Name: Fever and Sickness Behaviour: A Regulated Defence You Treat for Comfort, Not for the Number
Creator: Realised
Published: 2026-08-01

Summary

Fever and the pull to rest are regulated, evolved defence programmes rather than malfunctions — but the best human trial of treating fever found no harm from doing so, which means the honest advice is conditional: you don't need to fight the fever to zero, you treat for comfort rather than the thermometer, and a specific set of red flags means medication and urgent care, not endurance.

Why Moderate

The entry overall carries a Moderate rating, and the honest reason is the split beneath it.

Why not the tier above (Strong Evidence): the central practical thesis — that reflexively suppressing fever and sickness behaviour carries a real cost — is not established by strong human evidence. The human suppression trials are small, early-stopped, open-label or unreplicated, and the single largest, best-designed trial (HEAT) is null. A thesis whose best supporting human trials are Emerging-grade cannot rate Strong, whatever the mechanistic beauty behind it.

Why not the tier below (Emerging): too much of the entry's load-bearing content is better than that. The set-point physiology and the fever/hyperthermia distinction are Foundational. The red-flag framework is national clinical guideline. The febrile-seizure null is a Cochrane review of randomised trials. The sleep-susceptibility finding is a well-designed human challenge study. The entry's practical guidance — treat for comfort, know the red flags — rests on its strongest components, not its weakest.

Per-sub-area split: fever as regulated, conserved defence — Foundational for the physiology, Strong within the animal experiments, with the animal-to-human transfer explicitly discounted. Immune enhancement at febrile temperatures — Moderate (mechanistically solid, clinically unproven). Human cost of suppression — Emerging (Schulman, Graham, Earn), counterweighted by one Strong null (HEAT) and one Moderate surrogate-endpoint finding (Prymula). Febrile-seizure prevention — Strong (against antipyretics working). Sickness behaviour's adaptive function — Moderate (mechanism proven in animals, function inferred, no human outcome trial). Sleep and susceptibility — Strong within its design limits. Red flags — Foundational.

Practical takeaway

• Treat the person, not the thermometer. The question is never "what is the number?" but "how do they feel, and are any red flags present?" A fever of 38.5°C in a comfortable, drinking, responsive adult needs fluids, rest and time — medicating it is optional. The same number in a distressed, aching, sleepless person is a good reason to take paracetamol or ibuprofen, and doing so is legitimate treatment, not weakness or sabotage.
• Comfort is a real goal. If symptom relief lets you sleep, drink and eat, take it without guilt — sleep is itself immune-relevant, and the best human trial found no recovery penalty from paracetamol in far sicker patients than you. The conditional runs the other way too: if you feel tolerable without medicating, there is no obligation to chase a normal number on the thermometer.
• Follow the sickness behaviour. The exhaustion and withdrawal are the body's instruction to stop, and the honest evidence position is that overriding them is an unpriced trade, not a free lunch. Cancel what can be cancelled. If you cannot afford the day off, that is an economic reality, not a moral failure — but take the version of rest available to you rather than treating "functioning normally" as the goal of medication.
• Don't medicate in order to mingle. Feeling artificially fine is not being non-infectious. If antipyretics are what make going to the office or the party possible, the modelling evidence suggests the cost of that choice lands on other people.
• With children: treat distress, not the number; do not give paracetamol and ibuprofen at the same time; never give aspirin to children or teenagers with a viral illness; and know that keeping the temperature down does not prevent febrile seizures — so the reason to medicate a febrile child is their comfort, which is reason enough.
• Skip the pre-emptive dose at vaccinations. Don't give paracetamol routinely before or at an infant's vaccination appointment to head off a fever that hasn't happened; if the child is genuinely febrile and distressed afterwards, treating them then is fine. None of this changes whether or when to vaccinate.
• Count your paracetamol. Combination cold-and-flu remedies each contain it; stacking two products plus plain paracetamol is the classic route to accidental overdose, and the margin between the therapeutic and liver-toxic dose is narrow. One paracetamol-containing product at a time, and read the label of anything added.
• Know the difference between fever and overheating. Someone hot after exertion in heat, or after stimulants, may be in hyperthermia — antipyretics will not help, cooling and emergency care will.

Evidence detail

Why This Entry Exists

Fever sits at the centre of one of the most confidently argued and most poorly cited disputes in health content. One camp treats a raised temperature as a defect to erase — reach for the paracetamol at 37.9, alternate ibuprofen, get the number on the chart back to normal. The other camp treats fever as sacred — "your body's wisdom", never suppress, let it burn. Both camps quote studies. Both camps quote them badly.

