Modeling the Natural Course of Atopic Multimorbidity: Correlates of Early-Life States and Exposures

To the Editor:

A recent paper by Haider and colleagues (1) examined four UK population-based birth cohorts in the STELAR (Study Team for Early Life Asthma Research) consortium—Ashford, IOW (Isle of Wight), MAAS (Manchester Asthma and Allergy Study), and SEATON (Aberdeen cohort)—to delineate individual patterns of morbidity of eczema, wheeze, and rhinitis from birth to early adulthood. The authors provided a thorough analysis using different temporal frameworks and statistical methodologies to characterize the sequence of disease development. Based on their findings, the authors conclude that “atopic diseases fit a multimorbidity framework, with no evidence for sequential atopic march progression” (1).

Although a multimorbidity framework for atopic diseases is supported by the findings presented, the assertion that no evidence exists for sequential atopic march progression is not accurate. In fact, the risk of multimorbidity attributable to early-life eczema is higher than that attributable to wheeze or rhinitis, based on results presented in Table 3 (1). Moreover, the characterization of atopic march (life-course trajectory) absent the control for primary intervention, environmental, and treatment (medication) exposures could mask (confound) measures of effect (i.e., attributable risk). The argument that a low attributable risk of multimorbidity due to early eczema is associated with a low probability of transiting to multimorbidity is not consistent with several prospective studies that have shown that eczema is an important predictor of asthma risk (characterized by wheezing) (2). Furthermore, the analogy that the joint probability of wheeze (25%) and eczema (30%) at 0.075 would be expected by chance warrants a formal hypothesis test. Even so, the independent events assumption underlying such a joint probability is inconsistent with the multimorbidity framework, because evidence of event dependence and temporal sequence is well documented in previous prognostic studies (2). The use of a unidirectional latent Markov model to model transitions also ignores the likelihood of bidirectional transitions between multiple latent states (3).

The natural history of atopic diseases is heterogeneous, and a lot remains to be learned. Therefore, existing recommendations that suggest primary care physicians “should inform parents that children with eczema may later develop asthma” (1) should be moderated to communicate a range of risk considering other child- and parent-related risk factors (e.g., using a nomogram approach). The authors’ suggestion that physicians “should not make recommendations about ways to prevent atopic march or inform parents that children with eczema may later develop asthma” (1) goes further than is supported by their study findings. In fact, based on evidence of prognostic utility, most childhood asthma risk prediction tools incorporate early-life wheeze and eczema as critical predictors of school-age asthma risk (2). Although it is true that not all wheezers have (or develop) asthma, the exclusive modeling of wheeze symptoms often based on parent reports is prone to misclassification error or recall bias; this serves to undermine the extrapolation of “wheeze” symptom–based findings to asthma prevention recommendations. More cautious interpretations are warranted, especially because some evidence suggests that early-life preventive interventions, such as allergen avoidance, among children with early childhood eczema could reduce risk of current asthma during childhood (4, 5). This is consistent with a higher probability of transition to states with lower risk of multiple allergy sensitization and current asthma (3, 4). Furthermore, there is promising evidence for the potential impact of risk communication to assist shared decision making, individualize goal setting with at-risk populations, and reduce incidence and inequalities (6). The benefits of shared decision making regarding the need for preventive interventions (with a more holistic view of environmental influences) could outweigh the costs associated with communicating such risk. In addition, a growing body of research suggests that perturbations that occur in early childhood, a critical developmental period, may induce chronic conditions that extend into adulthood.