Customize your JAMA Network experience by selecting one or more topics from the list below.
Young SYN, Gunzenhauser JD, Malone KE, McTiernan A. Body Mass Index and Asthma in the Military Population of the Northwestern United States. Arch Intern Med. 2001;161(13):1605–1611. doi:10.1001/archinte.161.13.1605
Copyright 2001 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2001
Patients with asthma commonly have other medical problems such as obesity, but it is unclear if obesity independently relates to asthma occurrence.
To examine the association between asthma and obesity.
We studied enrollees aged 17 to 96 years in region 11 of TRICARE, a military managed health care program encompassing Washington, Oregon, and northern Idaho, using an enrollment questionnaire from January 1997 to December 1998. We performed case-control analyses on 2788 asthma cases and 39 637 controls. From these cases and controls, we selected a random sample of 1000 asthma cases and 1000 controls, linking them to a computerized military health record system to verify if medications indicated for asthma therapy were prescribed. After excluding cases not prescribed bronchodilator medications and excluding controls prescribed bronchodilator medications or steroids, we used logistic regression to estimate associations among asthma, body mass index, and demographic, lifestyle, and comorbid risk factors in 386 verified cases and 744 verified controls.
Increasing body mass index, younger age, female sex, non–active duty beneficiary status, and arthritis were significant independent predictors of asthma prevalence in both our larger analysis and our verified substudy, whereas stomach ulcer, depression, hypertension, and white race are also independent predictors of asthma prevalence in our larger analysis.
Increasing body mass index is a key factor predicting prevalence of asthma and, if determined to be etiologically related to asthma incidence, is a potentially modifiable risk factor for asthma.
IN 1990, health care expenditures for illness related to asthma were estimated at $6.2 billion or nearly 1% of all US health care costs.1,2 Asthma affects approximately 5.6% of the general population.3 Based on National Health Interview Survey results from 1980 through 1990, the age-adjusted prevalence rate for self-reported asthma increased 38%, from 3100 to 4290 per 100 000 population (from 6.8 million to 10.3 million persons affected).1 During the 1980s, asthma mortality increased by 6.2% per annum (±1.2%).4
In the US population, 54.9% of adults are overweight (body mass index [BMI] ≥25.0 kg/m2) and 22% are obese (BMI ≥30.0 kg/m2).5 The health care costs attributable to obesity were an estimated $99.2 billion in 1995 (5.7% of health care expenditure).6
Hypertension, obesity, arthritis, and diabetes are common in asthmatic individuals.7-11 Hypertension, arthritis, and diabetes are common in obese individuals.7,12-14 Therefore, obesity might be associated with the likelihood of having asthma.7,8,11,15-19 The few studies7,8,11,15-19 to date that have examined the relation between BMI and asthma have involved limited population groups such as children or patients at tertiary care institutions. Our objective was to examine the relation between BMI and asthma in a well-defined study population ascertained via a military health care system.
This study was a population-based prevalence case-control study of patients enrolled in TRICARE region 11, which encompasses Washington, Oregon, and northern Idaho. TRICARE is a military health care system comparable to a traditional health maintenance organization. The population includes active and retired service members of the US Air Force, US Army, US Coast Guard, US Navy, and US Marine Corps and their family members residing in region 11. The Health Enrollment Assessment Review (HEAR) is a self-administered questionnaire developed by the Office for Prevention and Health Services Assessment, the National Center for Environmental Health, and the Batelle Memorial Institute for TRICARE regions 6 and 4 through a Memorandum of Agreement among Armstrong Laboratory Human Services Command, US Air Force Materiel Command, and the Centers for Disease Control and Prevention. The questionnaires were designed to assist the Department of Defense and the managed care contractor in identifying patient health care resource needs as TRICARE beneficiaries. Questionnaires are mailed to patients enrolled in TRICARE after enrollment and then annually. Questionnaire data include demographic information, height, weight, lifestyle information, and history of medical problems. Based on institutional data, the questionnaire response rate was estimated to be 50%. The study was approved by the institutional review board of Madigan Army Medical Center.
There were 50 075 questionnaires collected from January 1997 to December 1998. If a person filled out the questionnaire more than once, data from the most recent questionnaire were used, for a yield of 45 743 questionnaires. Asthma cases were defined as patients responding positively to the question, "Have you ever been told by a health care provider that you have asthma?" All other individuals were defined as controls.