The actual evidence base is stranger than either story. The mechanistic case that fever is a deliberate, regulated defence — an actively defended rise in the body's temperature set-point, conserved across vertebrates for something on the order of hundreds of millions of years — is genuinely strong. The human clinical evidence that suppressing fever costs you anything is genuinely weak: small trials, non-significant results dressed up as dramatic ones, and one large, well-run trial that found treating fever in the sickest patients on earth made no difference at all, in either direction. Meanwhile the standard justification parents are given for aggressive fever treatment in children — preventing febrile seizures — turns out not to be supported by the systematic-review evidence.

This entry also covers sickness behaviour: the lethargy, appetite loss and social withdrawal that arrive with infection. These are not weakness or a failure of willpower; they are a cytokine-orchestrated programme with a plausible adaptive function. The practical implication — rest is the intervention, not the thing you medicate your way past — is the most undersold message in this whole area, partly because nobody makes money from telling you to lie down.

One register note before anything else, because it is load-bearing for this entire entry: nothing here means "don't take medicine." Comfort is a legitimate treatment goal — the UK's own paediatric fever guideline says to treat the distressed child, not the thermometer, and that is permission, not compromise. And a small, specific set of situations (listed in Risks, unmissably) demands medication or urgent care regardless of any philosophy about the body's wisdom.

What bad advice this protects against, in all directions:
• "Never treat a fever — you're fighting your body's healing." Wrong on the best available human evidence: the largest, blinded trial of treating fever in infected intensive-care patients found no harm from paracetamol on any outcome. Wrong on comfort: a miserable, sleepless, aching person is not being served by fever purism, and relief that lets them sleep may itself help recovery. And dangerous at the edges: this advice, taken seriously, delays care in the situations where fever is a red flag.
• "Get the temperature down — otherwise your child could have a seizure." The Cochrane systematic review found antipyretics do not prevent febrile seizure recurrence. This is the standard folk rationale for aggressive fever treatment in children and it is not supported. (Febrile seizures still require assessment — see Risks — but the thermometer-chasing they motivate does not deliver the protection promised.)
• "Take a max-strength cold-and-flu remedy and power through your day." This overrides sickness behaviour — the evolved instruction to stop — with no evidence that doing so is free, some weak evidence it carries a cost, a modelling argument that it spreads infection to others, and a real overdose hazard when multiple combination remedies each contain paracetamol. Sometimes powering through is an economic necessity, not a choice, and this entry does not sneer at that. But it should be recognised as a trade, not a triumph.
• "Fever means your immune system is working — a high fever is a good sign, ride it out." A temperature at or above roughly 40°C, a fever in an infant under three months, a fever with a non-blanching rash or stiff neck, or a fever during chemotherapy is not a wellness opportunity. It is a medical situation. The "fever is your friend" framing gets people to hospital late, and this entry exists partly to make sure Realised never contributes to that.
• "That study proved cold medicine makes your cold last longer." The most-quoted study here (Graham 1990) did not show significantly prolonged viral shedding — that was a non-significant trend in a trial of about fifteen people per arm, unreplicated in thirty-five years. Citing it as proof is building a case, not reading evidence.

OWNS: the biology of fever as a regulated set-point rise; the fever-versus-heatstroke distinction; the human evidence for and against routine fever suppression; sickness behaviour and the case for rest; febrile-seizure evidence; fever red flags and the comfort-treatment doctrine.

DEFERS: the general decision framework for when to blunt any defensive symptom (nausea, cough, diarrhoea, pain) to defensive_symptoms_and_when_to_blunt_them; what actually prevents and shortens colds to common_cold_prevention_and_treatment; how infection and inflammation wreck sleep and what to do about it to illness_inflammation_and_sleep_disruption; day-to-day immune resilience to immune_function_cross_pillar_optimisation; deliberate, chosen stressors and adaptation to hormesis_and_adaptive_stress.

Evidence

Read the tiers, not the thesis. The story in this entry is seductive — evolution built the fever, so leave it alone — and the evidence under each part of that story ranges from rock-solid to a fifty-year-old lizard experiment. Every claim below carries its evidence type and its strength label inline. The pattern to notice: the mechanism is well-supported, the human clinical cost of suppression is weak and mixed, and the strongest single human trial points against the purist reading.

