Abstract
Inequalities in health between socio-economic groups are a major public health concern. The current authors studied associations between parental socio-economic status (SES) and children's respiratory and allergic symptoms in 13 diverse countries, including the Russian Federation, North America (Canada and the USA), and countries across Eastern and Western Europe.
Data of 57,000 children aged 6–12 yrs, originating from eight cross-sectional studies, were analysed. SES was defined by parental education. Respiratory and allergic symptoms were defined by parental questionnaire reports.
Multiple logistic regressions showed that low parental education was associated with a decreased risk of inhalant allergy and itchy rash in school children. Furthermore, low parental education was associated with an increased prevalence of wheeze and nocturnal dry cough. No clear association was found between parental education and prevalence of doctor-diagnosed asthma and bronchitis. Part of the difference between socio-economic groups with regard to their children's symptoms was explained by established risk factors, such as parental allergy, smoking during pregnancy, pet ownership, crowding, mould/moisture in the home, use of gas for cooking, and air pollution (particulate matter with a diameter of <10 µm). However, differences remained after adjusting for these variables.
Children's health was associated with parental education. The association could not fully be explained by established risk factors.
Inequalities in health between socio-economic groups are a major public health concern. Numerous epidemiological studies have found higher rates of mortality and morbidity among groups with a lower socio-economic status (SES) defined by education, income or occupation within countries and across countries 1–8. Social inequalities in health are not confined to poor health among the most deprived. Marmot and coworkers 9, 10 reported increasing mortality and morbidity with decreasing SES for British civil servants, not considered to be poor by any usual standard.
Allergies seem to be one of the exceptions with regard to their association with SES. A number of studies performed in industrialised Western countries reported decreased prevalence of eczema 11, 12, hay fever 11, 13, and atopy determined by skin-prick testing and measurements of specific immunoglobulin E 11, 14–16 among children and adults with low SES. The current authors' knowledge about the association between SES and health in populations of the formerly communist countries of Eastern Europe is limited. However, Heinrich et al. 17 and Krämer et al. 18 reported an increased prevalence of eczema, allergic rhinitis and allergy for children from highly educated parents living in the former East Germany, where differences between social groups with regard to income were relatively small. Asthma which is at least partly attributable to allergy 19 does not follow the same pattern as eczema, hay fever and allergic sensitisation. Mielck et al. 20 reviewed 24 studies on the association between childhood asthma and SES and found that the studies did not reveal a clear picture; positive associations were as frequent as negative ones, and most studies showed no association at all. The International Study on Asthma and Allergies in Childhood 21 reported higher prevalence of asthma for urbanised, more “Westernised” countries compared with less developed countries.
The mechanisms behind all these (positive and negative) associations between SES and mortality and morbidity are not clear. Differences in susceptibility and differences in exposure were proposed to influence, and to be influenced by, SES and each other 22. Both exposure and susceptibility may lead to unequal health outcomes, which can then cycle back to affect SES 22. A number of hypotheses have been proposed as potential partial explanations for between-country and within-country disparities in asthma prevalence including differences in “hygiene”, diet, cigarette smoking, traffic pollution, antenatal exposures, and physical activity/obesity, but their significance for explaining between-region or between-country disparities in asthma prevalence is uncertain 23. Within the European Union-funded Pollution and the Young (PATY) project a unique database was established, comprising data on 57,000 children aged 6–12 yrs, originating from eight cross-sectional studies performed in the Russian Federation, North America (Canada and the USA), and countries across Eastern and Western Europe. The current authors used this database to: 1) describe the association between SES and children's respiratory and allergic symptoms in these diverse countries; and 2) find out whether known risk factors for respiratory and allergic symptoms do account for differences between socio-economic groups in terms of their children's respiratory and allergic symptoms.
MATERIALS AND METHODS
Study design and study population
The present study was designed as a combined analysis of data collected within the framework of eight cross-sectional studies performed in Europe, North America, and the Russian Federation. Cross-sectional studies were considered which had assessed respiratory symptoms and individual risk factors by questionnaire, included cough and wheeze as primary health outcomes, and allowed the calculation of annual mean particulate matter measures by study area. Table 1⇓ gives an overview of the selected studies. Between two and 29 study areas were defined per country. These study areas were school catchments in some cases and geographic areas in others. Study areas are not necessarily representative for the respective countries. The study population of the present study was defined as the participants of the original studies who were between 6–12 yrs. Thus, data from 57,363 children from 126 study areas in 13 countries were analysed.
Health outcomes
The health outcomes of interest (respiratory and allergic symptoms of wheeze, asthma, bronchitis, nocturnal dry cough, inhalant allergy and itchy rash), collected in the original studies by parental questionnaire, were chosen on the basis of the compatibility of their wordings, their scientific interest and their availability across the studies. However, the wording of the questions used to define a certain symptom/disease was not exactly the same in all studies and some outcome variables are not available for all studies. The exact definition of the wording of the questions used for the definition of outcomes is given in Appendix 1.
