Early goal-directed nutrition versus standard of care in adult intensive care patients: the single-centre, randomised, outcome assessor-blinded EAT-ICU trial

Allingstrup. Intensive Care Medicine 2017; 43:1637–1647. doi:10.1007/s00134-017-4880-3

Clinical Question

  • In critically ill adult patients, does early goal-directed nutrition (EGDN) during ICU stay compared to standard care nutrition delivery result in improved physical quality of life at 6 months?


  • The importance of nutritional support during critical illness in the ICU is well established. However, the various aspects of delivering this nutritional care is unclear because of diverging conclusions from recent trials. This may contribute to the variations seen in clinical practice and in national and international guidelines
  • As an example, ESPEN (European) guidelines recommend early-initiation of parenteral nutrition (PN) whilst ASPEN (American and Canadian) guidelines recommend later initiation of PN in adults who are receiving insufficient enteral nutrition in the intensive care unit (ICU)
  • Individualised goal-directed nutrition therapy with measured requirements has been proposed as one of several potentially beneficial strategies. EAT-ICU is the first randomised trial to assess the effect of individualised energy and protein supply to ICU patients, based on indirect calorimetry and 24-h urinary urea excretion (nitrogen balance) on physical quality of life


  • Single centre randomised controlled trial
  • Blinded outcome assessment
  • Computer generated randomisation schedule with varying block sizes (prepared by 2 people independent of the trial)
  • Allocation was concealed by consecutively numbered sealed opaque envelopes
  • Randomised 1:1, stratified according to presence or absence of active haematologic malignancy
  • Patients were randomised as soon as possible, with the intervention commenced at the latest 24 hours after admission to the ICU
  • 200 patients were required to demonstrate a 15% relative reduction in physical component summary (PCS) score at 6 months, with a significance level 0.05 and 80% power


  • Single ICU (Copenhagen University Hospital) in Denmark
  • June 2013- October 2016


  • Inclusion: Acutely admitted, mechanically ventilated adult patients with a central venous catheter (for PN if required), expected to stay longer than 3 days in ICU
  • Exclusion mostly based on any particular nutrition requirements
    • Contraindications to use enteral nutrition or parenteral nutrition, e.g. hypersensitivity towards fish-, egg or peanut protein, or any of the active substances in the PN products
    • Receiving a special diet
    • Burns > 10% total body surface area
    • Severe hepatic failure (Child-Pugh class C) or severe hepatic dysfunction: Bilirubin ≥ 50 µmol/l (3 mg/dl) + alanine aminotransferase ≥ 3 times upper reference value
    • Traumatic brain injury
    • Diabetic ketoacidosis or hyperosmolar non-ketotic acidosis
    • Known or suspected hyperlipidemia
    • BMI ≤ 17 or severe malnutrition
    • Pregnancy
    • The clinician finds that the patient is too deranged (circulation, respiration, electrolytes etc.) or that death is imminent
  • Screened: 2265 patients admitted to the ICU. 586 met inclusion and 383 were excluded. 203 (35%) were randomised (102 to EGDN and 101 to standard care)
    • 1 in each group died before randomisation and 1 in each group withdrew consent for data, therefore, 199 patients were analysed (100 in EGDN and 99 in standard care)
  • Baseline characteristics: no statistically significant difference between groups (EGDN vs Standard care)
    • Most common source of ICU admission was a hospital ward: 45% vs 38%
    • Admission type:
      • Medical: 52% vs 43%
      • Emergency surgery: 43% vs 54%
    • Age, years, median (IQR): 63 (51-72) vs 68 (52-75)
    • SAPS II score, median (IQR): 8 (6-11) vs 8 (5-10)


  • Early Goal Directed Nutrition
    • Calculated nutrition requirements and delivery were based on:
      1. Energy delivery directed by indirect calorimetry (IC) at randomisation and every other day
      2. 24-h urinary urea excretion (nitrogen use) was assessed daily and converted to metabolic protein consumption using Bistrian’s equation
    • Nutrition provision was titrated to IC and nitrogen balance with the aim to meet 100% of energy and protein needs on the first full trial day and for the duration of ICU stay (to a maximum of 90 days)
    • Protein was provided as at least 1.5 g/kg/day at all times during admission, regardless of urea excretion
    • Energy from propofol was included in energy calculations
    • Enteral nutrition was initiated within 24 h of randomisation and supplemented with PN if necessary to reach goal requirements
      • In case of sustained hyperglycaemia (defined as insulin requirement of at least 5 IU/h for > 12 consecutive hours), glucose was reduced and at a plasma urea above 20 mmol/l, protein was reduced by 0.2 g/kg/day
    • Supplemental PN was used if energy and protein needs could not be met by EN alone


