|Home | About | Journals | Submit | Contact Us | Français|
Asthma and obesity are major public health challenges of increasing importance. The prevalence of obesity (defined as body mass index [BMI] >= 30 kg/ m2) has increased steadily among adults in the United States and Europe.(1–4) Obesity prevalence approximately doubled from the early 1960's to the late 1990's, increasing from 13.4% to 27.6% among men and from 15.8% to 33.2% among women.(3) During the same time period, asthma prevalence also increased. From 1980 to 1996, self-reported prevalence of current asthma among adults 18 and older in the United States increased from 3.14 to 5.46%.(5)
Although it continues to be debated, the relationship between obesity and asthma has been suggested by a variety of studies in the epidemiological literature,(6–9) and is unlikely to be explained by changes in physical activity (10) or medication use.(11) Prospective studies suggest that obesity and weight gain increase risk for subsequent diagnosis of asthma among adults, particularly women,(10,12) although discussion continues as to the accuracy of these diagnoses and possible mechanisms underlying the association.(13) Improvement in asthma symptoms(14,15) and lung function(15, 16) after weight loss and weight-reduction surgery (16,17) has also been described. One randomized trial of an 8-week weight loss intervention saw significant improvements in FEV1, FVC, and symptoms maintained for one year after weight loss. (15)
Residents of urban and low-income neighborhoods in the United States are particularly affected by both asthma and obesity. The prevalence of adults and children who reported ever having asthma was 9.7% in the United States in 1997 (5) but was 12.0% in 2003 in New York City.(18) Urban centers such as Chicago, New York, and Phoenix contribute disproportionately to high national asthma mortality rates.(19) Obesity is also more prevalent among low-income neighborhoods.(20) Numerous studies have confirmed the relationship between low socio-economic status, community disadvantage, and obesity in developed nations.(21–23)
The aim of our study was to evaluate neighborhood differences in the relationship between asthma and obesity. Given the overlapping epidemics of obesity and adult asthma in low-income neighborhoods, we hypothesized that the population burden of adult asthma attributable to obesity would be greater in low-income than in middle-to-upper-income neighborhoods in New York City.
From May to June of 2002, the New York City Department of Health and Mental Hygiene conducted the Community Health Survey, a representative, population-based, random digit-dialed telephone survey of adult residents of the five boroughs of New York City. Trained investigators from the Baruch College survey research unit performed the interviews. One adult (age ≥ 18) was identified from each participating household. Studies were pre-translated into eight other languages. Interviews were discontinued if the quota for that neighborhood had been met. Ten attempts were made to reach each telephone number.(24) An estimated 2.6% of households in New York City did not have telephone service at any given time.(25) The overall cooperation rate was 64%, resulting in an unweighted sample size of 9,674.(26,27)
The current study sample was restricted to respondents age less than 55 years (n=6,664), in order to minimize misclassification with chronic obstructive pulmonary disease (COPD), and to respondents who reported BMI > 15 and < 60 kg/m2 and height > 50 inches (127cm; n=6,119), in order to exclude likely erroneous height and BMI values.
The Community Health Survey used a stratified sample explicitly designed to produce neighborhood-specific estimates. Neighborhoods were defined using the United Hospital Fund's 42 aggregations of New York City zip codes. These aggregates correspond to community planning districts, and were created in consultation with staff from the New York City Department of Health and Mental Hygiene, Department of City Planning, and others.(27) Several neighborhoods were combined in the design phase due to demographic similarity, resulting in a total of 33 neighborhood groups.(28)
The median annual family income for each of the 33 neighborhoods was estimated by using population-weighted zip-code-level data from the 2000 U.S. Census files.(29) Low-income neighborhoods were defined as neighborhoods in the lowest quartile of median family income in New York City.
Survey respondents were asked to report age, gender, race/ethnicity, educational attainment, weight in pounds, and height in inches. Respondents were asked, “About how much do you weigh without shoes?” and “About how tall are you without shoes?” BMI was calculated as weight (kg) divided by height (m) squared. Obesity was defined as BMI ≥ 30. Current asthma was defined as positive answers to both of the questions: “Have you ever been told by a doctor, nurse, or other health professional that you had asthma?” and “During the past 12 months, have you had an episode of asthma or an asthma attack?”(28) Of note, no further respiratory questions were included in the Community Health Survey.