To eliminate the misclassification of asthma with emphysema, persons responding positively to the question, "Have you ever been told by a health care provider that you have emphysema/chronic bronchitis?" were excluded (n = 1973 positive, n = 1345 data missing). Of the remaining 42 425 persons, 2788 were defined as cases and 39 637 were defined as controls. Of the 2788 cases, we excluded 193 (6.9%) who had no information for BMI and 18 (0.6%) who had implausible data (such as BMI <7 kg/m2 or >60 kg/m2), leaving 2577 cases for our general analysis. Of the 39 637 controls, we excluded 3098 (7.8%) who had no information for BMI and 192 (0.4%) who had implausible data, leaving 36 347 controls for our general case-control analysis.
To analyze agreement of self-reported asthma diagnoses, a random sample of 1000 cases was selected from the 2788 asthma cases and linked to the Composite Health Care System (CHCS), a computerized military health record system that includes information such as medication profile history, to determine if medications indicated for asthma therapy were prescribed. Verified asthma cases were those cases who were prescribed medications commonly indicated for treatment of asthma, including inhaled bronchodilators (such as albuterol inhalers, metaproterenol sulfate inhalers), oral bronchodilators (terbutaline sulfate tablets, albuterol tablets, metaproterenol sulfate syrup), and orally inhaled steroids (triamcinolone inhalers). Since nasally inhaled beclomethasone dipropionate and inhaled triamcinolone aqueous nasal spray are indicated for allergic rhinitis, persons who were prescribed beclomethasone nasal inhalers alone or triamcinolone aqueous nasal spray alone were excluded. Since oral steroids are indicated for therapy of other conditions, such as autoimmune disorders, persons who were prescribed oral steroids alone were excluded. Persons prescribed a combination of beclomethasone nasal inhaler, triamcinolone aqueous nasal spray, and/or oral steroids were also excluded as cases. Since inhaled ipratropium bromide is indicated for the treatment of bronchospasm associated with chronic obstructive pulmonary disease, persons who were prescribed an ipratropium inhaler alone were also excluded.
In verifying the random sample of 1000 asthma cases linked to the CHCS, we excluded 248 (24.8%) not found in the CHCS, 46 (4.6%) listed as "no prescription information," 2 (0.2%) younger than 17 years, and 1 (0.1%) listed as having died, with no prescription information listed. Of the 703 remaining cases, the following were excluded: 213 cases not prescribed any bronchodilators or steroids; 73 prescribed intranasal beclomethasone, intranasal triamcinolone, or oral steroids alone; and 1 prescribed an ipratropium inhaler alone.
Most (99%) of the remaining 416 had been prescribed an inhaler of some sort, including bronchodilators such as albuterol, steroids such as triamcinolone, and other inhaled medications such as cromolyn sodium. Four patients (1%) had been prescribed oral bronchodilators alone, and 174 patients (41.8%) had been prescribed an oral steroid in addition to other asthma therapy. Of the 416 cases remaining, we excluded 28 (7.3%) who had no information for BMI and 2 (0.5%) who had implausible data. This left 386 verified cases.
From the initial 39 637 controls, we reviewed all medication prescription information on a random sample of 1000 controls linked to the CHCS to determine if medications indicated for asthma therapy were prescribed. Controls prescribed asthma medication were excluded. We included as controls those listed in CHCS as "no prescription information" and controls not found in the CHCS, since most probably were not in the CHCS because they did not use any prescription medications.
In verifying the random sample of 1000 controls, we excluded 3 (0.3%) who had no prescription information because they were listed as having died and 189 (18.9%) prescribed an inhaled bronchodilator, oral bronchodilator, inhaled steroid, or oral steroid. Of 808 controls remaining, we excluded 59 (7.3%) with no information for BMI and 5 (0.6%) with implausible data for BMI. This left 744 verified controls. Using our 386 verified cases and 744 verified controls, we performed a substudy case-control analysis.
Selected self-reported demographic (age, sex, race, marital status, and beneficiary status), lifestyle (smoking, drinking, and physical activity), and comorbid medical condition variables were studied. The BMI was calculated as weight in kilograms divided by the square of height in meters from self-reported weight and height in individuals 17 years and older and was classified as follows: normal, less than 25; overweight, 25.0 or higher; preobese, 25.0 to 29.9; class I obesity, 30.0 to 34.9; class II obesity, 35.0 to 39.9; and class III obesity, 40 or higher.5
We determined the prevalence of asthma for different age groups, sexes, racial/ethnic groups, marital status, and military status. We also determined the prevalence of possible confounding factors such as smoking, drinking, and physical activity.
