Copyright ©ERS Journals Ltd 2007 Socioeconomic status, asthma and chronic bronchitis in a large community-based study1 Centre for Research in Environmental Epidemiology, Municipal Institute of Medical Research (IMIM), and 3 Dept of Experimental and Health Sciences, Pompeu Fabra University, Barcelona, Spain. 2 Centre for Public Health Research, Massey University, Wellington, New Zealand. 4 Dept of Respiratory Epidemiology and Public Health, Royal Imperial College London, London, UK. 5 Dept of Medical Sciences, Uppsala University, Uppsala, Sweden. 6 Dept of Social Medicine, Medical School, University of Crete, Heraklion, Crete, Greece. CORRESPONDENCE: L. Ellison-Loschmann, Centre for Research in Environmental Epidemiology (CREAL), Municipal Institute of Medical Research (IMIM), 80 Dr Aiguader Rd, Barcelona 08003, Spain. Fax: 34 932216448. E-mail: eellison-loschmann{at}imim.es Keywords: Asthma, atopy, bronchitis, socioeconomic status
Received: August 3, 2006
The present study investigated the relationship between socioeconomic status, using measures of occupational class and education level, and the prevalence and incidence of asthma (with and without atopy) and chronic bronchitis using data from the European Community Respiratory Health Survey (ECRHS). Asthma and chronic bronchitis were studied prospectively within the ECRHS (n = 9,023). Incidence analyses comprised subjects with no history of asthma or bronchitis at baseline. Asthma symptoms were also assessed as a continuous score. Bronchitis risk was associated with low educational level (prevalence odds ratio (POR) 1.9; 95% confidence interval (CI) 1.42.8) and occupational class (1.8; 1.22.7). Incident bronchitis also increased with low educational level (risk ratio (RR) 2.8; 95%CI 1.55.4). Prevalent and incident asthma with no atopy were associated with low educational level. Subjects in the low occupational class (incident risk ratio (IRR) 1.4; 95%CI 1.21.7) and education group (IRR 1.3; 95% CI 1.11.6) had higher mean asthma scores than those in higher socioeconomic groups. Lower educational level was associated with increased risk of prevalent and incident chronic bronchitis and asthma with no atopy. Lower socioeconomic groups tended to have a higher prevalence and incidence of asthma, particularly higher mean asthma scores. Adjustment for variables associated with asthma and bronchitis explained little of the observed health differences by socioeconomic status. The relationships between socioeconomic status (SES) and asthma prevalence and incidence are not well understood. Previous studies in adults have reported no association 1, 2, while others have reported an increased asthma prevalence with lower SES 3, 4. Some of the inconsistencies may be due to a lack of standardisation between studies, particularly with regard to definitions and measurement of asthma and SES. Not only are there difficulties in defining asthma 5, but in addition the relationship between asthma prevalence and incidence is not easy to disentangle 6, 7. Furthermore, as with other chronic conditions such as diabetes and coronary heart disease, asthma may have shifted from being more prevalent among the affluent to becoming a condition more strongly associated with poverty in recent years 8, 9. Additionally, differing patterns of SES have been observed in the prevalence of atopic and nonatopic asthma 10, 11. In general, little is known about the pathways and mechanisms by which SES affects respiratory disease in adults. A number of risk factors that may be involved in the interrelationship between SES and asthma and chronic bronchitis have been identified: 1) smoking 12; 2) exposure to environmental tobacco smoke (ETS) 13; 3) mould or mildew in the home 14; 4) allergen sensitisation 15; and 5) obesity 16. Some of these factors, e.g. tobacco smoke 12, show a stronger association with chronic bronchitis than with asthma. The European Community Respiratory Health Survey (ECRHS) previously examined the relationship between SES and asthma prevalence 3. An increased asthma prevalence amongst lower socioeconomic groups was observed at the individual level, with education also being a determinant of asthma risk at the centre level. The ECRHS II study was undertaken 10 yrs later to assess changes over time in the prevalence and incidence of asthma and associated respiratory symptoms. The objective of the current analysis was to investigate the relationship between SES, based on measures of occupational class and educational level, with the prevalence and incidence of asthma (with and without atopy) and chronic bronchitis.