Responses were weighted according to the probability of selection to participate in the survey (number of adults in each household / number of residential phone lines) and post-stratification weight, determined by weighting each record up to the population of the neighborhood, while taking into account the respondent's age, sex, and race. Neighborhood population was estimated using the 2000 U.S. Census files.(29)
We first calculated age-adjusted prevalences of current asthma and obesity weighted by sampling schema and corresponding one-sample 95% confidence intervals. Prevalence differences were calculated as the weighted prevalence of asthma among obese individuals minus the weighted prevalence of asthma among normal-weight individuals. A description of how multivariate risk differences, attributable risks, and risk ratios were calculated can be found in the Supplement.
Analyses were performed using SAS version 9.(30) The study was approved by the Institutional Review Boards of the Department of Health and Mental Hygiene and Columbia University.
The Community Health Survey included 6,119 respondents age 18–54 years with acceptable height and weight measures. This sample corresponded to a weighted population estimate of 3,947,000 persons in New York City. The prevalence of obesity was 16.1% (95% CI: 14.9%, 17.2%) and the prevalence of current asthma was 4.5% (95% CI: 3.9%, 5.1%).
Table 1 shows characteristics of the adult population estimates of New York City from this survey stratified by categories of BMI. Obese and very obese respondents (BMI > 30) were more likely to be of Black or Hispanic race/ethnicity, have lower educational attainment and income, and to have been born in the United States compared to normal weight individuals (BMI 18–25) (p<0.001 for all comparisons). Obese and very obese respondents were less likely to never have smoked (p = 0.075) and to have exercised in the past thirty days (p<0.001).
The age-adjusted prevalence of current asthma was 6.7% among obese (>30 kg/m2) individuals and was 4.0% among normal weight (18–25 kg/m2) individuals, yielding an age-adjusted prevalence difference for current asthma related to obesity of 2.7% (95% CI: 1.0%, 4.4%; p=0.001) across all of New York City. After additional adjustment for age, gender, race/ethnicity, smoking status and educational attainment, the multivariate prevalence difference for current asthma was 2.0% (95% CI: 0.5%, 3.6%; p=0.01).
Multivariate prevalence differences can be interpreted as attributable risks if the observed relationship is assumed to be causal and unique. If this assumption is made, the absolute risk of asthma among obese New Yorkers attributable to obesity was 2.0% (95% CI: 0.5%, 3.6%). The attributable risk percent, which is the percent of asthma cases among obese subjects attributable to obesity, was 33.8% (95% CI: 8.3%, 52.1%). The population attributable risk percent, which represents the percent of asthma attributable to obesity among all (obese and non-obese) New Yorkers, was 8.5% (95% CI: 2.1%, 13.1%).
Median annual family income in low-income neighborhoods was approximately $25,000, compared to approximately $56,000 in middle-to-upper income neighborhoods. Table 2 shows age, gender, race/ethnicity, education, and smoking status in low-income neighborhoods compared to middle-to-upper income neighborhoods. A map illustrating neighborhoods in New York City stratified by median household income (Panel A), prevalence of obesity (Panel B), and prevalence of asthma (Panel C) can be found in the Supplement.
The prevalence of obesity was 23.2% (95% CI: 20.9%, 25.6%) in low-income neighborhoods and 13.7% (95%CI: 12.4%, 15.0%) in middle-to-upper-income neighborhoods. The prevalence of current asthma was 5.8% (95% CI: 4.6%, 7.0%) in low-income neighborhoods and 4.1% (95% CI: 3.3%, 4.8%) in middle-to-upper-income neighborhoods.(Table 2)
In low-income neighborhoods, the prevalence of current asthma was 6.7% among obese individuals and 5.2% among normal-weight individuals, which yielded an age-adjusted prevalence difference of 1.5% (95% CI: −1.1%, 4.5%; p=0.24). In middle-to-upper-income neighborhoods, the prevalence of asthma was 5.7% among obese individuals and 3.5% among normal-weight individuals, which yielded an age-adjusted prevalence difference of 2.2% (95% CI: 0.3%, 4.2%; p=0.03).