Sub-area 1: Fever is an evolved, regulated defence

1. Infected ectotherms seek heat, and preventing that heat-seeking kills them — Strong Evidence within its animal domain; evidence type: reptile experimental (not human). Kluger and colleagues infected desert iguanas with Aeromonas hydrophila; infected animals behaviourally selected warmer microclimates, raising body temperature by roughly 2°C, and animals held at cool ambient temperatures — physically prevented from developing behavioural fever — died at much higher rates (Kluger, Ringler & Anver, Science, 1975). A separate companion paper showed the same mortality increase when fever was blocked pharmacologically with sodium salicylate (Bernheim & Kluger, Science, 1976), and a third paper demonstrated behavioural fever in goldfish (Reynolds, Casterlin & Covert, Nature, 1976). These are three distinct experiments and are routinely welded into one in secondary sources; they are kept separate here deliberately. Confounds: ambient temperature changes the pathogen's growth conditions as well as the host's immunity, so the design cannot fully separate "defence improved" from "bacterium grew worse when warm"; samples were small; and an ectotherm warms itself for free while a human fever costs roughly 10–13% extra metabolic rate per degree — the cost–benefit does not transfer automatically. Cui bono: evolutionary-medicine authors and anti-pharma content profit from stretching a lizard experiment into human clinical advice; the multi-billion over-the-counter antipyretic market profits from the opposite default, that every fever needs a product.

2. Fever is phylogenetically ancient and is a regulated set-point rise, categorically different from heatstroke — Foundational; evidence type: evolutionary reasoning plus human-verified physiology. The febrile response is conserved across warm- and cold-blooded vertebrates over an estimated 600 million years (Evans, Repasky & Fisher, Nature Reviews Immunology, 2015 — note this figure is a comparative-phylogeny inference, not a measurement). The set-point mechanism itself is textbook-level human physiology: pyrogenic cytokines drive prostaglandin E2 production in the hypothalamus, which raises the thermoregulatory target; the body then defends that higher target with its normal machinery — which is why a rising fever feels cold and makes you shiver. Antipyretics lower the set-point; they do not "cool" you. Hyperthermia — heatstroke, drug toxicity, malignant hyperthermia — is the opposite condition: a normal set-point overwhelmed by heat load. Antipyretics do not work in hyperthermia; it requires active cooling and is an emergency. This distinction is load-bearing and any account that blurs it is dangerous. Cui bono: nobody — this distinction has no commercial constituency, which is partly why wellness content omits it: it complicates the story.

Sub-area 2: What febrile temperatures do to immunity

3. Fever-range temperatures measurably enhance specific immune functions — in mice and dishes — Moderate; evidence type: in-vitro and rodent work synthesised in review, not human clinical-outcome evidence. At 38–40°C: enhanced trafficking of lymphocytes into lymph nodes (an IL-6-dependent effect), heat-shock protein induction, improved neutrophil, natural-killer and dendritic-cell function, and direct temperature inhibition of some pathogens' replication (Evans, Repasky & Fisher, 2015). The honesty label matters more than the list: there is no human trial showing that these enhancements translate into faster clearance or better clinical outcomes in people. The mechanism is real; the clinical payoff is inferred. Stating this mechanism as proven human benefit is the single most common overclaim in this literature, and the HEAT trial below is the reason to resist it. Confounds: heating cells in a dish is not fever (no cytokine milieu, no set-point); standard-housed laboratory mice are mildly cold-stressed relative to their thermoneutral zone, so their immune baselines differ from ours. Cui bono: hyperthermia-device and "fever therapy" sellers profit from mechanism-to-clinic inflation; pharma has little incentive to fund a trial asking whether its own antipyretic blunts recovery — a genuine reason this human evidence is thin rather than negative.