Study population The ECRHS sampling framework includes a random and asymptomatic sample. Details have been described elsewhere 17, 18. ECRHS I subjects were 2044 yrs of age and randomly selected from the general population in centres from throughout Europe, the USA, Australia and New Zealand during 19911993. All participants completing ECRHS I were invited to take part in a follow-up study, ECRHS II, during 19992001. The study population for the current analyses comprises those subjects who participated in both surveys and had occupational information collected in ECRHS II (28 centres from 13 countries).
SES was based on the subject's occupation and education level. Occupational class was derived from the longest-held job during the follow-up period between ECRHS I and II. Categories were based on the major group classification, using the first digit of the International Standard Classification of Occupations (ISCO) 19. If a subject held multiple jobs for the same time duration during the follow-up period, then the lower ISCO category (i.e. higher skill level) was used. The categories were: I for managers and professionals (nonmanual) of major groups 1 and 2; II for technicians and associate professionals of major group 3; III for other nonmanual workers of major groups 4 and 5; IV for skilled manual workers of major groups 6 and 7; V for semi-skilled or unskilled manual workers of major groups 8 and 9; and VI for unclassifiable or unknown. Occupational class group VI comprised any individual not occupationally active during follow-up or who could not be assigned an ISCO code. Each occupational class is presented in table 1
Educational level was based on age of the subject at completion of full-time study. To enable comparability of education level between countries, country-specific tertiles were constructed to provide a relative educational level measure, therefore, the cut-points for each country are different. Tertiles of education level were categorised as high (reference category), medium and low.
The prevalence analyses included 9,023 subjects (response rate 59%; fig. 1
Prevalent bronchitis was defined as the presence of both cough and phlegm on most days for 3 months during the previous year 20. Discordant responses (n = 871), i.e. subjects reporting at ECRHS II either only chronic cough or only chronic phlegm but not both, were excluded. No subjects reported both chronic cough and chronic phlegm in Tartu (Estonia), so this centre was excluded from the analysis (n = 259), leaving 7,915 subjects. The cumulative incidence of asthma was defined as the proportion of subjects without asthma symptoms at ECRHS I who subsequently reported asthma symptoms at ECRHS II. In total, 1,743 subjects were excluded after reporting any of the following symptoms: current asthma and/or shortness of breath or wheeze (with no cold) at the time of ECRHS I. A further 1,604 subjects were excluded due to missing data on atopic status, leaving 5,645 subjects in the incident asthma analyses. Both the asthma prevalence and incidence analyses were stratified according to atopic status. The cumulative incidence ratio for chronic bronchitis was calculated based on the proportion of subjects having neither cough nor phlegm at ECRHS I who then reported having both symptoms at ECRHS II. A total of 1,796 subjects were excluded who responded "yes" to having cough or phlegm at ECRHS I. Subjects with discordant responses to the questions on cough and phlegm (n = 470) were excluded, in addition to respondents from Tartu (n = 178) and Bordeaux, France (n = 124), where there were no incident cases of bronchitis reported for the follow-up period, leaving 6,455 participants. In the incidence analyses, responses to six questions on asthma symptoms were combined into an asthma score ranging 06 7. The items were: 1) breathless while wheezing in the previous 12 months; 2) waking with a feeling of chest tightness in the previous 12 months; 3) attack of shortness of breath at rest in the previous 12 months; 4) attack of shortness of breath after exercise in the previous 12 months; 5) waking by attack of shortness of breath in the previous 12 months; and 6) the presence of asthma ever. These analyses were conducted in those subjects (n = 5,924) reporting none of the six asthma symptoms at baseline.
