Associations of Obesity, Sleep-disordered Breathing, and Wheezing in Children

Asthma and obesity are increasingly prevalent conditions in both adults and children (1-5). Several reports indicate an association between body mass and asthma and/or frequent wheeze (6-9). These associations have been postulated to be caused by several factors (10). Both conditions may have common risk factors, such as minority ethnicity and lower socioeconomic status (2, 3, 11, 12). In addition, genetic factors, common biomediators, and hormonal influences may contribute to both obesity and lung function impairment (10, 13-18). Causal pathways may also link these conditions. Although there has been speculation that asthma predisposes to obesity by reducing physical activity (19), this has not been shown to be the case. Rather, recent research has suggested that obesity precedes the development of asthma (20, 21). However, the mechanistic basis for this association is unclear.

Sleep-disordered breathing (SDB), a condition increasingly recognized in obese children, also may explain some of the association between asthma and obesity, as well as the potential heterogeneity in the literature in the reported association between these conditions. SDB is characterized by snoring, repetitive upper airway collapse, marked swings in intrathoracic pressure, and intermittent hypoxemia and upregulated inflammatory pathways (22). It also is more prevalent in minority children (18, 23, 24). Pathophysiologically, SDB could he in the causal pathway mediating obesity and asthma, with its effects possibly associated with SDB-mediated gastric reflux, enhanced vagal tone, or airway and systemic inflammation (10, 25). However. there has been a paucity of asthma research that has explicitly addressed SDB as a risk factor for lower respiratory symptoms, such as wheezing, and its potential role in the obesity-asthma relationship.

To address this knowledge gap, we analyzed data from a community-based cohort from the Cleveland area and assessed the relationships between obesity with two outcomes: wheeze and current asthma. We hypothesized that obesity is associated with asthma and asthma symptoms, and that these associations are partially explained by SDB.

METHODS

Study Sample

The Cleveland Children's Sleep and Health Study is an urban, community-based cohort of 907 children studied at ages X to 11 years. This cohort was originally assembled as a stratified random sample of full-term and preterm (< 37 weeks' gestational age) children horn between 1988 and 1993. The cohort was assembled to overrepresent African American and former preterm children to have internal validity and stable estimates of the relationship of these factors to SDB and other health outcomes as previously reported (24). Institutional review boards at participating hospitals approved the protocol. The child s legal guardian provided informed written consent and the child gave assent.

Measures

Children were studied in their homes or other convenient locations when they were 8 to 11 years old. Demographic and medical data were assessed with the Child Sleep Questionnaire (26). Direct measurements at the time of study included the following: spirometry, a home-sleep study, height, and weight. Spirometry was performed by trained research assistants. Predicted values of FEV^sub 1^ and FVC are based on published equations (27). Sleep apnea was assessed with in-home cardiorespiratory monitoring of thoracic and abdominal inductance plethysmography, pulse oximetry, heart rate, and body position (PT-2 system: SensorMedics, Yorba Linda, CA). Details on the methods for spirometry, the overnight in-home cardiorespiratory studies, and their validation compared with in-laboratory polysomnography, and scoring of respiratory events can he found on the online supplement and in previous reports (24, 28).

Definitions

Outcome variables. Children were classified as having "wheezing" if the parent reported that the child had wheezy symptoms (i.e., child's chest sounding wheezy or whistling occasionally apart from colds, most days or nights or during the night during the past year) or if asthma medication was used in the past 3 months (n = 188). Children who met the criteria for active wheeze were also classified as having "asthma" if the child was also reported to have asthma (n = 121). Children were classified as "no active wheeze/asthma" if asthma medication use, wheezy symptoms, and diagnosis of asthma were not reported by the parent (n = 600).

Predictors. Obstructive sleep apnea (OSA) was defined as an obstructive apnea-hypopnea index of five events or more per hour and/or an obstructive apnea index of one event or more per hour (24, 29). Children without OSA were categorized as habitual snorers if the caregiver reported that the child snored loudly in the last month, at least one to two times per week. Children who habitually snored and/or had OSA were categorized as "SDB," and children who neither snored nor had OSA were categorized as "no SDB." Body mass index (BMI) was calculated as weight (kg)/height (m^sup 2^), and obesity was defined as a BMI in the 95th percentile or higher for age and sex (30). Atopy was considered present if the parent reported that the child had hayfever, eczema, a positive skin test for allergies, or an allergy to mold, dust, cats, roaches, or pollen.

