ICU and You · For the general public · Number 2
What doctors actually look at when they decide a child is sick — and why 37 degrees was never magic
Most of this site is written for doctors and nurses. This part is not. It is for anyone — parents, grandparents, people who look after other people's children, and anyone who has ever stood in a hallway at two in the morning holding a thermometer and wondering.
No medical knowledge is assumed. Nothing here can take the place of someone actually examining your child. But a lot of what makes a sick child frightening is that nobody has ever explained what the professionals are looking at — and that is a fixable problem.
Here is something that will annoy you. When a doctor walks into a room to see a feverish eight-year-old, the temperature is rarely the first thing they want to know, and it is almost never the thing that decides what happens next.
This is not because the number is meaningless. It is because the number answers a different question from the one you are actually asking. You want to know how sick is my child. The thermometer tells you is the thermostat turned up. Those are not the same question, and for a school-age child they are only loosely related.
So let us start with where the number came from, because it turns out to be a wonderful story about how a piece of nineteenth-century arithmetic became something the whole world believes.
In 1868 a German physician named Carl Reinhold August Wunderlich published a book on body temperature in disease. He had, by his own account, taken something in the order of a million readings from around twenty-five thousand patients. He did it with a thermometer about a foot long, tucked into the armpit, which took roughly twenty minutes to give a reading.
From all of that he concluded that normal human temperature was 37.0 °C, and that anything at or above 38.0 °C counted as a fever. Those two numbers have been on every thermometer, in every textbook and in every parent's head ever since.
They have also been drifting. When researchers at Stanford compared temperature records from Union Army veterans in the nineteenth century with modern American data, they found that average body temperature appears to have fallen by roughly half a degree over about 150 years. The best modern estimate of an adult's average is nearer 36.6 °C than 37.0.
Is that real, or is it just that old thermometers were badly made? Honestly — partly the second, and the scientists themselves say so; nobody knows how those nineteenth-century instruments were calibrated. But the decline also shows up within the old records, over decades in which the instruments did not change, which is harder to explain away. Take it as probably real and not fully settled.
Either way, the point for you is simpler. 37.0 was an average, not a boundary. It was never the line between a well child and an unwell one, and it was never meant to be.
A fever is not the illness. A fever is your child's immune system deliberately turning the thermostat up, because many bacteria and viruses do less well when it is warmer and parts of the immune response do better. It is a response, and a reasonably competent one.
Which leads to the two things parents are most surprised to be told, both of which are the settled position of paediatric guidelines in Australia and Britain.
One real exception, and it is a big one. Everything above is about older babies and children. In a baby under three months, a temperature of 38 °C is itself a reason to be seen straight away, whatever the baby looks like. Very young babies get seriously ill quietly, and the rules are genuinely different. If you have a newborn with a fever, stop reading and ring someone.
Roughly five things, in this order, and none of them requires equipment.
You will notice that a thermometer is not involved in any of them.
Is he playing?
Not playing well. Not playing happily. Just — at any point in the last few hours, did this child pick something up, argue with a sibling, ask for the iPad, complain about being bored, or take an interest in anything at all?
A child who is still, in some reduced way, behaving like a person with preferences is usually a child whose illness is being handled. A child who has stopped caring what happens to them is the one that worries us. This is not a scientific instrument and it will not appear in any guideline. It is nonetheless how a great many experienced clinicians make the first cut.
Most Australian parents know the glass test: press a clear glass against a rash, and if the rash does not fade, seek help urgently. That advice is correct, and you should keep doing it.
The problem is what people take from a negative result.
A study of 448 children and teenagers with meningococcal disease looked at when each symptom appeared. The famous ones — the rash that will not fade, the stiff neck, the drowsiness — turned up at a median of 13 to 22 hours into the illness. But around 72 per cent of the children had already shown early signs of blood poisoning at a median of eight hours:
Read that again, because it is the most useful thing in this piece. By the time the glass test works, you are typically half a day past the point at which the child was already showing signs. The rash is a late sign. It can also be scanty, or start off fading under the glass, or never appear at all.
So: a rash that does not fade means go now. A rash that does fade, or no rash at all, means nothing. It is a reason to act and never a reason to relax. The organisations that exist specifically for this illness say the same thing — get help if you are worried, rash or no rash.
This is the part I would most like people to take away, because parents routinely apologise for it and they should not.
A major review in The Lancet looked at which features best identify serious infection in children. Near the top of the list was not a vital sign or a blood test. It was a parent saying that this illness is different from their child's previous illnesses.
And in 2025 a Melbourne hospital published something better still. Over two years they asked caregivers one question before anyone took any measurements: are you worried your child is getting worse? Children whose caregiver said yes were admitted to intensive care 6.9 per cent of the time, against 1.8 per cent for those whose caregiver did not.
In that study, the caregiver's worry predicted admission to intensive care better than the child's heart rate and better than the child's breathing rate. A parent's answer to a single plain question outperformed two of the numbers we write down and put on a chart.
So when you say I know this sounds silly, but he's just not himself — that is not a silly thing you are saying to fill the silence. That is the highest-value sentence in the conversation, and a good clinician's ears prick up when they hear it. Say it early, say it plainly, and do not apologise for it.
A six-year-old can tell you what is wrong. A ten-year-old can tell you exactly what is wrong, in order, with adjectives. This is the great advantage of the age group and also its trap, because you start to rely on it.
Children this age are often being brave. They know that admitting to pain may mean a hospital, or a needle, or missing the thing they were looking forward to. They are also, quite often, managing you — a nine-year-old who can see that their mother is frightened will frequently decide to be fine.
I should be straight with you that this is clinical experience rather than proven fact. The research on children hiding symptoms is mostly about teenagers with long-term pain, not school-age children with sudden illness. But every paediatric nurse you ever meet will tell you the same thing, and the practical response costs nothing: believe the behaviour over the report. If a child says they are fine while lying completely still and refusing to get up, the stillness is the accurate statement.
Everything above is for the ordinary situation: a child who is unwell and you are weighing up what to do. Some things skip the weighing up entirely. This list is drawn from what Australian health services publish for parents, and it is worth knowing before you need it.
Any baby under three months with a temperature of 38 °C or more should be seen the same day, without waiting to see how things go.
Between those extremes, the honest advice is the one healthdirect gives and that no algorithm has ever improved on: you know your child. If something is different in a way you cannot explain, that is a reason to get them looked at, and a temperature of 37.4 is not an argument against you.
And if you are already in a hospital and feel you are not being heard, every state now has a formal route for families to escalate their own concern, independently of the treating team. In New South Wales it is called Raise It; in Queensland, Ryan's Rule; Victoria has a Kids Urgent Concern Helpline. They exist precisely because the evidence above is taken seriously. Ask the nurse in charge how to use the one in your state — that is what it is for, and using it is not rude.
Really — anything. There is no such thing as a question that is too basic.
Was there a time you took a child to be seen and felt foolish about it? Or a time you did not, and wished you had? Was there something a doctor or nurse said that you have never forgotten, for better or worse? I would like to hear about it, and those experiences shape what I write next.
I cannot give advice about a particular child or a particular illness, and I will say so if that is what is being asked. Everything else is fair game.
The button opens an email straight to me. If your device blocks it, write to icuandyou@icloud.com instead.