Do most children with eczema go on to get asthma and hay fever?
Only a small minority of children with allergy-type symptoms follow the classic path from eczema to wheeze to hay fever; most have just one or two of these conditions in their own distinct pattern.
Source
Developmental profiles of eczema, wheeze, and rhinitis: two population-based birth cohort studies
Study at a glance
- Design
- Cohort — Prospective population-based birth cohorts (ALSPAC and MAAS) with parent-reported symptoms at five ages from infancy to 11; Bayesian latent class machine learning grouped children by their symptom trajectories.
- N
- N=9801 · 9,801 children with data at two or more time points (8,665 from ALSPAC, 1,136 from MAAS); sensitisation trajectories used MAAS only, and the moderate/severe eczema sub-analysis used a reduced MAAS sample.
- Population
- Children in Avon (mothers recruited 1991-92) and in South Manchester/Cheshire (mothers recruited 1995-97), England, enrolled before birth.
- Outcome
- Latent classes of joint eczema, wheeze and rhinitis trajectories from age 1 to 11; allergic sensitisation by skin prick test across classes.
Structured fields used in claim comparison tables when every cited study has a complete layer.
What they did
The authors pooled data on 9,801 children from two English birth cohorts whose parents answered validated questionnaires about eczema, wheeze and rhinitis at five points between infancy and age 11. They fitted three Bayesian machine-learning models: one treating each symptom as independent, one hard-wiring the atopic-march sequence, and an unconstrained latent-class model that let the data define groups of children with similar symptom histories. Skin prick tests measured allergic sensitisation.
What they found
The best-fitting model found eight classes. About half of children (51.3%) had no disease, and only 3.1% followed an atopic-march-like profile, in which eczema also rarely resolved; the other classes included eczema only (15.3%), rhinitis only (9.6%), transient wheeze (7.7%) and persistent wheeze with later rhinitis (5.7%). The atopic march class had by far the highest risk of sensitisation (odds ratio about 22 versus no disease), and some classes with very different symptoms had similar sensitisation patterns. Results held in each cohort separately and in children with moderate/severe eczema.
The limits
What it doesn't show
Symptoms were reported by parents rather than diagnosed, and question wording differed slightly between cohorts, which could blur the classes. The models describe patterns of co-occurrence and cannot show what causes a child to follow one path or another, and class labels are the authors' interpretation. Sensitisation over time was available only for the smaller Manchester cohort, food allergy was not included, and both cohorts are English children born between 1991 and 1997, so the findings may not generalise to other populations.
Key terms
- Atopic march
- The idea that allergic diseases unfold in sequence within a child: eczema in infancy, then wheeze or asthma, then allergic rhinitis, with the eczema often fading.
- Latent class
- An unobserved group inferred from data; children in the same class share a similar pattern of symptoms over time.
- Birth cohort
- A group of people recruited around birth (here, during pregnancy) and followed forward with repeated measurements.
- Atopic sensitisation
- Having an IgE response to an allergen, measured here by a skin prick test wheal at least 3 mm larger than the control.
- Model evidence
- A Bayesian measure of how well a model explains the data while penalising unnecessary complexity, used to choose between models.
Flashcards
0 of 11 answers reviewed
Research intelligence for this paper
See its role on concept claims, tensions it is part of, placement history, and related discoveries.
Quiz yourself
Roughly what share of all children were in the atopic march class?
Common questions
If the atopic march is rare, why did it look real in earlier studies?
At the population level, eczema is most common in infancy and rhinitis becomes more common later, so cross-sectional prevalence looks like a sequence. Following individual children shows that most don't actually move through all three conditions.
Does this mean eczema never leads to asthma?
No. A small group did follow that pattern and were very likely to be sensitised. But most children with eczema, or with wheeze, did not go on to develop the other conditions, so using early eczema to pick children for asthma prevention may be poorly targeted.
What does 'machine learning' add here?
Instead of assuming a sequence, the unconstrained model let the data reveal groups of children with similar trajectories across all three symptoms at once, which conventional single-time-point analyses tend to over-simplify.
More on Paediatrics