Study variables Objective measurements of the subject's height and weight were obtained in both the ECRHS I and II questionnaires 21. Body mass index (BMI, kg·m2) was calculated as weight (in kg) divided by the square of height (in m) 21. Information on smoking status was obtained at each ECRHS survey. Participants were divided into three categories: nonsmokers, ex-smokers and current smokers. To assess levels of ETS, participants were asked about regular exposure to cigarette smoke in the previous 12 months. Rhinitis was classified using the question: "Do you have any nasal allergies, including hay fever?" Occupational exposures were defined as exposure to biological dusts, mineral dusts, gases or fumes during the follow-up period 22 and classified as none, low or high exposure.
Statistical methods Interaction terms were included to determine whether the associations of occupational class and educational level with health outcomes were the same in males and females. The interaction terms for educational level and occupational class were not significant (p = 0.21 and p = 0.18, respectively) for either asthma or bronchitis (p = 0.52 for educational level and p = 0.65 for occupational class). Thus, the results are presented with the data for males and females combined.
Prevalence Table 1 Almost one third of subjects belonged to occupational class I (managers and professionals) ranging from 15% in Verona (Italy) to 49% in Paris (France). Approximately 6.9% of subjects were unclassified. Of these, 44% were housepersons and 30% were currently employed but without an occupational ISCO code. The remainder were distributed amongst the unemployed, in poor health, retired or student categories. Heterogeneity was assessed in the association between education level and asthma prevalence by measuring the prevalence of asthma against the percentage of low or medium educational level by country (adjusted for age, sex and centre). No heterogeneity was found for either the medium (p = 0.76 for heterogeneity) or low (p = 0.93 for heterogeneity) education categories.
Table 2
Treatment and healthcare utilisation among asthmatics (data not shown) were examined and no differences were found according to occupational class. With regard to educational level, some small nonsignificant differences were observed, e.g. asthmatics in the low education group were less likely to have been prescribed medicines for their breathing (OR 0.77; 95% confidence interval (CI) 0.511.2) or to have seen a doctor (OR 0.72; 95% CI 0.501.03) compared with those in the high education group.
Cumulative incidence As with the prevalence analyses, the present authors modelled effect estimates for heterogeneity to assess the association between asthma incidence and education group at the country level. For the medium educational and low educational levels there was no heterogeneity (p = 0.97 and p = 0.72, respectively) with asthma incidence. The present authors also assessed heterogeneity in the association between bronchitis and education group at the regional level (due to small numbers in some countries) composed of Scandinavia, Central Europe, Southern Europe and English-speaking countries. p-Values for heterogeneity were 0.18 and 0.52 for the medium and low education groups, respectively.
Generally, no large differences were observed for cumulative incidence of respiratory symptoms by occupational class (table 3
Table 4
The present authors examined the prevalence of respiratory symptoms in ECRHS II and the cumulative incidence of respiratory symptoms in relation to occupational class and educational level in the 10-yr follow-up period between ECRHS I and II. Prevalent bronchitis was increased in low occupational classes, while low educational level was associated with an increased risk of both prevalent and incident bronchitis. Lower socioeconomic groups tended to have a higher prevalence (particularly for asthma with no atopy) and incidence of asthma, with higher mean asthma scores. Known risk factors for asthma and chronic bronchitis explained only a small part of the observed differences by SES. Some 3, 4, but not all 1, 25, studies have reported an increased risk of asthma with lower SES. ECRHS I found an increased prevalence of asthma in low SES groups 3, with the odds ratios being higher than those found in the current analyses. This difference is probably a combination of different sampling, since ECRHS II includes only a subset of ECRHS I, and improved living and working conditions and availability of treatments. It is unlikely that education directly affects the risk of developing respiratory symptoms, but it may capture long-term influences of early-life circumstances on adult health and is a predictor of future employment and income 26. There are difficulties in the comparability of educational achievement across countries where changes in the education systems within populations and differences in the meanings of various educational categories between populations may vary 27. Previous ECRHS analyses of SES 3 used tertiles of educational level, based on the age of the subject at completion of full-time study, with the same cut-off points applied across the whole ECRHS study population. In the current analyses, tertiles specific for each country have been calculated to provide a relative measure of educational level and minimise problems associated with educational levels