Disease Trajectories in Childhood Atopic Dermatitis

An Update and Practitioner's Guide

A.D. Irvine; P. Mina-Osorio


The British Journal of Dermatology. 2019;181(5):895-906. 

In This Article

The Challenge of Analysing Epidemiological Data

The clustering of atopic comorbidities has long been recognized.[46] Cross-sectional and longitudinal analyses of epidemiological studies have suggested a phenotype known as the 'atopic march', described as a natural progression of atopic manifestations from AD to asthma and allergic rhinitis (Figure 2).[47–50] This has created a common perception that AD is the 'entry point' for the development of other allergic diseases and that most paediatric patients with AD either follow the atopic march trajectory or outgrow the disease. However, the concept of the atopic march as the most common phenotype in paediatric patients with AD has been challenged by newer longitudinal studies applying more sophisticated statistical techniques.[51]

Figure 2.

The atopic march has been defined as the sequential progression from atopic dermatitis to asthma and allergic rhinitis. The common perception is that patients either follow this trajectory or outgrow the disease. This graph shows the percentage of patients with each of the indicated allergic conditions (y-axis) at different age groups (x-axis) in the subset of patients who follow what is traditionally known as the atopic march trajectory. This corresponds to the cross-sectional analysis of the data described by Belgrave et al.41

One of the most problematic aspects of analysing epidemiological data in AD is the difficulty of identifying incident phenotypes and establishing a causal relationship between AD and other concurrent atopic comorbidities. Many allergic diseases share common pathophysiology and genetic risk factors with AD, possibly explaining their co-occurrence.[52,53] The Mechanisms of the Development of Allergy (MeDALL) project has proposed the term 'multimorbidity' to describe the co-occurrence of allergic diseases in patients with AD, as the primary allergic disease is often not easily established. Accordingly, a variety of phenotypes and disease trajectories have been described, with the classic atopic march not applicable to every patient.[54]

Reporting bias can occur when clinicians are requested to score the presence or absence of concomitant comorbid conditions.[55] This is due to the fact that allergic comorbidities, such as allergic rhinitis and asthma, can appear at the point in the disease trajectory at which AD severity has already decreased significantly, meaning recall bias may lead to under-reporting of AD and other atopic comorbid conditions.[56] For example, in an Italian cohort (205 children, age 6–36 months, average 16·9 years of follow-up) diagnosed with AD by clinical examination (48·3% moderate, 19% severe), 60·5% of patients showed remission of AD by the end of follow-up.[57] However, when the investigators assessed the presence of other allergic conditions, such as asthma or rhinoconjunctivitis, only 20·5% of participants were reported as free of AD symptoms.[57]

Other issues encountered in the interpretation of epidemiological data include: (i) heterogeneity in data collection methods, (ii) use of different AD diagnostic criteria across cohorts, (iii) lack of consistency in the use of AD severity scales and (iv) absence of studies that stratify AD persistence by baseline AD severity.[58–63] All these factors make harmonizing datasets and defining and analysing disease severity and persistence across studies difficult.[58–63] Additionally, patients are frequently assigned to a phenotype based on the presence of individual symptoms listed in a questionnaire, when ideally syndromes of coexisting symptoms would be used, as they may represent patient subsets more accurately.[64]