Sub-area 3: What happens when humans suppress fever — the trials, honestly reported

4. Aggressive fever suppression in trauma intensive care: a dramatic-looking signal that did not reach significance — Emerging; evidence type: human randomised trial, small, stopped early. Schulman and colleagues randomised trauma ICU patients to aggressive antipyresis (paracetamol above 38.5°C plus physical cooling above 39.5°C) versus permissive treatment (nothing until 40°C). Eighty-two patients enrolled. Deaths: 7 in the aggressive arm versus 1 permissive — p=0.06, not significant. Per-patient infections: also not significant (p=0.26). The trial stopped at its first interim analysis because of the mortality imbalance — and trials stopped early systematically exaggerate effect sizes, because the stopping rule selects extreme interim results (Schulman et al., Surgical Infections, 2005). A further confound: the aggressive arm received external cooling as well as drug, and cooling a patient whose set-point is raised triggers shivering and a catecholamine surge — the harm signal, if real, may belong to the cooling rather than the paracetamol. This is a legitimate hypothesis-generating result. Presented as "treating fever killed seven people versus one," it is a miscitation. Cui bono: "let it burn" content leans on this trial precisely because its raw numbers are dramatic and its statistics are not; ICU culture's historic default of suppressing every fever was institutionally convenient rather than evidence-led.

5. The counterweight, and the strongest human evidence in this entry: the HEAT trial found treating fever in infected ICU patients caused no harm and no benefit — Strong Evidence; evidence type: human randomised trial, large, multicentre, blinded, placebo-controlled. Seven hundred ICU patients with fever and known or suspected infection were randomised to intravenous paracetamol every six hours versus placebo. Primary outcome, ICU-free days to day 28: 23 versus 22 days — a null result, absolute difference zero days (p=0.07 is a null, not a near-miss to be spun). Mortality: no difference at 28 days, and at 90 days 15.9% with paracetamol versus 16.6% with placebo (p=0.84) (Young et al., New England Journal of Medicine, 2015). This is the biggest and best-designed trial in the field, and it must carry at least as much weight as Schulman's. Confounds cut the other way here, and honesty requires saying so: these patients were on antibiotics and full organ support — the setting where fever's contribution to defence is most plausibly redundant — and the temperature difference between arms was modest (about a quarter of a degree), so the trial tested moderate suppression, not fever abolition. A null in a supported ICU patient does not prove suppression is free in an untreated infection at home; that extrapolation runs in the opposite direction from the usual one and gets labelled just the same. Cui bono: paracetamol manufacturers benefit from a "safe, no harm" headline; the natural-health constituency has an incentive to bury this trial because it is the strongest single piece of evidence against fever purism — and Realised's commercial neighbourhood is the second one, which is exactly why this trial appears here at full prominence.

6. Routine pre-emptive paracetamol at infant vaccination blunted antibody responses — Moderate; evidence type: human randomised trials (two, open-label, 459 infants). Infants given prophylactic paracetamol in the 24 hours after vaccine doses — routinely, before any symptom — had fewer febrile reactions and significantly lower antibody concentrations across ten pneumococcal serotypes and several other antigens; the authors concluded routine pre-medication at vaccination should not be recommended (Prymula et al., The Lancet, 2009). Qualifiers the headline version always drops: the trials were open-label; antibody titres, though lower, generally stayed above protective thresholds, so a real-world disease-protection difference was not demonstrated; and this tested prophylactic dosing — a different act from comforting a distressed febrile infant afterwards, which this finding says nothing against. It is also not, in any reading, a reason to alter or delay vaccination — the paper's own conclusion supports vaccination and questions only the routine pre-dose. Cui bono: anti-vaccine content weaponises this trial as "vaccines plus medicine harm your baby," which inverts its meaning; on the other side, "give them Calpol before the jab" was widespread standard advice with a non-trivial paediatric paracetamol market behind it, and this trial overturned it.

7. Aspirin and paracetamol during an experimental cold suppressed antibody responses and worsened nasal symptoms — in a tiny, unreplicated study — Emerging; evidence type: human randomised trial, very small (60 volunteers across four arms), experimental challenge model. Volunteers infected with rhinovirus and randomised to aspirin, paracetamol, ibuprofen or placebo: aspirin and paracetamol were associated with suppressed serum neutralising-antibody responses and increased nasal symptoms versus placebo (Graham et al., Journal of Infectious Diseases, 1990). The claim this study is famous for — prolonged viral shedding — was a non-significant trend, routinely miscited as a finding. Roughly fifteen people per arm, many outcomes tested, no convincing replication in thirty-five years. Cui bono: this is the most-quoted study in "cold medicine prolongs your cold" content because the true result is modest and the miscitation is dramatic; the multi-symptom cold-and-flu category depends on the opposite belief, that suppression is purely beneficial.

