NICE tells clinicians not to treat a child's fever just to bring the number down

The UK's national clinical guideline on fever in children under five contains an instruction that surprises most parents: "Do not use antipyretic agents with the sole aim of reducing body temperature in children with fever" [s1]. What it recommends instead is to consider paracetamol or ibuprofen in children with fever "who appear distressed", and to continue only as long as the child appears distressed [s1].

The distinction is between treating a symptom and treating a reading. The guideline treats the child's discomfort as the target and the thermometer as information.

What the guideline actually says

NICE guideline NG143 was published in November 2019 and last updated in November 2021 [s1]. Its antipyretic recommendations are short and mostly negative.

Antipyretic agents do not prevent febrile convulsions and should not be used for that purpose [s1]. Tepid sponging is not recommended for treating fever [s1]. Children with fever should be neither underdressed nor over-wrapped [s1]. When paracetamol or ibuprofen is used, the guideline says to continue only while the child appears distressed, to consider switching to the other agent if distress is not relieved, not to give both simultaneously, and to consider alternating them only if distress persists or recurs before the next dose is due [s1].

The guideline also validates something parents are often made to feel sheepish about: reported parental perception of a fever "should be considered valid and taken seriously by healthcare professionals" [s1]. On measurement it is specific — forehead chemical thermometers are described as unreliable and are not to be used by healthcare professionals, and the oral and rectal routes are not to be used routinely in children aged 0 to 5 [s1].

For care at home following contact with a healthcare professional, NICE lists the circumstances in which parents should seek further advice: the child has a fit, develops a non-blanching rash, seems less well than when advice was last sought, the carer is more worried than before, the fever lasts five days or longer, or the carer feels unable to cope [s1].

The trial behind the "distress, not temperature" framing

The evidence that antipyretics reliably lower temperature is not in doubt. What the trials repeatedly fail to show is that lowering it makes the child feel better.

The PITCH trial randomised children aged 6 months to 6 years with axillary temperatures between 37.8°C and 41.0°C, managed at home in England, to paracetamol plus ibuprofen, paracetamol alone, or ibuprofen alone [s3]. On temperature, the combination worked: it produced 55 minutes less time with fever in the first four hours than paracetamol alone (95% CI 33 to 77 minutes, P<0.001) and 4.4 hours less over 24 hours (95% CI 2.4 to 6.3, P<0.001), and 2.5 hours less over 24 hours than ibuprofen alone (95% CI 0.6 to 4.4, P=0.008) [s3].

On how the children actually felt, it did not. The trial's second primary outcome was the proportion of children reported as normal on a discomfort scale at 48 hours, and the authors report no benefit for discomfort or other symptoms — while noting that statistical power was low for those outcomes [s3]. Adverse effects did not differ between groups [s3].

That is the shape of the evidence NICE is responding to: a demonstrable effect on the number, an unproven effect on the child.

The adult version of the question

Fever in critical illness poses the same question with higher stakes, and it was tested directly. The HEAT trial randomised 700 intensive care patients with a temperature of 38°C or above and known or suspected infection to 1 g of intravenous paracetamol or placebo every six hours, continued until ICU discharge, resolution of fever, cessation of antimicrobial therapy, or death [s2].

The primary outcome, ICU-free days to day 28, was 23 days in the paracetamol group and 22 in the placebo group — a Hodges-Lehmann absolute difference of 0 days (96.2% CI 0 to 1, P=0.07) [s2]. By day 90, 55 of 345 patients on paracetamol (15.9%) and 57 of 344 on placebo (16.6%) had died, a relative risk of 0.96 (95% CI 0.66 to 1.39, P=0.84) [s2].

The trialists' conclusion was that early paracetamol for fever due to probable infection did not affect the number of ICU-free days [s2]. It is a null result in the population where a benefit, if one existed, should have been easiest to detect.

What is still unknown

None of this establishes that fever is beneficial, a claim sometimes made on the strength of these same trials. HEAT was not powered to detect small mortality differences and its confidence interval is wide [s2]. PITCH was underpowered for the discomfort outcome its authors most wanted to measure [s3]. What the evidence supports is narrower: that reducing a temperature has not been shown to improve the outcomes that matter, which is why a guideline can recommend against treating the number on its own.

None of the above is guidance for a particular child or patient. NG143 covers children under five in the UK, its recommendations sit inside a full clinical assessment framework, and the same guideline lists specific circumstances requiring urgent review [s1]. Decisions about medication for a feverish child belong with a clinician.