenance fluids only.
    • Bolus limits/escalation: after the initial 2-hour protocol window, the only routine indication for further bolus was hypotensive shock; maximum bolus volume was 80 mL/kg per 24 hours (protocolised maximum).
  • Comparison:
    • No bolus control (FEAST A only): maintenance fluids only (“low volume maintenance fluids” per national guidelines), with no immediate volume expansion.
    • Rescue in control arm: if hypotensive shock developed, rescue 0.9% saline 40 mL/kg over 1 hour.
    • Prespecified comparisons: bolus (albumin+saline) vs no bolus; and albumin vs saline.
  • Blinding: Open-label; primary endpoint (mortality) is objective.
  • Statistics: Independent monitoring and prespecified stopping framework; the DSMB recommended halting recruitment early due to harm signal.1
  • Follow-up: Primary endpoint at 48 hours; secondary endpoints included outcomes to 4 weeks.

Key Results

This trial was stopped early for harm.

Outcome Intervention Comparison Effect p value / 95% CI Notes
48-hour all-cause mortality Saline bolus: 110/1047 (10.5%) No bolus: 76/1044 (7.3%) RR 1.44 p=0.01; 95% CI 1.09–1.90 Prespecified primary comparison
48-hour all-cause mortality Albumin bolus: 111/1050 (10.6%) Saline bolus: 110/1047 (10.5%) RR 1.01 p=0.96; 95% CI 0.78–1.29 Prespecified primary comparison
48-hour all-cause mortality Any bolus: 221/2097 (10.5%) No bolus: 76/1044 (7.3%) RR 1.45 p=0.003; 95% CI 1.13–1.86 Prespecified secondary (often-cited) comparison
4-week all-cause mortality Any bolus: 255/2097 (12.2%) No bolus: 91/1044 (8.7%) — p=0.004 Bolus groups vs no bolus
Pulmonary oedema or raised intracranial pressure Any bolus: 50/2097 (2.4%) No bolus: 18/1044 (1.7%) — p=0.12 Rare events; not the dominant signal
New neurological sequelae at discharge Any bolus: 44/2097 (2.1%) No bolus: 21/1044 (2.0%) — p=0.92 No signal of difference
  • Fluid bolus therapy increased early mortality compared with no bolus.
  • Albumin conferred no mortality benefit over saline.
  • Classic “fluid overload” syndromes were uncommon and did not explain the mortality signal.

Internal Validity

  • Randomisation and allocation: Large, multicentre randomised design with protocolised approaches and prospective outcome ascertainment.
  • Performance/detection bias: Open-label, but the primary outcome (mortality) is objective. Subsequent adjudication/secondary analyses explored mechanisms of harm.2
  • Early stopping: Recruitment was halted following DSMB recommendation because of excess mortality in bolus arms; early stopping can inflate effect estimates, but the direction of effect was consistent across bolus types.1

Conclusion on Internal Validity: High for the primary outcome signal (mortality) given randomisation and objective endpoint; mechanistic interpretation relies on post hoc analyses but is supported by multiple independent evaluations.23

External Validity

  • Population representativeness: Highly representative of severely ill children presenting to African hospitals with febrile illness and impaired perfusion (including high malaria and anaemia burden).
  • Applicability to high-resource PICU practice: Controversial. FEAST occurred in settings with limited access to invasive ventilation, vasoactive infusions, and continuous monitoring—resources that often co-exist with aggressive fluid resuscitation in high-income settings.
  • Important exclusions: FEAST excluded children with gastroenteritis and severe acute malnutrition; therefore FEAST should not be used as a blanket argument against IV rehydration in diarrhoeal dehydration or in SAM.

Conclusion on External Validity: Generalisability is strongest to similar low-resource settings and to shock phenotypes dominated by severe infection/malaria physiology; extrapolation to dehydration-driven hypovolaemia (especially in SAM) requires separate evidence (see GASTRO and GASTROSAM/GASTRO-SAM).56

Strengths & Limitations

  • Strengths: Large pragmatic RCT; clinically meaningful primary endpoint; contemporaneous comparison of two commonly used bolus fluids; robust follow-up to 4 weeks; clear and practice-changing signal.
  • Limitations: Open-label; early stopping; context-specific resource constraints; shock physiology in FEAST (often without severe hypotension) differs from classic “PICU septic shock” cohorts; excluded key real-world subgroups (gastroenteritis and severe malnutrition).

