The sixteenth birthday

The question

Where should a critically ill fifteen-year-old be looked after? Not in principle — in practice, tonight, in the hospital you are standing in. The United Kingdom answered that with a line: under sixteen to the paediatric intensive care unit, sixteen and over to the adult one. That line has been policy since a Department of Health document in 1997.

This paper is the first to go and look at what actually happens. Not what the policy says, but where 37,320 adolescents were actually admitted over eight years, what was wrong with them when they arrived, how long they stayed and how many died.

What they did

Two national audit databases, which between them capture nearly everything. The Case Mix Programme holds the adult units — 212 of them, about 94 per cent of general critical care units in England, Wales and Northern Ireland. The Paediatric Intensive Care Audit Network holds every paediatric unit in England, Wales, Scotland, Northern Ireland and the Republic of Ireland. Both were searched for every admission aged 12 to 19 between January 2007 and December 2014.

Numbers in each setting were remarkably similar overall — about 18,400 admissions to adult units and about 18,900 to paediatric ones. Strip out elective surgery, though, and the picture changes: 16,830 emergency admissions to adult units against 10,612 to paediatric ones. A great deal of what fills a children's unit is planned.

What happens on the birthday

The policy is followed most of the time. Some 83 per cent of twelve to fifteen year olds went to a paediatric unit, and 90 per cent of sixteen to nineteen year olds went to an adult one. But the interesting part is the shape of the transition, and it is not a slope.

Nothing happens to a person between the last day of being fifteen and the first day of being sixteen. The physiology does not move, the family does not change, the disease does not know. What changes is which building they are taken to, who is trained in what, whether the ward round uses their first name, whether a parent can stay overnight, and whether anybody on the unit has looked after somebody their age this month.

What comes in the door — and it is not the same illness

Excluding elective surgery, the two settings are seeing genuinely different adolescents.

• Adult units: trauma, then overdoseTrauma accounts for 25 per cent of adolescent admissions, and poisoning with drugs or alcohol for a further 13.5 per cent. Between them, nearly two in five.
• Paediatric units: respiratory, and hardly any toxicologyRespiratory disease accounts for 25 per cent of admissions, rising to 31.5 per cent in the over-sixteens who are still there. Drug and alcohol toxicity is uncommon, and adolescents over sixteen in a paediatric unit are rarely admitted after trauma.
• The exception proves the patternTwelve to fifteen year olds admitted to adult units are not a random sample: their commonest diagnoses are trauma and overdose, and their median stay is 0.9 days against about 2 days for everybody else.

The question the study could not answer

Which is better? The paper is admirably direct about this: it cannot say. Unadjusted mortality was 6.0 per cent in paediatric units and 5.7 per cent in adult ones, and those two numbers should not be compared, because the two databases do not use comparable mortality risk scoring. There is no correction for severity of acute illness and none for the burden of chronic illness — and we have just established that the two populations differ systematically in exactly those respects.

What evidence there is comes from elsewhere. Previous studies, restricted to adolescent trauma and mostly American, have consistently found no difference in mortality between adult and paediatric centres, but a higher intervention rate in adult ones. Make of that what you will; more procedures is not automatically worse care, and it is not automatically better either.

And now the Australian question

Read this as a British paper about a British rule and it is interesting. Read it from a regional Australian hospital and it becomes something else, because the study describes a choice that a great many of us do not have.

When a fourteen-year-old arrives critically ill at a hospital several hundred kilometres from a paediatric intensive care unit, nobody triages them by diagnosis into the most appropriate setting. They go to the intensive care unit that exists. The retrieval conversation happens around them, and depending on distance, weather and the state of the child, the answer is sometimes that they stay.

Which makes the paper's closing sentence land harder here than it does where it was written: services planned for the majority of adult and paediatric intensive care patients may not be optimal for critically ill adolescents. In a regional unit, that is not a policy observation. It is a description of Tuesday.

What to take to the bedside

• Adolescents are a small group and a real oneAbout 4,600 admissions a year across an entire country, which is a small share of intensive care and a large number of young people. Small enough to fall between services; large enough to plan for.
• Know which adolescent you are gettingIf you work in an adult unit, the adolescent arriving is statistically likely to be trauma or overdose and to be gone within a day. That is a specific set of skills, and one of them is not clinical.
• The short stay is the risk, not the reassuranceA median stay under a day in the twelve to fifteens in adult units means the mental health assessment, the child protection thinking and the follow-up all have to happen inside that day, in a unit not built to do them.
• Ask who is responsible for the non-clinical partConsent, confidentiality, education, the parent sleeping in the corridor. In a paediatric unit somebody owns these. In an adult unit it is worth finding out whether anybody does.

What to read sceptically

• The data are old nowCollected 2007 to 2014 and published in 2018. The paper itself notes that Major Trauma Networks were changing where injured children go, and that was eight years ago. The pattern is probably still broadly right; the precise proportions are not current.
• The denominators are not symmetricalThe paediatric database covers every unit in the British Isles; the adult one excludes Scotland, the Republic of Ireland and some non-participating units. The authors say plainly that the absolute numbers admitted to each setting should be interpreted with caution.
• Mortality before unit discharge is a narrow outcomeIt is what audit databases hold. It is not what an adolescent or their family would nominate as the thing that mattered about the admission.
• It is descriptive, and says soThis is a paper about what happens, not about what should. It was designed as the first step of a mixed-methods project. Treating it as evidence for a service model is asking it to do something it did not attempt.