For univariate analysis, we assessed the unadjusted association of asthma with other factors and compared proportions with the χ2 test with 2-sided significance at the .05 level. We calculated crude odds ratios (ORs) and 95% confidence intervals (CIs) as an approximation of relative risk estimates and to measure the association between overweight or obesity and prevalence of asthma.20,21
To evaluate trend patterns in the BMI association, we compared the proportions of enrollees with asthma within incremental categories of BMI and tested the differences using the χ2 test for trend.
We performed multiple logistic regression analysis using SPSS Base 8.0 for Windows (SPSS Inc, Chicago, Ill) to determine the odds of having asthma, given the BMI plus possible confounders such as physical activity. We adjusted for the demographic and lifestyle variables listed previously, related to the prevalence of asthma, obesity, and comorbid conditions in the analyses, by fitting a series of hierarchical models.22 We also adjusted for the following comorbid conditions: stomach ulcer, depression, arthritis, hypertension, cancer, diabetes, elevated cholesterol level, myocardial infarction, heart disease or angina, kidney disease, liver disease, neurologic disease, and stroke.
Compared with those without asthma, our initial analysis showed that TRICARE enrollees with asthma were more likely to be female and younger and less likely to be active duty service members and to engage in exercise at least 3 times per week (Table 1). Enrollees with and without asthma were similar in terms of race, marital status, smoking history, and drinking history.
Verified asthma cases were more likely to be female and married and less likely to be active duty service members compared with verified noncases (Table 1). The verified cases and controls were similar in terms of age, race, smoking history, drinking history, and exercise history.
In the overall analysis, TRICARE enrollees with asthma were more likely to be preobese (defined as BMI of 25.0-29.9) than enrollees without asthma, even when the analysis was simultaneously adjusted for age and sex (OR, 1.2; 95% CI, 1.1-1.4; Table 2). Similarly, in the substudy, TRICARE enrollees verified with asthma were more likely to be preobese than enrollees verified without asthma, even when the analysis was simultaneously adjusted for age and sex (OR, 1.4; 95% CI, 1.1-2.0; Table 3). Both in the entire group of subjects and the verified subgroup, asthma cases had an increased odds for being obese (defined as BMI ≥30) compared with controls, even when the analysis was simultaneously adjusted for age and sex. In the entire group, the risk of having asthma increased with increasing BMI, with the greatest increase among individuals with a BMI between 40 and 60 (OR, 2.8; 95% CI, 2.3-3.5; P for trend <.001). We found similar results in the verified subgroup of subjects in that verified asthma cases had increased odds of being obese, with the greatest increase in risk for asthma prevalence seen among individuals with a BMI between 35 and 40 (OR, 4.8; 95% CI, 2.6-9.1; P for trend <.001). In both analyses, adjustment for race had no material effect.
In our final multivariate regression model for the overall group, increasing BMI, younger age, non–active duty beneficiary status, female sex, and a history of being diagnosed as having stomach ulcer, depression, arthritis, or hypertension were significant independent predictors of asthma, with white race being a marginally significant independent predictor (Table 4). In our final multivariate regression model for the substudy, increasing BMI, younger age, non–active duty beneficiary status, female sex, and a history of being diagnosed as having arthritis were significant independent positive predictors of asthma prevalence (Table 5).
Asthma has various causal factors and multiple etiologic pathways, the interrelations of which are not clearly understood. In this prevalence case-control study, increasing BMI was linearly related to the risk of having asthma, independent of potential confounders. Other independent associated factors were younger age, non–active duty beneficiary status, female sex, and arthritis in both our main analysis and our verified substudy. Surprisingly, smoking history was not associated with asthma prevalence.
Stomach ulcer, depression, and hypertension are also independent predictors of asthma prevalence in our main analysis, whereas white race is a marginally significant predictor. It is unclear by what mechanism arthritis, stomach ulcer, depression, or hypertension may be independent predictors, and verification of these medical conditions would be necessary in future studies.
An association between gastroesophageal reflux disease and asthma has long been recognized.23 The possible mechanisms that may explain this association include the following: gastric contents enter the bronchial tree, triggering spasms; gastric contents rise into the esophagus, triggering bronchospasm by stimulating the vagus nerve; and both mechanisms may be at work in some patients.24 The HEAR questionnaire did not ask patients if they had ever been diagnosed as having gastroesophageal reflux disease, so it is unclear if patients who responded that they had been diagnosed as having stomach ulcer included patients who had been diagnosed as having gastroesophageal reflux disease.
