Classic Trial

Restriction of Intravenous Fluid in ICU Patients with Septic Shock

Meyhoff T. NEJM 2022; DOI: 10.1056/NEJMoa2202707

Clinical Question

  • In adult patients in ICU with septic shock, does a restrictive approach to fluid therapy compared to a standard care result in fewer deaths at day 90?


  • Fluid therapy in the critically unwell has been widely studied across many patient populations
  • Higher volumes of fluid in patients with sepsis and septic shock has been associated with harm, although a meta-analysis showed that the quality of evidence was very low
  • The ANDROMEDA SHOCK trial demonstrated in those with septic shock, that by 2 hours only ~25% of patients were deemed to still be fluid responsive
  • The CLASSIC pilot study showed that a restrictive fluid strategy was feasible and showed significantly less fluid was used in the first 5 ICU days in the restrictive group
    • Mortality at day 90 was an exploratory outcome and this was reduced in the restrictive group, although not significantly


  • Stratified, parallel-group, open-label, international, randomised controlled trial
  • Randomised 1:1 using a central web based system
    • Block sizes of 6 or 8
    • Stratified according to site and presence/absence of metastatic or haematological malignancy
  • Treatment group assignments not masked for clinicians, patients or investigators
    • Those on data safety monitoring committee, statisticians assessing the outcomes, and the management committee writing the abstract draft were blinded
  • Received assigned intervention from randomisation until ICU discharge (for a maximum of 90 days)
    • If re-admitted within 90 days then treatment assignment continued
  • Data collected from medical records and registries
    • Patients or relatives contacted if further information needed
  • Pre-specified subgroups included need for respiratory support, AKI, Lactate > 4 mmol/L, weight > 76kg, > 30mls/kg of fluids administered in 24 hours prior to randomisation
  • Power calculation
    • 1554 patients needed to show a 7% absolute reduction in 90 day mortality from a baseline of 45%
    • Type I and II error rates of 5% and 20% respectively
    • Based on pilot study, prior RCTs and systematic reviews
  • Registered with


  • 31 ICUs in Europe
  • November 2018 to November 2021


  • Inclusion:
    • Age ≥ 18
    • Septic Shock according to Sepsis-3
      • Suspected or confirmed infection, plasma lactate > 2mmol/L and vasopressor requirement
    • 1L of fluid administered in the prior 24 hours
    • Within 12 hours of screening
  • Exclusion:
    • Septic shock > 12 hours
    • Life threatening bleeding
    • Burns > 10% TBSA
    • Known Pregnancy
  • 2223 screened –> 1554 randomised
    • 770 to restrictive group
    • 784 to standard group
  • Comparing baseline characteristics of restrictive vs. standard group
    • Age (yrs): 71 vs 70
    • Male (%): 59.9 vs 58.2
    • Median time from ICU admission to randomisation (hrs): 3 vs 3
    • Median predicted mortality (%): 40 vs 40
    • Source of admission (%)
      • ED: 39.3 vs 38.5
      • Wards: 34.2 vs 38.7
    • Source of infection (%):
      • GI: 36.8 vs 38.3
      • Respiratory: 27.7 vs 26.5
      • Urinary: 15.8 vs 17.1
    • Median highest lactate (mmol/L): 3.8 vs 3.9
    • Median highest dose of noradrenaline (μg/kg/min): 0.25 vs 0.23
    • Median volume of IV fluids in prior 24 hours (L): 3.2 vs 3.0
    • Systemic glucocorticoid use (%): 28.6 vs 29.1
    • Use of respiratory support (NIV, IMV): 52.6 vs 48.6


  • Restrictive Group
  • Fluids could be given for:
    • Severe hypoperfusion:
      • Lactate > 4 mmol/L
      • MAP < 50 mmHg (with or without vasopressors/inotrope)
      • Mottling beyond kneecap
      • Urine Output < 0.1ml/kg (only within first 2 hours post randomisation)
      • If met any of these criteria then 250-500mls bolus of crystalloid could be given followed by re-evaluation
    • Overt losses (e.g. D+V) but only to correct for loss and not more
    • If oral or enteral route contraindicated for water or electrolyte solutions then IV fluids could be given to correct electrolyte derangements and to ensure a 1L/24hrs total intake
      • Fluids with medications and nutrition count as input
      • Fluids for medications should be reduced to lowest possible volume


  • Standard Care
    • No upper limit for use of fluids
    • IV fluids should be given based on SSC guidelines in case of circulatory impairment
    • IV fluids should be given to replace observed or expected losses to correct dehydration or electrolyte derangements
    • IV fluids should be given as maintenance in places with protocols that recommend this

Management common to both groups

  • Isotonic crystalloids used
    • Albumin only allowed following abdominal paracentesis
  • Concomitant interventions for septic shock detailed in trial protocol (appropriate antibiotics, noradrenaline as a vasopressor, source control and RRT for routine indications)
  • All other treatment including diuretic use at discretion of treating clinicians


  • Primary outcome:
  • Death by day 90
    • Restrictive 42.3% vs Standard 42.1%
    • Adjusted absolute difference 0.1 (95% CI -4.7 to 4.9), p = 0.96
  • Secondary outcomes:
    • No significant difference in:
      • Number of Serious Adverse Events
      • Number of days alive without life support
      • Number of days alive and out of the hospital

Authors’ Conclusions

  • In adult ICU patients with septic shock, IV fluid restriction did not result in fewer deaths at 90 days compared to standard IV fluid therapy


  • Randomised
  • Multi-centre trial involving 31 ICUs across 8 European countries increases external validity
  • Pre-published statistical analysis plan
  • Intention to treat analysis
  • Minimal loss to follow up (0.6%)
  • Balanced baseline characteristics
  • Unwell patient cohort
  • Achieved a separation in fluids administered via trial protocol


  • Unblinded
  • The most common infection source was GI – this may have necessitated large volumes of replacement fluid if diarrhoea or vomiting were prominent features
  • Powered for a 7% reduction in mortality
  • Large volumes of fluid given outside of the trial protocol
    • Nearly 3L of fluid given to each group in 24 hours prior to randomisation
    • The total difference in fluids administered by day 5 was only ~2L (8.8L vs 10.8L), with a difference of only ~800mls in the cumulative day 5 fluid balance
    • These similarities between the groups may result if the standard care delivered in ICUs is already a relatively restrictive (e.g. no maintenance) fluid strategy
      •  This potentially biases the results towards the null
  • Large numbers of protocol violations (% patients with at least one violation)
    • 21.5% in restrictive group vs 13.0% in standard group
      • In restrictive group 64% of violations were for fluid bolus use without criteria being met
        • Violations resulted in median of 97ml fluid administered/day
      • In standard group 100% of violations were for no IV fluid being given on any single day in ICU
        • This was 13% of all patients in the standard group
    • However, there was no difference in outcome when violations removed in per-protocol analysis

The Bottom Line

  • This study did not confirm my pre-existing biases, in that a restrictive fluid strategy was not shown to be superior, with respect to 90 day mortality, to standard fluid management currently being practiced in European ICUs
  •  However, the 5 day cumulative fluid balance in the standard group was not much greater than the restrictive group. Therefore care is still required to ensure that every patient in ICU has a vigilant and considered fluid management strategy. This includes the de-escalation or evacuation phase of fluid therapy.

External Links


Summary author: George Walker @hgmwalker
Summary date: 17th June 2022
Peer-review editor: @davidslessor

Picture by: Pixabay/Pexels

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.