Interpretation & Why It Matters

  • Boluses Increased Mortality
    In this setting and population, bolus fluids (whether albumin or saline) increased early and 4-week mortality compared with no bolus, overturning a long-held assumption that boluses are uniformly beneficial in paediatric shock.
  • Mechanism: Not “Fluid Overload”
    Post hoc mechanistic analyses and later re-analyses suggest the excess mortality was largely due to cardiovascular collapse rather than pulmonary oedema/raised intracranial pressure as the dominant pathway.243
  • How GASTRO-SAM Changes Interpretation
    FEAST is often (mis)applied as a general “anti-fluid” argument. However, because FEAST excluded gastroenteritis and severe acute malnutrition, its findings should not be extrapolated to diarrhoeal dehydration—particularly in SAM. The later GASTRO trial (non-malnourished severe dehydration) and the 2025 GASTROSAM (GASTRO-SAM) trial in children with SAM and dehydrating gastroenteritis found no signal that IV rehydration strategies increased short-term mortality or fluid overload compared with cautious control regimens. This supports a more nuanced interpretation: FEAST cautions against liberal bolus resuscitation in severe infection/impaired perfusion in low-resource settings, not against appropriately targeted rehydration in dehydration phenotypes (including SAM).56

Controversies & Subsequent Evidence

  • Generalising to high-resource care
    • International guidelines in higher-resource contexts continue to recommend fluid boluses for paediatric septic shock, typically with close monitoring and escalation to vasoactive support when needed.89
  • Guidelines after FEAST
    • Post-FEAST, guidance has been heterogeneous and sometimes slow to incorporate the trial’s implications in low-resource settings, prompting ongoing debate about context-specific algorithms and the need for phenotyping shock.710
  • Re-analysis and timing of harm
    • Secondary evaluations (including time-to-death analyses and re-analyses) have reinforced that harm emerged early after bolus administration and aligns more with circulatory collapse than classic fluid overload syndromes.43
  • Evidence synthesis
    • Systematic reviews and later meta-analyses highlight that FEAST dominates the randomised evidence base in similar settings and that conclusions should be interpreted in light of population/setting differences and evolving supportive care.1112
  • What GASTROSAM adds (for FEAST interpretation)
    • GASTROSAM directly addresses a major FEAST “gap”: children with SAM and dehydrating diarrhoea. The absence of a mortality harm signal with IV rehydration strategies in that phenotype supports keeping FEAST’s conclusions tightly anchored to its enrolled population and shock physiology.6

Summary

  • Bolus fluids increased 48-hour mortality compared with no bolus (RR ~1.45 in prespecified combined analyses).
  • Albumin was not superior to saline for mortality.
  • Excess mortality was not explained by overt pulmonary oedema/raised ICP events; later analyses support cardiovascular collapse as a key pathway.24
  • Because FEAST excluded gastroenteritis and SAM, its findings should not be extrapolated to diarrhoeal dehydration or to SAM rehydration. GASTRO and GASTROSAM provide separate evidence in those phenotypes.56

Further Reading

Other Trials

Key Secondary Analyses / Re-analyses

Systematic Review & Meta Analysis

Guidelines

Notes

  • FEAST remains a uniquely influential RCT because few trials have randomised fluid boluses vs no bolus in comparable low-resource contexts.
  • Interpretation should be phenotype-driven: severe infection with impaired perfusion (FEAST) is not the same as dehydration-driven hypovolaemia (GASTRO/GASTROSAM).
  • Guideline recommendations may diverge by setting, monitoring capability, and availability of vasoactive support and ventilation.

Overall Takeaway

In African children with severe febrile illness and impaired perfusion, fluid bolus therapy (albumin or saline) increased early mortality compared with no bolus, with no benefit of albumin over saline. Subsequent mechanistic and re-analytic work supports cardiovascular collapse—not classic fluid overload—as a key pathway. Critically, FEAST excluded gastroenteritis and severe acute malnutrition; subsequent trials in dehydration phenotypes (including GASTROSAM/GASTRO-SAM) do not show the same mortality harm signal with IV rehydration strategies, reinforcing a phenotype- and setting-specific interpretation of FEAST. 2 3 6 5

Overall Summary

  • Bolus fluids increased 48-hour mortality compared with no bolus (RR ~1.45 for any bolus vs no bolus).
  • No mortality benefit of albumin over saline.
  • Mechanistic work suggests cardiovascular collapse—not overt fluid overload—drives much of the harm signal. 2 4
  • FEAST excluded gastroenteritis and SAM; GASTROSAM/GASTRO-SAM supports that FEAST should not be extrapolated to dehydration phenotypes (including SAM) as a blanket “no IV fluids” rule. 6 5

Bibliography