An association between obesity and the development of asthma in children and adults has been reported recently in several studiesReference Camargo, Weiss, Zhang, Willet and Speizer1–Reference Gililand, Berhane, Islam, MacConnell, Gauderman and Gililand4. However, whether obesity predisposes to bronchial hyper-responsiveness in pre-school children has not been clarified. Wheezing is a clinical manifestation more characteristic of asthmaReference Ball, Castro-Rodriguez, Griffith, Holberg, Martinez and Wright5 and pre-school children with wheezing have an elevated risk for the development of asthma later in lifeReference Martinez, Wright and Taussig6. Given this, the detection of an association between overweight and wheezing in pre-school children may indicate that the influence of obesity on the risk for asthma may occur quite early in somatic development. Such information would be of use in the design of preventive strategies to decrease the risk for the development of asthma, by defining the period where weight control should be implemented.
The present study reports a significant association between wheezing and overweight in a large sample of pre-school children living in small towns of the states of Bahia and São Paulo, Brazil.
Methods
The study was approved by the Ethics Committee of the School of Public Health of the University of São Paulo. Participant families were instructed about the objectives of the study and in all cases gave informed consent after such information had been obtained.
Selection of study locations
The present investigation is part of a larger study that focused on the living conditions, nutrition and health of children between 0 and 5 years of age living in small towns in the states of Bahia and São Paulo, Brazil.
The communities were selected based on the following criteria: high urbanisation rate, small territory extent ( < 500 km2), small population ( < 15 000 inhabitants) and an adequate representation of different geographic profiles, in order to reflect some of the economic and social diversity of Brazil.
The study was conducted between March 1999 and March 2000 in Bahia, and during November and December 2001 in São Paulo. In Bahia, the towns of São Felix, Itiruçu, Cipó, Milagres, Acajutiba, Gongogi, Santa Inês, Presidente Dutra and Serrolândia were studied, whereas Bady Bassit, Bofete, Jaborandi, Morungaba and Riversul were studied in São Paulo.
Sampling strategy
As mentioned above, the present investigation is part of a larger study which focused on more general conditions of health and socio-economic variables in children, with more emphasis on malnutrition. In Bahia, the determination of sample size took into account the prevalence of stunting in the Northeastern region of Brazil7, whereas in São Paulo the sample size was computed using as reference the predicted prevalence of stunting for each community based on the model proposed by Benício and Monteiro8. After selecting the communities, children were enrolled in the study using the following approaches. In Bahia, a two-stage sampling was employed, considering the home as the primary unit and the child as the secondary unit. In São Paulo, the strategy was to select the sample in a single-stage approach, i.e. obtaining data on all children under 5 years of age in selected homes. Table 1 presents the number of homes evaluated for each community and the corresponding sample fraction.
Table 1 Number of houses selected and the fraction of houses in rural and urban zones in each community
Data acquisition
Data on housing, socio-economic conditions and maternal and child health were obtained by trained personnel, who applied structured questionnaires to mothers. The interviewer was also trained to measure height and weight according to World Health Organization (WHO) recommendations9. Anthropometric measurements were determined twice, using adequate instruments (electronic scale and stadiometer), and the difference between these two measurements was kept below 0.1 cm and 100 g for height and weight, respectivelyReference Martorell, Habicht, Yarbrough, Guzman and Klein10.
Mothers were asked whether their children had presented any of the following clinical conditions in the previous 15 days: wheezing, upper respiratory tract infection (URTI), toothache, diarrhoea, fever, cough, prostration, or parasite elimination. In addition, information on previous hospital admissions (if present) was also recorded. Nutritional status was characterised using weight-for-height to define overweight and height-for-age to define malnutrition, using the National Center for Health StatisticsReference Hamil, Drizd, Johnson, Reed, Roche and Moore11 population as reference in accordance with WHO recommendations12. A deviation from − 2 Z-scores was selected as cut-off point. The software EpiInfo version 6.0 (Centers for Disease Control and Prevention, CDC) was employed for analysis of nutritional status.
Family income was expressed in terms of units of the minimum national salary (MNS), categorised in two levels: < 0.5 MNS or ≥ 0.5 MNS.
Statistical analysis
Descriptive statistics of the measured variables were computed for each town. Thereafter unconditional logistic regression models were calculated, using as dependent variable a binary indicator for wheezing and different combinations of predictive variables: overweight, age, sex, URTI, fever, parasite elimination. In addition to health data, models also incorporated the following indicators of housing and socio-economic conditions: sleeping alone in the bed (yes/no), water treatment, availability of electricity and state. Statistical analysis was done with the aid of the software Stata version 7 (StataCorp).
Results
A total of 3453 children were studied, 2439 (70.6%) in Bahia and 1014 (29.4%) in São Paulo. A slight predominance of boys was observed (50.4%). The age distribution of the studied sample is shown in Fig. 1.
Fig. 1 Age distribution of the studied children from 14 towns in Brazil, 2000–2001 (SD – standard deviation)
For the overall sample, the prevalence of malnutrition was 9.5% (95% confidence interval (CI) 8.4–10.6). Malnutrition was more frequent in Bahia (11.3%, 95% CI 9.9–12.8) than in São Paulo (5.2%, 95% CI 4.0–6.9).
The prevalence of overweight was 3.2% (95% CI 2.7–3.9). There were differences in the prevalence of overweight between states, being estimated as 1.8% (95% CI 1.3–2.4) in Bahia and 6.7% (95% CI 5.3–8.5) in São Paulo. Descriptive statistics of the collected data are presented in Table 2.
Table 2 Prevalence and 95% confidence interval (CI) of clinical conditions exhibited by the children enrolled in the study. Brazil, 2000–2001
URTI – upper respiratory tract infection.