After multivariate adjustment for age, gender, and race/ethnicity, the prevalence difference for asthma due to obesity was 1.3% (95% CI: −1.5%, 4.0%; p=0.36) in low-income neighborhoods and 2.0% (95% CI: 0.1%, 3.9%; p=0.04) in middle-to-upper income neighborhoods. Due to modest sample sizes and the modeling approach, the multivariate model did not converge when smoking status and educational attainment were included as covariates. Separate models in which smoking status or educational attainment were retained at the exclusion of other variables, however, did not differ in the direction or magnitude (data not shown).
The corresponding attributable risks, assuming causality, of 1.3% and 2.0% were therefore not higher among obese New York City adults living in low-income neighborhoods compared to in middle-to upper-income neighborhoods, respectively (P=0.66). On a percentage basis, the percent of asthma among obese individuals attributable to obesity was 20.6% (95% CI: −29.9%, 51.5%) in low-income neighborhoods and 37.9% (95% CI: 3.9%, 59.8%) in middle-to-upper-income neighborhoods.
The population attributable risk percent, which is the percent of asthma in the entire population attributable to obesity, was also not greater in low-income neighborhoods (7.3%; 95% CI: −10.6%, 18.3%) than in middle-to-upper-income neighborhoods (7.7%; 95% CI: 0.8%, 12.1%). These results suggest that the risk of asthma attributable to obesity among New Yorkers in low-income neighborhoods was not greater than that among New Yorkers in middle-to-upper-income neighborhoods.
We further examined the relationship of obesity and asthma further on a multiplicative scale for the city as a whole (Table 3) and stratified by neighborhood income (Table 4). Risk ratios from these multivariate models are of similar magnitude and in the same direction as prevalence difference and attributable risk results. Further description of these results can be found in the Supplement.
Obesity was confirmed as an important risk factor for current asthma in this study of the adult population of New York City. However, in spite of the higher prevalence of both asthma and obesity in low-income neighborhoods in New York City, the risk of asthma attributable to obesity was not higher in low-income neighborhoods compared to middle-to-upper-income neighborhoods. These findings make it unlikely that obesity causes excess asthma prevalence in low-income neighborhoods in New York City. Instead, alternative causes of excess asthma in low-income neighborhoods should be sought.
This study is the first to examine whether the association between obesity and adult asthma differs by neighborhood. Prior studies have shown an effect of neighborhood beyond individual health characteristics using ecologic or multi-level analysis,(31) but few studies have examined how neighborhood and individual characteristics may interact for specific health outcomes.(32) A recent study of the effect of aeroallergens on adult asthma hospitalizations found a stronger association among subjects with lower education levels, but neighborhood was not examined, and no difference was found by income level.(33) When the same group looked at air pollutants, neither education nor income level modified the effect of air pollution on asthma.(34) In a multi-level model created by Cagney and Browning, “physician diagnosis of a weight problem” was a significant predictor of asthma diagnosis but was unchanged by the addition of neighborhood characteristics, such as collective efficacy, to the model.(35)
Although the prevalence of obesity was higher in low-income neighborhoods, the relationship between obesity and asthma was weaker in low-income neighborhoods than in medium-to-high neighborhoods. Our finding suggests that other factors might contribute to the pathogenesis of asthma in low-income neighborhoods and modify or lessen the importance of obesity as a cause of asthma in these neighborhoods. Ambient air pollution, indoor air pollution, allergens, stress, and violence are all potentially worse in low-income compared to middle-to-upper income neighborhoods. Ambient air pollution is a well-established cause of emergency room visits and hospitalization for asthma.(36) Nitrogen oxides, ozone, and diesel exhaust are asthma triggers that may be higher in low-income neighborhoods, particularly with nearby bus and truck routes.(37–40) Indoor air pollution (41) and allergens, including cockroach allergen (42) and mouse allergen,(43) may be higher in low-income neighborhoods. Matsui and colleagues found that settled dust concentrations of mouse allergen were a log-fold higher in inner-city homes compared to suburban homes.(44) Stress is a well-described asthma trigger,(45) which is likely to be worse in low-income neighborhoods. Community violence may play a role as well; the Inner-City Asthma Study found an association between exposure to violence and asthma symptom-days in a model which controlled for socioeconomic status, housing deterioration, and negative life events.(46) Furthermore, Wright and Steinbach recently described cases from the Boston City Hospital where exposure to violence appeared to precipitate asthma symptoms.(47)
Recent work supporting a genetic mechanism for the link between asthma and obesity should also be considered (48). In an analysis of twin pairs in the University of Washington Twin Registry, Hallstrand and colleagues found that 8% of the genetic component of obesity is shared with asthma. While environmental determinants were considered in the analysis, neighborhood of residence was not explored. Genetic variation by neighborhood may contribute to the differences in asthma and obesity prevalence, but further studies are needed; our survey did not include questions about family history or collect genetic material.