  • Standard Care
    • Energy requirements calculated by 25/kcal/kg/day
    • EN commenced within 24 hours of randomisation and gradually increase as tolerated
    • If on day 7 energy needs were not met, supplemental PN was provided

Management common to both groups

  • The same EN and PN solutions
  • The same blood glucose level (BGL) aims (6-10 mmol/l)
  • Supplemental trace elements and vitamins based on measurements
  • The same gastric residual volume (GRV) cut off and protocol for management (based on usual practice in the ICU)
  • The same mobilisation protocol (based on usual practice in the ICU)


  • Primary outcome: physical quality of life 6 months after randomisation as assessed by the physical component summary (PCS) score of the Medical Outcomes Study 36-item short form health survey version did not differ between the two groups
    • Mean PCS score 22.9 vs standard care: 23.0; p = 0.99
    • Similar results in the predefined sensitivity analyses of the complete case population, the per-protocol populations, the two predefined subgroups and in the analyses adjusted for the stratification variable (haematological malignancy) and the predefined baseline variables (age and SOFA score)
  • Secondary outcomes
    • Mortality at day 28, day 90 and at 6 months: no statistical difference between groups
      • day 28: 20% dead in EGDT vs 21% in control group
      • day 90: 30% dead in EGDT vs 32% in control group
      • 6 months: 37% dead in EGDT vs 34% in control group
    • Cumulative energy and protein balances at day 1, 3 and 7 and over the course of the ICU stay was higher in the EGDN group
      • day 1: -211 vs -1011 kcal & 0.07 vs -0.7 g/kg protein
      • day 3: -220 vs -924 kcal & -0.59 vs -0.83 g/kg protein
      • day 7: -298 vs -702 kcal & -0.65 vs -0.75 g/kg protein
      • last study day: -249 vs -747 kcal & -0.56 vs -0.65 g/kg protein
        • all p<0.0001
    • Insulin requirement: the EGDN group received a greater median (IQR) dose of insulin and had a greater proportion of BGL ≥ 15 mmol/L
      • Insulin: 86 (2-530) vs 0 (0-39), p=0.008
      • BGL ≥ 15 mmol/L: 52% vs 25%, p=0.0001
    • Episodes of hyperglycaemia: More patients in the EGDN group experienced at least one episode of hyperglycaemia
    • No difference in:
      • length of stay among survivors in ICU or hospital
      • new organ failure
      • time to any infection or type of nosocomial infection
      • rates of hypoglycaemia

Authors’ Conclusions

  • The EGDN strategy resulted in greater energy and protein delivery in the ICU compared to standard care, but no differences were observed in PSC score at 6 months or any other clinically important outcomes


  • A highly relevant clinical question and hot research topic
  • The primary outcome was a long term, physically focused outcome
  • Study design attempts to address and provide titrated nutrition based on metabolic changes that occur during critical illness. This is only attempted in a few previous RCTs
  • Provides further evidence about the utility of IC in critical illness


  • This is a small, single centre study, and so the results may not be generalisable to all populations
  • The primary outcome was a long term outcome, but the intervention was early and very short duration. It may not be biologically plausible that the intervention would effect the outcome
  • Missing data was imputed. It seemed to be well understood and conducted by the authors but there is always a risk of inappropriate imputation
  • The patients may not have been likely to benefit from intervention as they were not long stay patients (7 days)
  • Early nutrition intervention when metabolic processes are aimed to mobilise endogenous energy stores may place the patient under further metabolic stress

The Bottom Line

  • In critically ill adults, EGDN provided early in ICU stay, and for a short duration, resulted in better nutritional outcomes but there was no difference in clinically important outcomes
  • This study provides a base from which future research can be conducted into questions such as when nutrition should be commenced and how metabolic processes differ during critical illness

External Links


Summary author: Emma Ridley
Summary date: 18th October 2017
Peer-review editor: Steve Mathieu and Duncan Chambler

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.