having different meanings in different countries, which is only partially solved by adjusting for country. Using tertiles calculated over the whole ECRHS population yielded little difference in the risk estimates; however, the results were less consistent in terms of the direction of the gradient seen between high, medium and lower educational level and increased risk for all respiratory outcome measures compared with the results using country-specific tertiles. Using IRR in the analyses, no association was found with occupational class and asthma risk, but an effect was seen when asthma symptoms were analysed as a continuous score. With a condition such as asthma, where there is a high prevalence and low incidence, bias due to disease misclassification may be substantial 7. The higher mean asthma scores with lower occupational class suggest that misclassification of asthma status at baseline may explain the absence of an association between asthma incidence and occupational class when the IRR measure was used. The present findings are consistent with Montnémery et al. 1 who examined social position as a risk factor for asthma and chronic bronchitis in a random sample of 12,071 adults. Montnémery et al. 1 found an increased risk of bronchitis, but not asthma, in those individuals with a low social position compared with a middle/high social position. Chronic bronchitis has been found to be more consistently associated with lower social class 28 and unemployed people have a higher risk of bronchitis-type symptoms than their employed counterparts 20. Some of the observed associations with occupationally defined social class may be due to respiratory symptoms caused by occupational exposures 29, although several studies have reported that confounding by occupational exposure does not fully explain this association 30. A socioeconomic gradient has been reported with smoking, an important risk factor for bronchitis 30. No statistically significant interaction between either occupational class or educational level and smoking status was found, suggesting that the findings for SES and bronchitis were not dependent on smoking status. The response rate for the current study was 59%, ranging 2580%, across the participating centres and thus the potential for selection bias must be acknowledged. There were no differences between responding and nonresponding subjects by sex, but subjects from a high occupational class were more likely to respond (63%) than those from a low occupational class (57%). Responding subjects with asthma were slightly more likely to participate than those without asthma (62 versus 60%, respectively). The reverse pattern was seen for chronic bronchitis, with a higher proportion of those responding reporting no bronchitis at baseline (60%) compared with those with bronchitis (56%). In total, 22% of subjects were excluded due to missing data on atopy. The present authors assessed the effect of this by comparing the results among the study population, including those with missing atopy data, and among the population with atopy data. The results did not change, however, as no significant difference was found with occupational class (p = 0.88) or educational level (p = 0.81) for those with and without atopy data. There may have been some misclassification of asthma or bronchitis, as defined by the questionnaire which has been previously validated against bronchial hyperresponsiveness 31. The overall effect of this type of misclassification would be to underestimate the true association of asthma or bronchitis with SES. Several potential explanatory factors were integrated in the fully adjusted models, including obesity, respiratory infections in childhood, exposure to allergens, smoking and exposure to ETS, which have been identified as being more common among lower SES groups 8, 28. It is possible that some of these factors may be intermediate variables on the causal pathway between lower SES and asthma or bronchitis, and may be highly correlated with each other. In that case, it would be expected that risk associations would reduce with widening CI. However, there were no dramatic changes seen in either the risk estimates or CI between the minimally and fully adjusted models; e.g. the minimally adjusted RR estimate for asthma in the low education group was 1.32 (95% CI 0.991.77), which changed to 1.31 (95% CI 0.971.77) when BMI was added to the model. Inclusion of any one of the explanatory variables used in the fully adjusted model did not change the minimally adjusted risk estimate by >10%. In conclusion, the present study identified lower educational level to be associated with an increased risk of prevalent and incident chronic bronchitis and also with an increased risk of prevalent and incident asthma with no atopy. Lower socioeconomic groups had higher mean asthma scores, suggesting that misclassification of asthma status at baseline and follow-up may explain some of the absence of an association between asthma incidence and occupational class in these analyses. Adjusting for potential explanatory variables related to socioeconomic status did not modify much of the association, suggesting that other factors in adult life or in childhood may mediate the occurrence of socioeconomic differences in respiratory disease.
This article has been cited by other articles:
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||