There have also been many studies25-27 showing an association between asthma and psychopathologic conditions, particularly depression and anxiety, with estimates of psychopathologic conditions in severe asthmatic patients ranging from 30% to 63%. It is unclear whether patients with asthma have higher rates of depression and anxiety as a consequence of asthma or whether they have a comorbid genetic risk for affective or anxiety disorders.28
Asthma is a reason for exclusion from active military service, which likely explains beneficiary status as a family or retired member as significant independent positive predictors of asthma. This may also be related to our finding that female sex was an independent positive predictor of asthma, since active duty service members are still predominantly male. It follows that the female TRICARE region 11 enrollees are more likely to belong in the family member category as spouses of active duty service members, thus women having asthma are less likely to have been selected out in our population.
Few studies have been published regarding common correlates in patients with asthma. Bailey et al8 noted obesity was a common comorbidity in 479 adults with asthma seen by the University of Alabama at Birmingham Asthma Program. A limitation noted by the authors was that University of Alabama at Birmingham is a tertiary care institution; thus, the population of patients seen may not be representative of the general population.
Other published studies have focused on children. Luder et al17 examined 209 black and Hispanic children seen in a New York City medical center and compared them with 1017 black and Hispanic children enrolled in city schools. They found that the prevalence of overweight was significantly higher in children with moderate-to-severe asthma than in their peers and that being overweight was associated with significantly more severe asthma symptoms and outcomes (increased number of school days missed per year, peak expiratory flow rate [PEFR] less than or equal to 60% of the predicted PEFR, and increased asthma medications prescribed).
Unger et al7 noted that 18 (30%) of 61 obese children from Chicago, Ill, had asthma. The authors suggest that children may limit exertion to control exertional asthma symptoms, thereby reducing energy (caloric) expenditure and promoting weight gain.
Kaplan and Montana11 tested children from Miami, Fla, with no history of asthma for spirometry performance before and after an exercise challenge on a treadmill and found a greater frequency and degree of exercise-induced bronchospasm in 13 obese children compared with 14 control children. Eleven (85%) of the 13 obese children compared with 6 (43%) of the 14 control children showed a minimum decrease of 15% after exercise challenge for any of the following pulmonary function test parameters assessed: forced expiratory volume in the first second, forced expiratory flow between 25% and 75%, and PEFR. They questioned whether exercise-induced bronchospasm leads to exercise avoidance and obesity or whether obesity causes or enhances bronchial hyperreactivity to exercise.
A population-based cohort study by Siersted et al18 in Odense, Denmark, studied 495 children, half selected at random and half with a history of asthma. Subjects completed a questionnaire and underwent testing that included peak expiratory flows, anthropometric measurements, spirometry, treadmill exercise testing, and methacholine challenge. The study found undiagnosed asthma was associated with low physical activity and high BMI.
In a prospective cohort of 85 911 female nurses aged 26 to 46 years, Camargo et al15 found that after controlling for 9 potential confounders (including age, race, smoking, physical activity, and energy intake), BMI was a strong risk factor for asthma (relative risk, 2.7 for BMI ≥30.0 vs BMI 20.0-22.4; 95% CI, 2.3-3.1). In a prospective cohort of 16 862 children of female nurses, the same authors16 found that after controlling for Tanner stage, the relative risk of asthma for the highest compared with the lowest BMI quintile was 2.3 (95% CI, 1.3-4.1) among boys and 1.5 (95% CI, 0.9-2.6) among girls.
There are several strengths to our study. Our study was population-based, since all enrollees in TRICARE region 11 were asked to complete the questionnaire. Thus, our findings may be generalized to other similar populations. The large sample size allowed sufficient power so that we were able to simultaneously control for many potential confounders. Verification of self-reported asthma diagnoses was accomplished through linkage to a medication database.
There are also several limitations in this study. Similar to other studies, we were not able to answer the key question of causality; it is not clear if obesity or asthma occurs first. There are several potential explanations for the observed association between asthma prevalence and obesity: (1) asthma may lead to exercise avoidance and obesity; (2) obesity might cause or enhance bronchospasm; (3) severe asthma cases who die or experience extreme morbidity may be underweight—leaving asthmatic patients who survive and are healthy enough to complete a questionnaire with a higher average BMI; and (4) obese individuals may have altered immune function, which could lead to respiratory hyperreactivity. One mechanism proposed by Fredberg et al29 for the second potential explanation is that obesity may lead to an altered pattern of tidal lung inflation and/or less effectiveness of those inflations in stretching airway smooth muscle, which may result in hyperreactivity.
