Introduction
Asthma is a complex and heterogeneous airway disease, characterized by airway inflammation and bronchial hyperresponsiveness, resulting in recurrent episodes of dyspnea, wheezing and cough. Asthma is typically known as a disease that begins in early childhood and is associated with atopy and familial heredity. However, other forms of asthma have been recognized since the mid-twentieth century, and the clinical relevance of asthma starting in adulthood has since received more attention [
1,
2]. Adult-onset asthma is often a severe and uncontrolled form and poses a large burden on both patients and healthcare, due to frequent hospitalizations and high oral corticosteroid use [
1,
3]. Despite its burden, it remains largely unknown why asthma develops in adults [
4].
The incidence of adult-onset asthma over the past decades is stable, with an incidence rate ranging from 3.6 to 4.8 per 1000 person-years [
5‐
7]. While asthma in childhood is often associated with exposure to specific allergens such as house dust mite, animal dander and pollen, a different etiology has been proposed in adult-onset asthma including environmental and lifestyle factors. Several dietary hypotheses have been put forward, for example the shift to a more pro-inflammatory and energy-rich diet [
8‐
10]. This so-called ‘Western diet’ features high intake of animal products, saturated fat and refined sugars, while the nutritional value is relatively low in terms of antioxidants, omega-3 fatty acids and dietary fiber [
9]. In contrast, the more plant-based Mediterranean diet is believed to have anti-inflammatory properties, largely attributed to the amount of fruit, vegetables, grains and nuts that are consumed [
10]. These dietary factors can modulate epigenetics, gut microbiome composition, airway remodeling, innate immunity and inflammatory responses—all factors that have been implicated in the etiology of asthma [
11,
12].
Thus far, prospective studies focusing on dietary factors and incident adult-onset asthma report contradictory results [
13‐
29]. Yet, studies are very heterogeneous regarding the level of diet estimation—i.e., dietary patterns, diet scores, foods, or nutrients—and the definitions used for the assessment of asthma, making comparison of study findings a challenge [
30,
31].
Furthermore, the Western diet has also been pinpointed as one of the main causes of the global obesity epidemic [
31,
32]. Obesity itself is, however, considered an independent risk factor for adult-onset asthma [
30,
32] and a potential effect modifier in the relation between diet and asthma [
13]. This further complicates research into diet as risk factor of adult-onset asthma. To account for these complex interrelations, we aim to explore the relation of diet quality and food intake with incident adult-onset asthma at five year follow-up separately in normal weight and overweight adults from a general population.
Discussion
In a large population-based cohort, we explored the relation of diet quality and food intake with incident adult-onset asthma in normal weight and overweight adults. Our results show that diet quality—assessed by both the LLDS and MDS—was not associated with incident adult-onset asthma in both the BMI groups. Furthermore, no tendency was observed between the risk of incident asthma and the intake of a priori assumed favorable and unfavorable food groups. Although the dietary intake of several food groups differed between cases and controls, after adjustment for confounders only few food groups remained associated with incident adult-onset asthma. Overall, our results question the role of food as ‘simple’ risk factor of incident asthma in adults.
Several studies have been conducted on the association between dietary intake and incident adult-onset asthma, of which few assessed modification by BMI [
13‐
29]. Overall, these studies report heterogeneous results (see online supplement, table S3). Dietary pattern approaches show increased risk or no association between a western diet and adult-onset asthma [
13‐
17,
47]. ‘Fruit and vegetables’ or ‘nuts and wine’ dietary patterns were also not related to adult-onset asthma, as shown in a French cohort [
15]. This is supported by the data of Australian adults where indeed no association was found between the Mediterranean Diet Score and incident asthma [
18]. Yet, better diet quality—reflected by a higher Alternate Healthy Eating Index 2010 score—has been linked to lower risk in this Australian cohort, while no association was found in an US population [
18,
19]. Both studies did not assess subgroups of BMI. Our study largely confirms and extends their results by showing no association of diet quality and adherence to a Mediterranean-style diet with incident adult-onset asthma within specific BMI subgroups.
Regarding individual foods, we found a reduced risk of asthma with the intake of red and processed meat in normal weight adults. In other studies high processed meat intake was not associated with asthma incidence [
20‐
22], but was related to a higher asthma symptom score, with the effect being only partly mediated by BMI [
48,
49]. More in line to previous findings, we observed a moderate effect of increased intake of sugar-sweetened beverages and artificially sweetened beverages on higher risk of adult-onset asthma in normal weight subjects, albeit only borderline significant. Also non-obese US adults consuming sweetened beverages ≥ 2 times/day, were more likely to have asthma than non-consumers [
50]. In the Framingham Offspring Cohort, a strong effect of these beverages and fruit drinks—but not diet soda—on higher risk of incident asthma in adults was shown as well [
29], which was attributed to the high fructose:glucose ratio in these drinks. However, our results do not fully support this hypothesis as we also found an increased risk of asthma for higher intake of artificially sweetened beverages.
In the overweight group, we unexpectedly observed a decreased risk of asthma for the intake of savory and ready products and an increased risk for higher cheese and vegetable consumption. No longitudinal study has yet examined the intake of savory and ready products on adult-onset asthma, but one study in German adults prospectively assessed the risk of cheese intake on incident asthma, which observed no association [
22]. Furthermore, fruit and vegetables have been proposed to be protective for asthma development in two meta-analyses [
51,
52], but again few longitudinal studies have been performed in adults. These two Scandinavian studies show decreased risk of incident asthma for higher consumption of flavonoid-rich fruits and by a combined fruit and fish consumption score [
26,
28]. However, in line with our finding, in a large survey within the Global Allergy and Asthma European Network (GA
2LEN), the intake of fruity vegetables was also positively associated with a higher asthma symptom score in European adults [
53]. Interestingly, this association was most evident in the Dutch study center, which raises the question whether this can be attributed to the way vegetables are prepared in the Netherlands, as this may differ from habits in other countries. Indeed, 70% of Dutch people cook their vegetables by boiling, resulting in high loss of phytochemicals and consequently the nutritional quality of vegetables [
54]. The general conclusion from the authors of the GA
2LEN study was that there is no consistent evidence for an association of asthma with fruit and vegetable intake [
53].