Table 3 depicts crude odds ratios for reporting of wheezing and each individual explanatory variable. Table 4 shows the odds ratios for reporting of wheezing and overweight (yes/no), both in the univariate model and in the multivariate model when all parameters depicted in Table 3 were considered simultaneously. This last model showed that overweight children had a greater frequency of reported wheezing, with an odds ratio of 2.57 (95% CI 1.51–4.37).
Table 3 Crude odds ratio (OR) and 95% confidence interval (CI) for the association between wheezing and each of the clinical, housing and socio-economic conditions in pre-school children from 14 towns in Brazil, 2000–2001
URTI – upper respiratory tract infection; MNS – minimum national salary.
Table 4 Odds ratio (OR) and 95% confidence interval (CI) for the association between overweight and wheezing, crude (Model 1) and adjusted (Model 2) for potential confounding variables, in pre-school children from Brazil, 2000–2001
* Explanatory variables: sex, age in two categories, upper respiratory tract infection, loss of appetite, fever, parasites, hospital admission in last 12 months.
Discussion
The results of the present study disclosed a significantly higher report of wheezing in overweight pre-school children. This association was robust and was not affected by the inclusion of different controlling variables. In fact, the magnitude of the odds ratio was not modified significantly by the inclusion of several explanatory variables (Table 4), indicating that overweight pre-school children were two times more prone to present wheezing in the 15 days before the questionnaire was applied. Moreover, the association between obesity and wheezing was insensitive to the criteria employed to define obesity. Using the criteria of Cole et al. Reference Cole, Bellizzi, Flegal and Dietz13 and the CDCReference Kuczmarski, Ogden, Grummer-Strawn, Flegal, Guo and Wei14, the coefficients relating obesity to wheezing remained virtually the same. This finding reinforces the relationship between these two conditions, i.e. risk for wheezing and obesity, each of which is exhibiting an increasing frequency in childrenReference Popkin, Richards and Monteiro15, Reference Nafstad, Magnus and Jaakkola16.
Obesity is far from being a problem of developed countries. In fact, there is clear evidence of an increasing frequency of obesity in Brazilian childrenReference Abrantes, Lamounier and Colosimo17–Reference Saldiva, Escuder, Venancio and Benicio20. In our study, obesity in pre-school children was also detected, being higher in São Paulo (6.7%).
Classically, programmes against obesity in children were motivated to prevent bone and joint problems, as well as cardiovascular, endocrine and psychological conditionsReference Barlow and Dietz21. The possible role of obesity in increasing the risk of asthma provides additional justification to prevent obesity in children. According to the present results, the prevention of obesity must be made at an early age if avoidance of asthma is the health endpoint.
The increased risk of obese children to present recent episodes of wheezing found in the present investigation – 2.57 (Table 4) – was also observed in previous studies despite the age groups and the criteria employed to define obesity being different. An investigation of 9357 children between 5 and 6 years of age living in rural communities in Germany disclosed a relative risk for asthma of 2.12 and 2.33 for overweight and obese children, respectivelyReference von Kries, Hermann, Grunert and von Mutius3. In the UK, a study of 14 908 children revealed that obesity was associated with a risk for asthma of 1.28Reference Figueroa Munoz, Chinn and Rona2. In the USA, the risk for asthma in obese children ranged from 2.2Reference Gold, Damokosh, Dockery and Berkey22 to 8.1Reference Wang and Dietz23. Thus, although recognising that the present study focused on wheezing and not asthma, there is an apparent concordance between the findings observed in our population and those of earlier studies relating obesity to episodes of bronchial hyper-responsiveness in children, even at early ages as that of the present investigation.
The pathogenic mechanisms responsible for the increased frequency of asthma are not yet determined. Several mechanisms have been proposed, including reduction of airway size, different eating habits, sedentary life and changes in hormone levelsReference Shaheen, Sterne and Montgomery24. It is important to stress that wheezing in pre-school children may be caused by other pathological conditions, such as respiratory infections, intestinal parasites and gastro-oesophageal refluxReference Sears25. Other conditions such as exposure to tobacco smoke, low birth weight and low maternal age also increase risk for wheezingReference Klinnert, Price, Liu and Robinson26. However, in the age group evaluated in this study, the precise diagnosis of asthma is problematic. In fact, a recent study in São Paulo showed that 93% of pre-school children with episodes of wheezing were later in their lives diagnosed as having asthmaReference Benicio, Ferreira, Cardoso, Konno and Monteiro27. This finding indicates that wheezing is indeed a significant risk for asthma in the age group focused upon in our studyReference Clough, Keeping, Edwards, Freeman, Warner and Warner28.
In conclusion, obesity was shown to be a significant factor for wheezing in a large sample of pre-school children. Despite the limitations of a cross-sectional study and other restrictions already pointed out in our discussion, the large sample and maintenance of the magnitude and significance of the association across different model specifications are indicative that such association may be causal.
Acknowledgements
Sources of funding: Financial support was provided by the Financiadora de Estudos e Projetos (FINEP/MCT-Brazil).
Conflict of interest declaration: None.
Authorship responsibilities: S.R.D.M.S. contributed towards the design of the study, coordinated the data collection in São Paulo state, conducted the statistical analyses and wrote the first draft of the paper. M.M.E. contributed towards the design of the study and the analyses in São Paulo state. S.I.V. contributed towards the study design. M.H.A.B. contributed towards the study design and coordinated the study in São Paulo state. A.M.O.A. contributed towards the study design and coordinated the study in Bahia state. L.P.M.O. contributed towards study design and data collection. M.L.B. conceived the study, contributed towards study design, and supervised the data collection, analysis and interpretation. All authors contributed in interpretation of the results and writing the papers.
Acknowledgements: We wish to thank the administrations of the 14 studied municipalities for logistic and operational support during the data collection.
