THE EVIDENCE

Confounding

Studies in adults have found associations between reported asthma or symptoms (rather than doctor diagnosed asthma) and BMI and, in a few studies, height and weight were also self-reported. This has led to some scepticism that the association is spurious or due to confounding¹³ or, at most, the result of increased perception of symptoms among those who are overweight.¹⁴ Schachter et al found an association between symptoms and medication for asthma and increasing BMI, but not airway responsiveness (AHR), in 1971 adults aged 17–73 years,¹⁴ and in this issue of Thorax they present similar findings in 5993 children aged 7–12 years.¹⁵

The association is not simply due to concomitant trends in asthma and obesity, as suggested by Wilson.¹³ The association is not ecological but is found in individual data and, while in the UK the trends in BMI and asthma have been concurrent in children, the trend in BMI does not explain the trend in asthma due to the recent nature of the association.¹⁰ Wilson’s alternative explanation was that it was due to confounding. Confounding can never be completely ruled out in observational studies, but the factors suggested by Wilson—gastro-oesophageal reflux and obstructive sleep apnoea—are not potential confounders but intervening variables on putative causal pathways.¹³

Increased perception

That the association may be due to increased perception of symptoms in obese individuals is much more difficult to rule out. Indeed, it can be assumed that part of the association is due to perception as lung function decreases with increasing BMI within individuals,^16,17 although in cross sectional data an increase in lung function may be seen at lower BMI and a decrease only at higher values.¹⁸ The question is whether the association is entirely due to increased perception. The conclusions of Schachter et al were based on finding no trend of increasing AHR with greater BMI.^14,15 However, in 11277 participants in the European Community Respiratory Health Survey (ECRHS) a statistically significant trend was found even after adjusting for lung function,¹⁹ and a case-control study of men in the Normative Aging Study found a U-shaped relation with greater AHR at high and low BMI, also adjusted for lung function.²⁰

There is other evidence for not dismissing the association as being entirely due to increased perception. In children, case-control studies comparing those with diagnosed asthma and those without showed that the asthmatic children had greater mean BMI.^21,22 Furthermore, at least six longitudinal studies have shown an increased incidence of asthma in overweight or obese children and adults.^6,7 A delayed effect is more difficult to explain away than an immediate one by increased perception. In these studies incidence was calculated in those disease or symptom free at baseline. The lack of an agreed definition of asthma, and the difficulty of differentiating true incident asthma from recurrence of quiescent asthma, have provoked criticism of this approach.²³ However, the studies do make the reverse causation hypothesis—that lack of exercise in asthmatic patients promotes obesity—an unlikely explanation. These studies also provide evidence against the mechanical effects of obesity being the sole explanation, by the same argument as against increased perception and against a combination of perception and mechanical effects alone.

Studies of change in symptoms in obese asthmatic patients who lose weight have the potential to overcome the above scepticism. The one randomised controlled trial of 38 obese patients showed a reduction in symptoms and improvement in health status in the treated group compared with the control group, and an increase in lung function.²⁴ However, at high BMI a reduction in weight is likely to increase lung function irrespective of symptoms,^16,17 so this may not have convinced the sceptics. Airway responsiveness was not measured. A large trial including AHR as an outcome could provide evidence that the change in reported symptoms is not entirely due to reduced perception with weight loss, although on its own it cannot determine whether obesity is a cause of asthma.

Dietary factors

Review articles have considered many possible explanations apart from confounding, mechanical effects, and perception.^5,6 Obesity may modify the immune system, female sex hormones may play a role, physical inactivity may independently promote obesity and asthma, and a large number of dietary factors may be implicated.⁵ In the randomised controlled trial weight reduction was achieved through a diet which was modified in content as well as in calories,²⁴ and surgical reduction may lead to dietary changes. It may be feasible to collect dietary data in asthmatic patients motivated to lose weight, but a large study will be required to disentangle the candidate explanatory factors.

Sex differences

A number of studies have found an association between the prevalence or incidence of asthma in women but not in men, prompting a discussion of the mechanisms involving female sex hormones. However, the finding was not universal and, in the ECRHS, the association between AHR and obesity was, if anything, greater in men,¹⁸ but associations between symptoms and obesity were almost identical in men and women.²⁵ Part of the explanation for this heterogeneity in the findings may be in study size and methodology. In order to conclude a different effect in men and women, a statistically significant interaction is required. It is not sufficient to observe a statistically significant relation in one group and not in the other, but studies may lack the power to detect an interaction.²⁶ A number of studies have not reported a test of interaction but analysed data for men and women separately on a priori grounds because of the previous reports of differing associations. Others have reported a combined effect, so the evaluation of the evidence for and against a sex difference is quite difficult. In addition, some of the larger studies of adults showing a greater association in women analysed reported height and weight^27,28 which may have different validity in men and women. Only a large study with the power to detect an interaction effect can answer the question, but the interaction may genuinely differ between studies if the association is due to multiple mechanisms with difference contributions in different places.

Schachter et al found an association between BMI and atopy in girls,¹⁵ while Jarvis et al found no association with atopy in men or women, and no interaction between BMI and atopy on symptoms.²⁵ Huang et al found an association between high BMI and both atopy and AHR in girls but not boys in Taiwan,²⁹ the association with atopy explaining that with AHR. An association between BMI and atopy was reported in a study in Finland but no symptom data were included.³⁰

WHERE ARE WE NOW?

The scientific community is divided over the importance of the reported association between obesity and asthma, over whether the association is confined to women and girls or not, and whether atopy is also associated and perhaps on a causal pathway. In addition, there are a number of plausible mechanisms with little or no evidence for or against their role. Only large studies which include AHR as an outcome are likely to add further to the debate. However, we can surely all endorse the plea made by Redd and Mokdad²³ not to delay intervention programmes to tackle the obesity epidemic while we argue over the mechanisms for an association with asthma.