Statistical Analysis

Group differences were compared using two sample t tests, Wilcoxon rank-sum tests, and the Pearson ?^sup 2^ test for normally distributed, skewed, and categorical data, respectively. Logistic regression was used to assess the relationship between obesity and two outcomes, wheeze and asthma, controlling for traditional risk factors (age, sex, race, term status, and maternal history of asthma), as well as SDB. Main effects for covariates of secondary interest (caregiver education, caregiver marital status, smoker in the household) and all two-way interactions with SDB were evaluated using a stepwise method, and significant effects (p < 0.05) were retained. To determine the appropriate functional form of obesity, preliminary models, including traditional risk factors and various parameterizations of BMI percentile (including linear, quadratic, ordinal, and categorical terms) were fitted and compared using Akaike's information criteria (31). The results show that obesity defined as BMI in the 95th percentile or higher was the best fit to the data and the only obesity term that was a significant predictor of either outcome. Thus, obesity defined as BMI in the 95th percentile or higher was used in subsequent models. The final model was fitted with and without SDB, and SDB was considered to be a confounder of the obesity association if the estimated regression coefficient of obesity changed by at least 15%.

RESULTS

Of the 907 participants in this cohort, 835 had both a valid sleep study and height/weight measurements, enabling us to determine SDB and obesity status, respectively. We had sufficient questionnaire data to categorize 788 as wheezy/not wheezy and 721 as asthmatic/nonasthmatic. Descriptive statistics by asthma and wheeze status are shown in Table 1. These unadjusted results show that, compared with those without wheeze/asthma, a significantly greater proportion of children with wheeze were male, African American, were born prematurely, had a primary caregiver with a high school education or less, had a maternal history of asthma, were atopic, and had evidence of greater airflow obstruction on spirometry. Furthermore, compared with those with neither wheeze nor asthma, children with active wheeze had a significantly higher BMI and a greater prevalence of obesity (21.3 vs. 13.8%, p = 0.0141) and were more likely to have OSA (9.0 vs. 3.8%, p = 0.0045) and SDB (30.9 vs. 15.3%, p < 0.0001). There were no statistically significant group differences for underweight status.

Parallel analyses (Table 1) with asthma as the outcome variable yielded similar results. Compared with children without wheezing/asthma, children with asthma were significantly more likely to be obese (23.1 vs. 13.8%, p = 0.0097). Children with asthma had an almost twofold greater prevalence of both SDB and OSA; the latter association, however, was based on small numbers of children with asthma with OSA and did not reach statistical significance.

Logistic regression was used to model the association between obesity and wheeze (Table 2) and asthma (Table 3), while adjusting for potential confounders. A main effects model that included traditional risk factors for wheezing or asthma (age, race, sex, term status, and maternal history of asthma) in addition to obesity (BMI = 95th percentile) was first fit without considering the effect of SDB. The results show that, with the exception of age, all traditional risk factors were significantly associated with both outcomes, with odds ratios (ORs) ranging from 1.62 to 3.40. Moreover, obesity also was significantly associated with each outcome, with ORs of 1.57 and 1.82 for wheeze and asthma, respectively. When SDB was added to the model predicting wheeze, there was an 18.4% reduction in the regression coefficient for obesity and the effect of obesity was no longer statistically significant (OR 1.45; 95% confidence interval [CI] 0.93, 2.26). This finding suggests that the relationship between obesity and wheeze was partially explained by confounding with SDB. In contrast, the other traditional risk factors remained statistically significant predictors for active wheeze, even after considering both SDB and obesity in the model. Moreover, SDB is a significant predictor of wheeze, such that children with SDB have nearly twice the odds of wheeze compared with those without SDB (OR 1.89; 95% CI 1.26, 2.85). SDB was also significantly confounded with race; when SDB was added to the model, the regression coefficient for African American race was reduced by 15.6% (OR 1.72; 95% CI 1.21, 2.46).

When SDB was added to the model predicting asthma, there was only an 8.0% reduction in the regression coefficient for obesity, and obesity remained statistically associated with asthma (OR 1.73; 95% CI 1.03, 2.90). This modest reduction in the regression coefficient suggests that the relationship between obesity and asthma is not substantially confounded by SDB. After considering obesity and traditional risk factors, SDB was not significantly associated with asthma (OR 1.38; 95% CI 0.84, 2.28) and was not a significant confounder with the other traditional risk factors.