8. Population modelling suggests widespread fever suppression may modestly increase influenza transmission — Emerging; evidence type: mathematical modelling built substantially on animal-derived parameters — not measured human epidemiology. Earn, Andrews and Bolker combined animal-derived estimates of how fever suppression increases viral shedding with survey data on antipyretic use, and estimated that current use increases seasonal influenza cases by roughly 5% (95% confidence interval 0.2–12.1%) and pandemic influenza cases by roughly 1% (95% confidence interval 0.0–2.7%) — intervals that reach close to zero at one end and, for seasonal influenza, above 12% at the other (Earn et al., Proceedings of the Royal Society B, 2014). Secondary sources routinely swap the seasonal and pandemic figures; the attribution above is taken from the published paper. This is a reason to consider a population-level externality of "medicate and mingle"; it is not measured epidemiology. Cui bono: the "non-drowsy, keeps you going" product category is built on exactly the behaviour this model penalises.

Sub-area 4: The febrile-seizure question

9. Antipyretics do not prevent febrile seizures — removing the main justification parents are given for aggressive fever treatment — Strong Evidence; evidence type: Cochrane systematic review of human randomised trials. The review found antipyretic agents ineffective for preventing febrile-seizure recurrence, and ineffective even at lowering temperature during the episode that led to a recurrence; the drugs that did reduce recurrence (intermittent diazepam, continuous phenobarbital) carried adverse effects in up to 30% of children, weighed against a condition that is generally benign (Offringa et al., Cochrane Database of Systematic Reviews, 2021). What this does not mean: it is not "febrile seizures don't matter" and not "don't treat your child." A first febrile seizure needs medical assessment; any seizure over five minutes is an emergency; and antipyretics still do what they genuinely do, which is relieve discomfort. Cui bono: no commercial party gains from this finding, which is part of why it remains so poorly known against decades of thermometer-chasing advice.

Sub-area 5: Sickness behaviour and the case for rest

10. Lethargy, appetite loss and withdrawal during infection are an organised, cytokine-driven programme — not weakness — Moderate; evidence type: evolutionary reasoning plus rodent experiments; no human outcome trial. Hart's foundational analysis argued the behaviour of a sick animal is an adaptive strategy supporting the febrile response and reallocating energy to defence (Hart, Neuroscience and Biobehavioral Reviews, 1988); Dantzer and colleagues supplied the mechanism — proinflammatory cytokines acting on the brain produce a coordinated motivational state, reproducible experimentally by cytokine or endotoxin administration and attenuated by blocking cytokine signalling (Dantzer et al., Nature Reviews Neuroscience, 2008). Stated plainly: there is no randomised human trial of rest-versus-activity during infection measuring recovery — that absence is an absence, and the strength of "rest is the intervention" rests on mechanism plus the sleep evidence below, not on a trial of resting. Also honestly: "adaptive" does not mean "always beneficial" — the same cytokine programme, chronically activated, is the substrate of inflammation-associated depression. Cui bono: nobody profits from telling you to lie down, which is why the message is undersupplied; the entire presenteeism economy and the "power through" remedy category are built on overriding exactly this signal — though for many people working while sick is an economic constraint, not a choice, and deserves no sneer.

11. Short sleep before exposure roughly quadruples the odds of catching a cold — the supporting receipt for rest — Strong Evidence; evidence type: human experimental viral challenge with measured (not randomised) sleep. One hundred and sixty-four healthy adults had sleep measured by wrist actigraphy for a week, were then quarantined and experimentally given rhinovirus: those sleeping six hours or less had roughly four-fold higher odds of developing a clinical cold than those sleeping over seven (odds ratios 4.5 below five hours, 4.2 at five-to-six; Prather et al., Sleep, 2015). The experimental infection under quarantine removes the differential-exposure confound that wrecks observational cold studies — but sleep itself was measured, not randomised, so residual confounding by stress, health and circumstances remains. And note the direction: this is susceptibility before exposure, adjacent support for rest during illness rather than direct proof of it. Cui bono: the sleep-product and wearable industries quote this constantly — and Realised has a direct commercial interest in sleep mattering, which is stated here rather than hidden, and is why the study's limits are reported at full strength.