Definition of socio-economic status
SES refers to an individual's relative position in the social hierarchy and can be operationalised as level of education, occupation and/or income 32. In the present study, parental education was used to define parental SES. Parental education was classified according to a standardised scheme developed by the Organization for Economic Cooperation and Development (OECD) into “none”, “elementary”, “secondary, first stage”, “secondary, second stage”, and “post-secondary” as described by Kunst and coworkers 33, 34. Based on this classification, three categories of low, medium and high parental education were defined at country level, taking frequency distributions into account to make sure that every country/study had a sufficiently large number of subjects with low and high parental education, respectively. The current authors aimed to construct groupings, which were as comparable as possible across the studies, though it was not possible to do this perfectly. The definition of educational levels is presented in table 2⇓. Mother's and father's educational levels were classified separately, and the maximum educational level achieved by either parent was then used in analyses hypothesising that the parent with the highest educational level is the determining factor for a family's SES.
Definition of explanatory variables
In the current study, variables that might explain potential associations between parental education and children's health have been termed “explanatory variables”. Potential explanatory variables, such as parental allergy, maternal smoking during pregnancy, breastfeeding, pet-ownership, crowding, mould/moisture in the child's home, and the use of gas for cooking were identified from review articles (e.g. von Mutius 35 and Strachan 36) and defined by parental questionnaire reports. Explanatory variables were defined as similarly as possible between countries from the information that was available. However, definitions differ between countries and not all explanatory variables are available for all countries. This might limit to some extent the present authors' ability to explain SES effects by these variables in some countries as well as the comparability of the explanatory variables adjusted SES effects between countries.
Parental allergy was defined as parental asthma in the past 12 months and/or parental eczema/skin or nasal allergy ever for Bulgaria, Czech Republic, Hungary, Poland and Slovakia; for Germany, Italy and Switzerland, it was defined as parental asthma ever and/or parental eczema/skin or nasal allergy ever; for the Netherlands it was defined as parental asthma ever and/or nasal allergy ever; for North America and the Russian Federation it was defined as parental asthma ever; and for Austria it was defined as asthma, skin allergy, nasal allergy or hay fever ever in either parents or sibling(s) of the child. Pet ownership was defined as “ever had a furry or feathery animal in the home” for Austria, Bulgaria, Czech Republic, Hungary, Poland, Slovakia, Italy, The Netherlands and Switzerland; as “current regular contact with furry or feathery animals” for Germany; and as “furry/feathery animal currently in the child's home” for the Russian Federation and North America. Crowding was defined as the number of persons per room >75th percentile for all countries except the Russian Federation, Switzerland and North America; for the Russian Federation it was defined as the number of persons per m2 >75th percentile; for Switzerland as “child shares bedroom”; and for North America as the number of persons per bedroom >75th percentile. Exposure to particulate matter with a diameter of <10 µm (PM10) was assessed by measurements at fixed ambient monitoring sites in the study areas.
Statistical analysis
All analyses were restricted to children with non-missing information on parental education. Information on parental allergy was missing for up to 13% of the countries' study populations. Therefore, “missing information on parental allergy” was defined as a separate category to include these participants in the analysis. Country-specific logistic regression analyses were performed to analyse the effect of parental education on children's respiratory and allergic symptoms on the country level. First, adjustments were made for sex and age only. Afterwards, the current authors adjusted for the explanatory variables listed above and study area (as a proxy for explanatory variables on the study area level) and PM10, respectively, to determine if these variables explained potential associations between parental education and children's health. The current authors then compared country-specific and mean sex, and age-adjusted odds ratios for the comparison of low versus high parental education with sex, age, study area, and explanatory variables adjusted odds ratios. Mean odds ratios were calculated with a fixed effects approach. In case of heterogeneity (p<0.10) mean odds ratios were calculated using the random effects approach described in DerSimonian and Laird 37. Countries are presented ordered by per capita gross domestic product in tables and figures.
For comparability reasons, all associations between symptoms and parental education are presented as odds ratios with 95% confidence intervals (CI) using high parental educational status as the reference category, although prevalence ratios may be more suitable for common outcomes. The present authors are aware of the fact that the odds ratio depends on the underlying prevalence and therefore largely avoided comparisons of odds ratios between countries.
RESULTS
Description of the study population
A description of the study population including parental education is presented in table 3⇓. Frequency distributions of parental education differed considerably between countries. The percentage of low parental education, as defined here, was lowest in Slovakia (3.1%) and highest in the Netherlands (30.5%); the percentage of high parental education was lowest in Germany (13.8%) and highest in the USA (67.8%).
Prevalence of symptoms and diseases
The frequency distributions of the outcome variables are presented in table 4⇓. Wheeze during the past 12 months, asthma and inhalant allergy are available for all countries. The remaining outcomes were available for all countries except two or three. Variability of prevalence between countries differed considerably by outcome.