Alternative models that added atopy as a covariate to the final models were also fitted. As expected, atopy was significantly associated with both wheeze (OR 3.63; CI 2.52, 5.23) and asthma (OR 5.32; CI 3.41, 8.30). Similar trends were observed in the association of obesity and SDB with each outcome when atopy was included as a covariate. SDB remained a statistically significant predictor of wheeze, and the coefficient for obesity decreased by 12.5% and was no longer a statistically significant predictor of wheeze. When SDB was included in the model, obesity remained a statistically significant predictor of asthma, with the coefficient decreasing by only 2.7%.

Because our cohort was oversampled for preterm, we performed sensitivity analyses restricted to full-term children. Qualitatively similar findings were observed for the association of wheezing/asthma, SDB, and obesity in the restricted sample compared with the full sample. In addition, because it is possible that guardians may nonspecifically overreport "breathing" symptoms, such as wheezing and snoring, we repeated analyses defining our exposure as those with OSA who do not snore (n = 18) and excluded the 22 (55%) children with OSA with habitual snoring. In this subanalysis, children with OSA (without reported snoring) tended to have a higher prevalence of active wheeze compared with children without OSA (38.9 vs. 20.4%, p = 0.074), suggesting that nonspecific overreporting did not account for the observed associations between SDB and wheezing.

DISCUSSION

In this study, we confirmed that traditional risk factors, such as male sex, African American race, and prematurity, are positively associated with asthma and wheeze. We further assessed the nature of the association between obesity and wheeze/asthma and quantified the extent to which SDB contributed to this relationship. We observed a statistically significant increased prevalence of both wheeze and asthma in obese children (identified as those with a BMI percentile = 95%), with little evidence of linear or bimodal relationships of asthma/wheeze with changing BMI levels. Finally, our analyses suggested that SDB is associated with wheeze/asthma and that SDB may explain a part of the relationship between obesity and wheeze.

The basis for an increase of asthma in obesity is an area of active research. The rising prevalence of obesity in both children and adults highlights the public health importance of this area of research. There are several potential mechanisms that have been postulated for this association, including inflammation, mechanical factors, and common genetic factors. Leptin, a hormone produced by adipocytes with inflammatory and possibly immunoregulatory functions, is elevated in obesity and has been postulated to modulate the expression of the asthma phenotype (32-34). Obesity may induce mechanical airway changes, such as "latch state," which is related to stiffening of the smooth airway muscle from reduced tidal volume and functional residual capacity, increasing airway narrowing (15). Common genetic influences, including those that influence estrogen receptor expression and ß receptors, also may explain obesity-asthma associations (10). Nonetheless, the relative importance of these factors in the link between obesity and asthma is unclear.

Some studies have demonstrated a dose-response relationship between weight and asthma, whereas other studies have demonstrated a possible quadratic relationship, with the highest risk of asthma at both extremes of body weight (6, 35). Different dose-response relationships have been reported for boys as compared with girls (36). It is possible that some of the between-study differences in the reported dose-response relationship between obesity and asthma could be from the inclusion of different population subsets in various reports with other potential sources of heterogeneity, including ethnicity, preterm status, atopic background, or presence of SDB. In this sample, after considering potential confounders or effect modifiers, such as sex and race, we only observed an association between asthma/wheeze with obesity, as defined as a BMI in the 95th percentile or higher, with insufficient evidence of a linear association with BMI percentile and increased asthma/wheeze prevalence in children with BMI levels classified as "at risk" for obesity (BMI 85-95th percentile) compared with those not at risk (BMI < 85th percentile). We also did not observe evidence of increased asthma/wheeze in underweight children; however, the relatively small numbers of children with low BMI levels limited this analysis.

We based our measurements of SDB using both objective (i.e., an elevated apnea-hypopnea or apnea index) and subjective measurements (i.e., habitual snoring). Including snoring, a cardinal symptom of SDB, in our exposure definition may have reduced misclassification of children who may have had an unrepresentative single-night sleep study or who were in the lower spectrum of disease severity. However, the finding of similar behavioral morbidities in children classified on the basis of snoring compared with objective criteria (37-40) suggests similar or overlapping underlying physiology in both groups of children.