Sub-area 6: Red flags and the boundary of the thesis

12. A defined set of fever situations demands urgent care, and clinical guidance explicitly endorses treating for comfort — Foundational; evidence type: clinical guideline (NICE) plus case-control and historic cohort data (Reye's syndrome). The full red-flag list lives in Risks and Contraindications below; the guideline points worth anchoring here: any fever of 38°C or above in an infant under three months is a red-flag feature requiring urgent assessment under the UK's NICE fever guidance for under-fives, and the same guidance advises antipyretics for a feverish child who appears distressed — not with the aim of reducing the number — and advises against giving paracetamol and ibuprofen simultaneously. That is a national clinical guideline sanctioning comfort as the treatment goal, which this entry adopts with authority rather than apology. Aspirin stays off-limits for children and teenagers with viral illness: the Reye's syndrome association, though formally an association from case-control and historic cohort data rather than trial-proven causation, underpins a public warning that took US cases from hundreds per year in the early 1980s to under twenty annually, and the precautionary rule is universal and non-negotiable. Cui bono: none — no one profits from clear red flags, and the risk of burying them runs entirely one way.

13. What this entry's thesis cannot claim — framing statement; absence stated as absence. Influenza kills hundreds of thousands of people globally each year. There is no evidence that catching influenza — or any infection — confers net benefit, no "immune-training" payoff worth the risk, and nothing in the fever literature supports seeking or welcoming infection. The defensible claim is narrow: the body's response to infection is a functional evolved programme rather than a malfunction, and reflexively erasing every part of that response has, in some contexts, a measurable or plausible cost. That is a claim about the response, never about the infection. Nothing here touches vaccination guidance: nothing in this entry is a reason to decline or delay a flu vaccine, and the infant-vaccination finding above concerns routine pre-medication at the appointment, not the vaccine.

Mechanism

Fever is not overheating; it is re-targeting. Molecules from pathogens, and the body's own signalling cytokines released in response (interleukin-1β, interleukin-6, tumour necrosis factor), drive an enzyme cascade — cyclooxygenase-2 producing prostaglandin E2 — in the preoptic area of the hypothalamus, the body's thermostat. Prostaglandin E2 raises the thermostat's set-point. The body then uses its ordinary machinery to reach the new target: vasoconstriction to conserve heat, shivering to generate it. This is why the rising phase of a fever feels cold — you are below your new set-point — and why the breaking phase brings sweating, as the set-point falls back and the body sheds the excess. Antipyretics like paracetamol and ibuprofen work by inhibiting the prostaglandin step and lowering the set-point; they persuade the thermostat, they do not cool the room.

Why raise the set-point at all? The best-supported answers, from animal and laboratory work: febrile temperatures accelerate the delivery of immune cells to lymph nodes where the response is being organised, induce heat-shock proteins that support antigen handling, sharpen neutrophil and natural-killer function, and sit above the replication optimum of some pathogens. Fever is expensive — roughly a 10–13% metabolic surcharge per degree — and expensive traits conserved across vertebrate evolution usually earn their keep. But conservation is evidence of past selective value; a modern human with antibiotics and fluids is not the environment the trait was selected in, and the honest reading of the human trials is that within a moderate range, in treated patients, blunting fever costs little or nothing measurable.

Sickness behaviour runs on the same signals aimed at a different organ. The same proinflammatory cytokines act on the brain and produce a coordinated motivational shift: fatigue, sleepiness, loss of appetite, loss of interest in socialising. Experimentally, administering cytokines or bacterial endotoxin reproduces the whole syndrome; blocking cytokine signalling attenuates it. The functional account — energy reallocated from foraging and activity toward immunity and the metabolic cost of fever — is plausible and widely accepted, though not proven in humans by outcome trials. The transmission-limiting account (withdrawal protects the group) is a hypothesis and labelled as such.

The one mechanism distinction that can save a life: hyperthermia is not fever. In heatstroke, stimulant or MDMA toxicity, serotonin syndrome and malignant hyperthermia, the set-point is normal and heat load has overwhelmed the body's ability to dissipate it. Antipyretics act on the set-point, so they do nothing in hyperthermia. The treatment is active cooling, urgently. A very hot person who has been exerting in heat, or who has taken stimulants, is a different emergency from a febrile person with an infection.

