For eczema in under-twos, ask the parent what a doctor said — not what the rash did

Design

post-hoc diagnostic accuracy analysis within a randomised prevention trial

Population

933 US children aged 2 years or under; chart-diagnosed atopic dermatitis in 26.2% at 24 months

Primary outcome

sensitivity, specificity and predictive values against clinician chart diagnosis

Effect

proxy report 71.6%/93.2% sensitivity/specificity at 24 months; symptom questionnaire 50.9%/93.2%

A post-hoc analysis of the CASCADE prevention trial compared two remote ways of identifying atopic dermatitis in children aged two and under against a reference standard of a clinician's chart diagnosis, in 933 participants. Chart-diagnosed atopic dermatitis was present in 26.2 per cent at 24 months.

Asking the caregiver whether a clinician had diagnosed atopic dermatitis — a proxy report — gave sensitivity 65.8 per cent and specificity 91.3 per cent at 12 months, and 71.6 per cent and 93.2 per cent at 24 months, with a positive predictive value of 88.3 per cent at 24 months. The Childhood Eczema Questionnaire, which asks about symptoms, performed markedly worse on sensitivity: 46.3 per cent at 12 months and 50.9 per cent at 24 months. Most of its false negatives were cases where the caregiver had not observed or reported itch or rash.

The finding is about case identification for research and remote triage, not about replacing examination. But it says something about history-taking: in this age group the parent's recall of what a clinician said is more reliable than their account of the symptoms, which is the opposite of the usual assumption.

• In a remote consultation for a young child, ask directly whether a doctor has ever diagnosed eczema
• Do not exclude atopic dermatitis because a parent denies itch — half of confirmed cases were missed that way
• Use symptom questionnaires to rule in, not to rule out, in this age group
• Remember the reference standard here was a trained primary care clinician, not a dermatologist

Why it matters

It inverts the usual assumption that parental symptom report is the more reliable history in infants.

Don't overread it

This validates remote case identification for research in children with healthcare access — it does not validate remote diagnosis in clinic.

The statistics, in plain English

A specificity above 90 per cent with sensitivity near 50 per cent means the questionnaire rarely labels a well child but misses half the affected ones. In a population where a quarter have the condition, that combination gives a usable positive predictive value and a negative answer you cannot trust.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.