The role of SDB as a potential mediator between obesity and asthma has thus far received little attention. Most research in this area has addressed the association between SDB and lower respiratory symptoms without explicitly considering SDB as a confounder (or link) in obesity and asthma associations. Larsson and colleagues (41) have shown an increase in reported snoring and witnessed apneas in subjects with asthma compared with subjects without asthma. A British study showed that young wheezers and subjects with asthma have a higher prevalence of snoring compared with subjects without asthma (42). Teculescu and coworkers (43) suggested that snoring in French children was associated with exercise-induced asthma and a history of allergy. In a different population of family members of patients with OSA, we previously found that, after adjusting for obesity and race, persistent wheezing was a significant predictor of SDB (18). Although these reports underscore the potential importance of linked pathologies in the upper and lower airway, none explicitly addressed the interrelationships of obesity, asthma/ wheeze, and SDB.

SDB may modify the expression of asthma in the setting of obesity for several reasons. Gastroesophageal reflux occurs commonly in SDB (44) because of intrathoracic pressure swings that accompany intermittent upper airway obstruction, favoring acid reflux-induced bronchoconstriction (25). Abnormalities in common biomediators may explain some of this relationship. Augmented levels of leptin, elevated in obesity and considered to have immunomodulating functions, have been demonstrated in sleep apnea (45-48). SDB, obesity, and asthma also are each associated with augmented inflammatory states. In SDB, the oxidative stress during repetitive apneas may upregulate inflammatory cytokines and subsequent airway inflammation and smooth muscle constriction. A small, but intriguing, literature that has reported improved asthma control in patients with asthma with sleep apnea after treatment with nasal continuous positive airway pressure suggests that SDB may play a causal role in exacerbating asthma (49, 50).

SDB and asthma also may share common risk factors as proposed by the "unified airway" theory (51). Generalized abnormalities of the nasopharynx and lower airway may coexist either because of common airway responses to inflammatory or atopic stimuli, or because of other developmental factors that similarly affect the upper and lower airways. Although it is unclear how obesity might mediate common effects on both the upper and lower airways, it is possible that children with both upper and lower airway disease may be less physically active and thus be predisposed to obesity. In this study, the lack of objective measurements of physical activity prevented us from assessing this as a potential confounder.

The observed differences in the relative impact of SDB as a risk factor and confounder for our wheeze and asthma outcomes raise several interesting questions. It is possible that some of the small differences in parameter estimates are caused by chance, especially given the smaller sample of children classified as asthmatic compared with those classified as wheezing. However, if a failure to receive an asthma diagnosis despite wheezing or asthma medication use indicates a more atypical condition, then SDB may play a more prominent role in "atypical asthma" or "nonasthmatic wheezing." Alternatively, the absence of physician-diagnosed asthma in wheezing children may represent a diagnostic bias occurring in children of lower socioeconomic status, who may be at greater risk of SDB. Future studies that use refined assessments of asthma and atopy, relating these to measures of SDB and obesity, may ultimately produce improved phenotype descriptions that enable better descriptions of underlying pathophysiologic processes.

The strengths of this study include the collection of standardized, objective measures of SDB in a large sample of children representing population subgroups at increased risk for asthma. A study limitation is the lack of objective data on atopy, preventing us from fully describing the asthma phenotype and assessing differences in obesity and asthma between atopic and nonatopic subgroups. Another limitation relates to including parent-reported data on both wheezing and snoring in defining our outcome and predictors, respectively, with the potential for reporting biases. Exploratory analyses, however, showed an increased prevalence of wheezing among children with objectively measured OSA (without parent-reported snoring). These analyses suggested that the wheezing-snoring relationships were not solely an artifact of reporting bias.

In conclusion, we have confirmed that obesity, defined as BMI in the 95th percentile or higher for age and sex, is significantly associated with asthma and wheezing among a large, diverse pediatric population. Furthermore, our data indicate that unrecognized SDB may explain a portion of the association of obesity and childhood wheezing. Future epidemiologic studies of childhood asthma should minimally consider use of snoring data in assessments of obesity-asthma/wheeze relationships. Additional research is needed to address the extent to which SDB exacerbates the expression of asthma, and whether SDB treatment may lead to improvement of asthma in children with both conditions.

Conflict of Interest Statement: L.G.S. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; A.S.-I. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; C.L.R. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; H.L.K. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; S.R. received a grant from Sepracor with payment per participant enrolled (December 2002-June 2005) to evaluate levabuterol in a randomized controlled trial, and approximate payments for cost for participants' fees and study costs were $15,000.

Acknowledgment: The authors thank Sarah Bivins, Judith Emancipator, Najla Colebiewski, Heather Rosebrock, and Dina Tell for their expert assistance with data collection and processing. The authors also thank the participating families who made this study possible.

© 2005 American Thoracic Society Provided by ProQuest LLC. All Rights Reserved.