Risks And Contraindications

Seek urgent medical care — these override everything else in this entry:
• Any fever of 38°C or higher in a baby under 3 months. This is a red-flag feature in national paediatric guidance. Not a wait-and-see situation.
• Fever with a non-blanching rash (spots that don't fade under a pressed glass), stiff neck, severe headache with light sensitivity, a bulging fontanelle in an infant, unusual drowsiness or confusion — the meningitis/sepsis picture.
• Fever during chemotherapy or with a known weakened immune system. Febrile neutropenia is a same-day medical emergency, not a next-day one. Also urgent: fever with recent surgery or an indwelling line or device, and fever in pregnancy.
• Temperature at or above roughly 40–40.5°C (a conventional clinical threshold rather than a trial-derived one, and stated as such).
• A seizure with fever: a first febrile seizure needs medical assessment to exclude serious causes; any seizure lasting more than five minutes is an emergency. Febrile seizures are terrifying to watch and usually benign — both halves are true.
• Fever persisting beyond about three days, or returning after apparent recovery (the classic pattern of a secondary bacterial infection).
• Signs of dehydration — markedly reduced urine, dry mouth, no tears in a crying child — or breathlessness, chest pain, or confusion.
• A very hot person after heat exposure, exertion, or stimulant drugs: treat as possible heatstroke — active cooling and emergency care. Antipyretics do not work here.

Medication rules that are not optional: no aspirin for children or teenagers with viral illness (Reye's syndrome). Respect paracetamol's maximum daily dose and watch for it hiding in combination remedies — staggered accidental overdose is a leading cause of acute liver failure. Ibuprofen and other NSAIDs carry their own cautions (stomach, kidneys, some heart conditions, later pregnancy) — check with a pharmacist if in doubt.

And the standing rule for every entry in this family: never stop or avoid a medicine you have been prescribed on the strength of anything written here. If a clinician managing you says to treat the fever, treat the fever — they have information about you that no general-evidence entry does.

Controversy

Nature of the controversy: a genuine evidential gap being colonised from both sides by confident overstatement. The mechanism strongly suggests fever is functional; the human trials fail to show that moderate suppression costs anything measurable; and both camps resolve the tension by quoting only their half.

Position A — "fever is defence; suppression is sabotage." Its best evidence: the ectotherm survival experiments (blocking behavioural fever killed infected lizards, both physically and pharmacologically); the deep evolutionary conservation of an expensive trait; the demonstrated immune enhancements at febrile temperatures; the infant-vaccination antibody finding; and the small human signals (Schulman, Graham). Where it overreaches: it inflates a non-significant 7-versus-1 mortality imbalance in 82 patients into "treating fever kills"; it cites a non-significant shedding trend as proof colds are prolonged; it transfers lizard energetics to humans without pricing the 10–13%-per-degree metabolic surcharge; it ignores the largest and best trial in the field because the result is inconvenient; and at its worst it shades into advice that delays care for the infants, chemo patients and meningitis cases for whom fever is a red flag.

Position B — "fever is a symptom; normalise the number." Its best evidence: the HEAT trial's clean null — 700 blinded patients, no harm from paracetamol on any endpoint — plus decades of safety data for antipyretics used at proper doses, and the plain fact that relief of misery is a clinical good. Where it overreaches: it treats the thermometer as the patient — medicating a comfortable child to fix a chart number; it recruits febrile-seizure prevention as justification when the systematic-review evidence says antipyretics don't prevent them; it extrapolates an ICU null (antibiotics on board, modest suppression) to "suppression is free everywhere, always"; and its commercial wing sells the override of sickness behaviour as a virtue — "don't let a cold slow you down" — which prices neither the individual trade nor the transmission externality.

The funding/bias dimension — cui bono, both ways: the over-the-counter antipyretic and combination-remedy market is worth billions and benefits from the default that every symptom is a defect needing a product; the presenteeism economy benefits from workers who medicate and attend. On the other side, "your body's wisdom" is also a commercial position — it sells books, supplements, and anti-pharma content, and it gains from over-reading animal work into clinical advice. Note also the null-shaped hole: nobody has a strong incentive to fund large trials of not treating fever in ordinary community illness, so the thin evidence base is partly a funding artefact, in both directions.

Realised Position: Fever and sickness behaviour are regulated defences, not malfunctions — and that reframe changes how you relate to being ill more than it changes what you swallow. You do not need to fight a moderate fever to zero, and you do not need to fear it; the number is not the patient. Treat for comfort, freely and without guilt — the best human evidence says moderate suppression carries no demonstrated recovery cost, and national guidance explicitly endorses comfort as the goal. Respect the pull to rest as information, not weakness. And hold the red flags as hard overrides: young infants, the immunocompromised, the meningitis picture, very high temperatures, long fevers, seizures — these are medical situations, not philosophy. Realised's brand sits nearer the "body's wisdom" camp, so the discipline this entry imposes on itself is giving the strongest contrary trial full weight and keeping the thesis conditional.

