HEAT Header Image

Acetaminophen for Fever in Critically Ill Patients with Suspected Infection

Young P. NEJM 2015; published on line first 5th October 2015. DOI: 10.1056/NEJMoa1508375

Clinical Question

  • Does the regular administration of paracetamol to critically ill patients with fever and known or suspected infection, affect the number of ICU-free days?


  • Randomised controlled study
  • Intention to treat analysis
  • Doubled blinded
  • Block randomisation with block size of six and stratified according to participating centre
  • Sample size of 700 patients provided 80% power to detect an absolute difference of 2.2 ICU-free days at 28 days after randomisation, at an alpha level of 0.05
  • A Wilcoxon rank-sum test was used for the primary outcome analysis. A proportional-odds analysis with the number of ICU-free days categorised as 0 to 7 days, 8-14 days, 15 to 21 days, or 22 to 27 days was used for the subgroup analysis


  • 23 adult general ICUs in Australia and New Zealand
  • February 2013 – July 2014


  • Inclusion:
    • aged > 16 years with a temperature of >38°C + known or suspected infection + receiving antibiotics
    • Enrolment within 12 hours
  • Exclusion: Acute brain disorders; liver dysfunction (AST or ALT > 5x upper limit of normal or bilirubin > 2x upper limit of normal); post cardiac arrest where current or anticipated temperature control was required; rhabdomyolysis; pregnancy; previous enrolment
  • 700 patients were randomised and 690 were included in the analysis
  • Baseline characteristics similar, including age, existing conditions, APACHE II score, time from admission to randomisation, physiological characteristics, sepsis severity, organ support and causative organisms (62.5% vs 62.2%)


  • Intravenous paracetamol: 1g every 6 hours


  • Intravenous infusion of 5% dextrose (placebo): 100mls every 6 hours

In both groups

  • Patients received the study drug until 28 days after enrolment or until the occurrence of one of the following pre-specified cessation criteria: discharge from ICU, resolution of fever, cessation of antimicrobial therapy, death, or the development of a contraindication to the study drug
  • Rescue physical cooling was permitted if the body temperature rose to 39.5°C or higher
  • Use of open label paracetamol was permitted after the course of study medication was completed


  • Primary outcome:
    • median ICU-free days to day 28: no significant difference
      • 23 (IQR 13-25) in the paracetamol group vs 22 in the placebo group (IQR 12-25) CI 0-1; P=0.07
  • Secondary outcome: comparing patients who received paracetamol with patients who did not receive paracetamol
    • all cause mortality at 28 days: no significant difference
      • 13.9% vs 13.7%
    • all cause mortality at 90 days: no significant difference
      • 15.9% vs 16.9%
    • number of days alive from randomisation until 90 days
    • median length of stay in ICU (days): increased in nonsurvivors and reduced in survivors
      • nonsurvivors: 10.4 (IQR 4.1 – 16.9) vs 4.0 (1.7 – 9.4); P<0.001
      • survivors: 3.5 (IQR 1.9 – 6.9) vs 4.3 (2.1 – 8.9); P< 0.01
    • median length of stay in hospital (days): increased in nonsurvivors
      • nonsurvivors: 13.9 (IQR 7.1 – 22.2) vs 7.7 (2.9 – 17.0); P<0.001
      • survivors: 13.2 (IQR 7.7 – 25) vs 14.1 (8.1 – 27.1); P< 0.13
    • median organs support free days: no significant difference
      • mechanical ventilation: 27 vs 26  P=0.14
      • inotropes or vasopressors: 27 vs 27  P=0.36
      • renal replacement therapy: 28 vs 28  P=0.53
    • physiological and laboratory-related variables
      • mean and maximum axillary temperature at days 0 to 7:
        • lower overall mean peak body temperature in the paracetamol group
          • 38.4°C vs 38.6°C absolute difference (AD) -0.25°C; 95% CI -0.38 to -0.11; P<0.001
        • lower mean daily average body temperature in the paracetamol group
          • 37.0°C vs 37.3°C AD -0.28°C; 95% CI -0.37 to -0.19; P<0.0001
        • among patients in whom the study drug was discontinued owing to discharge from the ICU, 12.3% of the paracetamol group and 23% of the placebo group had a temperature of 38°C or higher on there last ICU day
      • liver dysfunction necessitating stopping study drug: lower in paracetamol group
        • 8.1% vs 9.9%
      • Others – no statistical difference
        • mean serum C-reactive protein (CRP) in ICU on days 1,3,5 and 7; proportion of patients in ICU with CK of > 5000 units on days 1,3,5 or 7; highest serum creatinine in ICU during first 7 days after randomisation
  • Subgroup analysis – no statistical difference
    • ICU-free days to day 28 in the following groups
      • presence or absence of septic shock
      • use or nonuse of aspirin
      • presence of absence of high fever (temperature >39°C in the 12 hours before enrolment
      • location of infection acquisition (community, hospital, or ICU)

Authors’ Conclusions

  • Early administration of paracetamol to treat fever due to probable infection does not affect the number of ICU-free days


  • A common question with important clinical outcome measures has been addressed
  • Protocol was reported before enrolment commenced
  • Statistical analysis plan was reported before the interim analysis was conducted
  • Allocation concealment and study drugs were masked by being packaged in an indistinguishable 100ml glass bottle
  • All analyses, including post-hoc analyses, were conducted before unmasking the study-group assignments
  • Only 10 patients (1.4%) were lost to follow up
  • The effects of other antipyretics including surface cooling and NSAIDs were considered and there was no difference between the two groups


  • ‘ICU-free days’ is a composite outcome combining mortality and ICU length of stay
  • 1/3 of patients in each group were exposed to paracetamol in the ICU after the course of the study drug administration
  • No information collected about the use of paracetamol before randomisation or after ICU discharge
  • Protocol deviations were high: 30% in paracetamol group and 28% in placebo group. Around a 1/3 of these were related to missed doses of study drug

The Bottom Line

  • Early administration of paracetamol to ICU patients to treat fever due to probable infection results in a moderate reduction in temperature but does not effect mortality at 28 or 90 days. The observation that it may result in a longer ICU stay in nonsurvivors and shorter ICU stay in survivors should be interpreted cautiously and only as hypothesis generating.

External Links


Summary author: @stevemathieu75
Summary date: 7th October 2015
Peer-review editor: @davidslessor
Conflicts of interest: None


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.