A limitation of this study is its use of self-reported height and weight to calculate BMI. Other studies have validated the use of self-reporting in this setting, however. An examination of self-reported BMI in an adult Scottish population found a sensitivity and specificity of 83% and 96% respectively for men, and 89% and 97%, respectively, for women.(49) If the misreporting of BMI and asthma were severe in low-income neighborhoods and absent in middle-to-high-income neighborhoods, this might have resulted in a stronger asthma-obesity relationship in the latter neighborhoods. However, it is unlikely that misreporting were differential in that direction: multiethnic studies have found that nonwhite subjects and subjects with less education report weight and height more accurately than White, more educated subjects.(50)
The cooperation rate for the Community Health Survey was lower than desired. Although the Community Health Survey targeted a random sample of the New York City population, not all contacted households agreed to participate in the study. Additionally, the survey did not capture households without a telephone, cell phone-only households, or institutionalized adults.(28) The relationship between asthma and obesity could differ among respondents and non-respondents; however, it is unlikely that any resultant bias would be strong.. Furthermore, the cooperation rate for the Community Health Survey is comparable to that of other cross-sectional population-based studies.(51)
The use of self-reported asthma diagnosis is a further limitation of this study. We defined current asthma as a doctor or nurse diagnosis of asthma plus self-report of asthma symptoms in the past 12 months. While self-report of asthma symptoms may be subject to misclassification, this combination of questions may be specific enough to reduce differential misclassification. One study comparing histamine challenge to the question of asthma during the last year found a sensitivity of 55% and specificity of 96% for this question, compared to a sensitivity of 44% and specificity of 98% for bronchial hyperresponsiveness on histamine challenge.(52) A 2002 study in Central Harlem, New York City found an asthma prevalence of 14%, (53) which is higher than the asthma prevalence reported in our study (5.8% in low-income neighborhoods). This difference in asthma prevalence is not surprising, however, as our definition of asthma was more stringent. The Central Harlem study did not ask about symptoms in the past year and did not exclude subjects over the age of 55, which may have resulted in misclassification of COPD cases as asthma cases. Nonetheless, we cannot eliminate the possibility of misclassification of asthma cases due to imperfect self-report.
Our study is unique in that it encompasses a large urban population and includes a high proportion of participants living in the inner-city. Although almost 6% of adult New Yorkers living in low-income neighborhoods had current asthma and 23% of them were obese, we did not find a stronger relationship between current asthma and obesity in low-income neighborhoods. Our findings suggest that the epidemic of inner-city asthma is not attributable disproportionately to obesity; other causes, such as allergen sensitization and indoor air pollution, also require further exploration.
The authors wish to thank Thomas Matte, MD and Bonnie Kerker, PhD for their helpful comments on the manuscript.
FUNDING: Robert Wood Johnson Generalist Physician Faculty Scholars Award, National Institutes of Health, National Research Service Award T32-HP10025
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
CONFLICT OF INTEREST: none
AUTHORS' CONTRIBUTIONS Emily S. Tonorezos – study conception and planning, data analysis, manuscript preparation.
Adam M. Karpati – study conception and planning, creation of figures, manuscript preparation
Yuanjia Wang – study conception and planning, data analysis, manuscript preparation.
R Graham Barr – study conception and planning, creation of asthma supplemental questionnaire dataset, data analysis, manuscript preparation, and funding.
All authors contributed to the drafting of the report.