Cross-Pillar Connections

• defensive_symptoms_and_when_to_blunt_them — the general framework this entry instantiates: fever and sickness behaviour are the flagship cases of symptoms that are defences, and the when-to-blunt logic (costless comfort versus real trade versus mandatory treatment) is developed there across cough, nausea, diarrhoea and pain.
• common_cold_prevention_and_treatment — what actually shortens and prevents colds; this entry supplies the fever-and-rest half of managing one, that entry covers the remedies, zinc-and-vitamin claims, and prevention evidence.
• illness_inflammation_and_sleep_disruption — the same cytokines that raise the set-point and drive sickness behaviour also reshape sleep architecture during infection; that entry covers sleeping while ill, this one covers why the sleepiness is there.
• immune_function_cross_pillar_optimisation — the day-to-day baseline (sleep, stress, activity, diet) that determines how often this entry's acute scenario happens; the sleep-susceptibility finding here is the bridge between the two.
• hormesis_and_adaptive_stress — the boundary case: fever is a body-initiated defensive stress, not a chosen hormetic one, and that entry's dose-response framing explains why "stress can be adaptive" never licenses seeking infection.

What would change our mind

Falsifiability: explicit upgrade/downgrade criteria from source

• Toward firmer "suppression has a cost": a replication of the rhinovirus-challenge antibody findings in a modern, adequately powered trial; a large randomised trial in community (non-ICU, non-antibiotic-supported) infection showing slower recovery or higher complication rates with routine antipyresis; direct human measurement (not modelling) that antipyretic use increases onward transmission; or demonstration that the infant-vaccination antibody differences translate into measured differences in disease protection.
• Toward firmer "suppression is simply free": a HEAT-style large blinded trial in community illness showing null results outside the ICU safety net; failed replications of the Graham and Schulman signals under better conditions; or trial evidence that aggressive antipyresis in children improves any hard outcome.
• On rest: any randomised human evidence on activity versus rest during acute infection would upgrade or downgrade the currently mechanism-based rest recommendation — in either direction.
• What would NOT move us: more mouse or in-vitro mechanism papers (the mechanism tier is already granted — the gap is human outcomes); more modelling built on the same animal-derived parameters; dramatic retellings of Schulman's raw death counts without its statistics; testimonials of fevers "burned through" or colds "caught early" with medication; and any argument, however eloquent, that infection itself is beneficial — that claim needs outcome data nobody has produced.

Industry bias note

Structural incentives the evidence base may reflect

The suppression end. The over-the-counter analgesic and cold-and-flu category is a multi-billion market whose core framing — fever and symptoms as defects to erase, "don't let a cold stop you" — precedes and outruns its evidence. Its commercial interest is served by the number-on-the-thermometer model of illness, by combination products that stack a paracetamol dose inside every remedy, and by advertising that presents overriding sickness behaviour as a virtue. The historic default of pre-dosing infants with paracetamol at vaccinations, and ICU protocols that normalised every chart, show how far convenience-shaped practice travelled ahead of trials. Employers and presenteeism culture push the same direction without selling a single tablet.

The naturalist end. "Never suppress your body's wisdom" is also a product. Evolutionary-medicine framing, anti-pharma channels and natural-health brands monetise the fever-is-sacred story, and their systematic tells are visible in this literature: the lizard experiments retold as human clinical evidence, Schulman's non-significant deaths retold as proof, Graham's non-significant shedding trend retold as "medicine prolongs colds," HEAT omitted, and the fever/heatstroke distinction dropped because it complicates the narrative. The worst versions bleed into anti-vaccine content, misusing the infant-vaccination trial against its own authors' conclusion. Realised operates in this end's commercial neighbourhood — the incentive to overclaim here is ours, and naming it is the cheapest honesty signal available.

The clean signal. The findings with no constituency behind them are the ones to trust most: the Cochrane conclusion that antipyretics don't prevent febrile seizures (profits no one, embarrasses standard advice); the HEAT null (disappoints both camps — no harm for the purists, no benefit for the medicators); the NICE instruction to treat the distressed child rather than the thermometer; and the fever-versus-hyperthermia distinction, which sells nothing and saves lives. An entry built on those four planks, with the contested material tier-labelled around them, is the honest version of this topic.

Fever and Sickness Behaviour: A Regulated Defence You Treat for Comfort, Not for the Number