Given the robustness of our results and the lack of (consistent) findings in previous studies, we might start to question the straightforward role of diet in the adult-onset of such a multifactorial disease as asthma. Focusing solely on dietary factors—be it dietary patterns, foods or nutrients—might be too simple to elucidate the complex relationship of a Western lifestyle with adult-onset asthma, and an integrated approach might be necessary that assesses lifestyle patterns. Besides diet, physical activity and body composition as ‘classic’ lifestyle factors [
30], this could also include smoking, drug use, socio-economic status and perhaps even stress or degree of urbanization. Although the current trend in studying diet-disease relationships is to assess dietary patterns and specific diet scores [
30], most observational studies— including ours by using the MDS and LLDS—have shown no effects on incident adult-onset asthma. By using this approach, important signals may be missed due to competing risks of food components. For example, foods that are rich in omega-3 fatty acids or antioxidants may also contain histamine or pesticide residues, which could potentially be harmful for some distinct asthma phenotypes. Indeed, asthma is also no longer considered a single disease, but rather an umbrella diagnosis for several diseases with various clinical appearances [
55]. These phenotypes vary according to the age of onset, atopic status, type of airway inflammation, and the association with obesity. Given this heterogeneity, better phenotyping is essential to understand the role of diet in the development and control of specific subtypes of asthma [
31].
The strengths of our study are the prospective design of the study and the large sample size, which allowed well-powered stratified analyses. Furthermore, highly valuable dietary data was available from the externally validated Flower-FFQ, which is also representative for the Dutch population [
37,
56]. In addition, the LLDS and food group categorization are in line with current scientific evidence on diet-disease relationships and reflect the adherence to the Dutch Dietary Guidelines [
42]. The LLDS is a relatively new score that corresponds to the well-established MDS as both scores include vegetables, legumes, fruit and nuts as positive foods. Although the LLDS has not been validated against other well-known dietary indices—such as the Alternate Healthy Eating Index—it has the ability to discriminate between groups based on socio-demographic characteristics and has been related to other disease outcomes, such as chronic kidney disease and Crohn’s disease in previous studies [
42,
57,
58]. Regarding the MDS, our Dutch study population may not have been representative enough for a traditional Mediterranean diet due to Westernized eating habits.
We next further confirmed the lack of association between diet and incident adult-onset asthma from a methodological point of view. First—as previously argued—combinations of foods and nutrients are consumed as meals rather than individual foods, which may interact with each other [
30,
31]. Therefore, a sensitivity analysis was performed with a model containing all 22 food groups. Similar effect estimates were observed, although significance could not be reached due to the high number of parameters in the model (see online supplement, table S2). This excludes a confounding effect of overall eating pattern.
Second, since the follow-up time of our study was limited, incident asthma cases may have already improved their dietary intake at the start of the study due to asthma-related complaints, potentially leading to reverse causation. This could also explain the lower energy intake in overweight asthma cases compared to overweight controls. Perhaps, asthma cases with overweight might have followed an energy-restricted diet more often to lose weight in order to improve their asthma symptoms. To address reverse causation, the dietary intake data of asthma cases who reported asthma symptoms at baseline (10%) were compared to those cases without symptoms at baseline. The comparison gave no indication to consider reverse causation as explanation for our study findings.
Third, the applied asthma definition was based on self-reported new-onset asthma rather than physician-diagnosed asthma. Although the applied asthma definition is a common way to select new asthma cases in large cohort studies, this might explain the lack findings in our study. Due to privacy regulations, it was not possible to validate disease status with medical records. However, 64% of incident asthma cases reported asthma medication use at follow-up, which is often only prescribed by a medical doctor. In addition, cases showed increased lung function decline from baseline to follow-up, as compared to controls, showing that cases indeed developed respiratory problems over time. Furthermore, strict criteria were applied to select a population-at-risk free of airway disease at baseline, but selection bias was judged unlikely after comparison of demographic characteristics of the selected population-at-risk and total adult Lifelines population (see online supplement, table S1).
Fourth, missing important confounders may influence the outcomes. Therefore, a sensitivity analysis was performed to adjust for some other well-known confounders, which did not affect our results (see online supplement, table S2). Despite, residual confounding remains possible by other unknown factors.
Finally, the effect of dietary intake on incident adult-onset asthma could be mediated by BMI, due to for instance the inflammatory effect of fat mass and adipokines on asthma onset [
30]. Although BMI is often considered a confounder in studies assessing diet-disease relationships, adjusting for a mediator might also lead to biased results [
30]. Since BMI was significantly different between cases and controls in the overweight group, a sensitivity analysis was performed with inclusion of BMI as covariate, which showed similar results (see online supplement, table S2). Hence, both confounding and mediation by BMI seemed unlikely, although the latter was not assessed formally.
In summary, the results of this study do not provide convincing evidence for a role of diet quality and food intake in the etiology of adult-onset asthma. However, this does not preclude diet as risk factor for asthma in other time windows across the life-span or as disease modifier in asthma patients. Yet, to elucidate the complex relationship between a Western lifestyle and the risk of adult-onset asthma an integrative approach might be necessary, including a range of modifiable lifestyle factors. This also includes appropriate asthma phenotyping to account for the heterogeneity of this common disease.