Missing or implausible BMI data led to the exclusion of 7.6% of the 2788 cases and 8.3% of the 39 637 controls in the general study and similar percentages in the substudy. Since the response rate on the HEAR questionnaire was only about 50%, there could have been selection bias, in which subjects with multiple medical problems such as stomach ulcer and hypertension were more likely to respond, thereby creating artifactual relations. Since asthma and obesity are both reasons for rejection from military service, the findings in this study may not reflect the general population. However, the prevalence of overweight and obesity in our study population closely reflects the prevalence of overweight and obesity in the US population.5
In view of the fact that asthma and obesity are both reasons for rejection from military service, it is interesting to observe that retired service members have twice the risk of asthma as active service members. Our data also indicate that obesity is more prevalent in retired service members who are asthmatic (15.9%) than in active service members who are asthmatic (7.3%). Although one cannot rule out the possibility that some of these retired service members may have developed asthma, then decreased their physical activity, and subsequently became obese, it is possible that retired service members decrease the physical activity they maintained during their active service years, increase their BMI, and then develop asthma. This question of causality may be answered with a prospective study in which a cohort with initial baseline pulmonary function test and BMI measurements is followed up with repeated measurements throughout subsequent years. Unfortunately, it is difficult to conduct a prospective study in a military population due to the frequent moves of active service members.
Since this study is questionnaire-based, it is possible that unverified asthma could lead to misclassification of patients. Verification analysis indicated that two thirds of the randomly sampled cases had been prescribed an inhaled bronchodilator, oral bronchodilator, inhaled steroid, or oral steroid, suggesting that most of the self-reported cases truly have asthma. It is possible that unverified BMI may lead to misclassification of overweight patients as not being overweight. Furthermore, women of reproductive age may be pregnant, with a resulting increased BMI. Indeed, 54.4% of overweight and obese women were in the 20- to 39-year age group compared with 45.1% of overweight and obese men.
This study was limited in that the racial/ethnic distribution was predominantly white. Furthermore, it was not possible to determine from the HEAR questionnaire if the study groups differed in socioeconomic status, because there were no questions available on income or educational level. The TRICARE system tends to ensure that everyone in the military population has equal access to care, although in some instances, beneficiaries who are not on active service may be seen on a "space-available" basis.
In summary, increasing BMI is a key factor predicting prevalence of asthma and, if determined to be etiologically related to asthma incidence, is a potentially modifiable risk factor for this disease. Prospective cohort data are needed to sort out issues of causality, ie, does obesity lead to asthma or vice versa. Future studies may include interventional trials to determine if a decrease in weight will cause a decrease in asthma symptoms for overweight asthma patients. One small study30 performed in Finland involved 2 groups of 19 obese patients with asthma, with the intervention of a supervised weight reduction program in the treatment group, showing an improvement in lung function, symptoms, morbidity, and health status. More studies are needed to evaluate the asthma-BMI association in black and other minority populations and the general, nonmilitary population. These further studies are key to better education and management of patients with asthma. Instead of parents or clinicians discouraging strenuous exercise in obese children or patients with "exercise-induced bronchospasm" to manage asthma symptoms, better management may prove to be a program stressing reduced energy intake and increased physical activity aimed at achieving an ideal body weight, thereby reducing asthma symptoms.
Accepted for publication October 31, 2000.
Presented as a poster at the 16th Annual National Preventive Medicine Meeting, Arlington, Va, March 18-21, 1999, and the Third Annual Force Health Protection Conference, Baltimore, Md, August 7-11, 2000.
Most of this work was completed while Dr Young was a public health resident at Madigan Army Medical Center, as part of her requirement to earn the Master of Public Health degree, Department of Epidemiology, University of Washington School of Public Health and Community Medicine, Seattle.
The opinions or assertions herein are the private views of the authors and should not be construed as official or as reflecting the views of the US Departments of the Army, Navy, or Defense.
Corresponding author: Jeffrey D. Gunzenhauser, MD, MPH, Preventive Medicine Service, Madigan Army Medical Center, Tacoma, WA 98431-5000. Reprints: Sylvia Y. N. Young, MD, MPH, PSC 824, Box 2760, FPO AE 09623.