Parental education and explanatory variables
The associations between parental education and potential explanatory variables are presented in figure 1⇓. Except for the Czech Republic and North America, the prevalence of parental allergy significantly increases with increasing level of parental education. Smoking during pregnancy and crowding were consistently more prevalent among families with a low parental education compared with high parental education. The reverse trend was seen for breastfeeding (except for Bulgaria), i.e. breastfeeding was more prevalent among families with a high parental education. Associations between parental education and pet-ownership, mould/moisture and use of gas for cooking were less clear.
Parental education and health outcomes
First, country-specific age and sex adjusted odds ratios for the association between parental education and respiratory and allergic symptoms were calculated (data not shown). Secondly, the current authors adjusted for a set of explanatory variables such as study area (as a proxy for explanatory variables the study area level), parental allergy, smoking during pregnancy, breastfeeding, pet-ownership, crowding, mould/moisture in the child's home, and use of gas for cooking that were associated with parental education and might account for the effect of parental education on children's health. The results are presented in figure 2⇓.
The additional adjustment resulted in a change in odds ratio (related to the age, and sex-adjusted odds ratio) of −25.4–23.3% (median −1.3%) and −69.9–26.7% (median −1.9%) for the comparison of medium versus high parental education and low versus high parental education, respectively. The absolute change in odds ratio exceeded 10% for 25.3% and 49.3% of the odds ratios, respectively. Statistically significant differences between low and high parental education with regard to the children's health remained statistically significant after additional adjustment although they became generally slightly smaller. SES effects were heterogeneous (p<0.10) for all outcomes except inhalant allergy and itchy rash (fig. 2⇓). Fully adjusted odds ratios were not less heterogeneous than minimum adjusted odds ratios.
The risk of wheeze during the past 12 months was found to increase with decreasing parental education in all Eastern European countries (Bulgaria, the Russian Federation, Poland, Hungary, Slovakia, and the Czech Republic) except (East) Germany along with Austria, but not in the remaining Western European countries of Italy, the Netherlands and Switzerland, and North America (fig. 2⇓). The associations between parental education and asthma and bronchitis are less systematic and less clear. Approximately half of the odds ratios for the comparison of low versus high parental education are smaller than one, whereas the other half is larger than one. Nocturnal dry cough is more prevalent among children with low parental education. With one exception each, prevalence of inhalant allergies and itchy rash were highest in children with high parental education and lowest in children with low parental education. Mean odds ratios were statistically significant for wheeze, nocturnal dry cough, inhalant allergy, and itchy rash (for itchy rash low versus high SES only). Mean odds ratios for hay fever and eczema (available for seven and five countries only, data not shown) were in line with mean odds ratios for inhalant allergy and itchy rash: there was a decreased prevalence of hay fever (mean OR (95% CI): 0.77 (0.49–1.23)) and eczema (mean OR (95% CI): 0.64 (0.53–0.77)) among children with low SES compared with high SES.
With some exceptions, the increase/decrease in risk was stronger for low parental education than for medium parental education. Odds ratios for the comparison of low versus high parental education varied between countries. Part of this variation arises from differences in prevalence of symptoms between countries. Furthermore, for wheeze, nocturnal cough and itchy rash the difference was negatively correlated with prevalence of low parental education (data not shown), i.e. the difference became larger when the definition of low parental education was stricter. For the other outcomes, no systematic pattern was seen.
Study area was used to control for the overall effect of (unmeasured) explanatory variables on the study area level. Since data on annual average exposure to PM10 calculated from monitoring sites in the study areas was available for all countries except Italy (table 3⇓), the current authors also calculated PM10 adjusted odds ratios and compared them with the study area adjusted odds ratios. The difference in odds ratios ranged from −14.7–7.5% (median −0.2%) and −18.9–12.7% (median 0.5%) for the comparison of medium versus high and low versus high parental education. Statistically significant associations between SES and symptoms remained statistically significant (data not shown).
DISCUSSION
Low parental education was associated with a decreased prevalence of inhalant allergy and itchy rash in school children aged 6–12 yrs living in Eastern and Western Europe and Northern America. Furthermore, low parental education was associated with an increased prevalence of wheeze and nocturnal dry cough; whereby the former was mainly restricted to children from Eastern Europe. No clear association was found between parental education and prevalence of asthma and bronchitis. Part of the differences between parental levels of education with regard to the children's health could be explained by differences in prevalence of parental allergy, crowding, mould/moisture in the child's home smoking during pregnancy, pet-ownership, and use of gas for cooking. However, differences remained after adjustment for these variables.
The association between parental education and inhalant allergy and itchy rash was consistent across Eastern and Western European countries and North America. The present authors think that this consistency is particularly noteworthy as the definitions and meanings of low and high SES as defined by parental education differ between countries. This indicates that social inequalities resulting in differences in health outcomes existed not only in Western European countries, but also in former socialist countries with a relatively uniform distribution of resources.
Selective under-reporting or access to healthcare might explain the decrease in prevalence of bronchitis with decreasing SES that was observed in some countries. Selective access to healthcare due to socio-economic reasons is unlikely for those countries whose systems are based on solidarity and equity, where entitlement and level of health insurance benefits do not depend on the amount of contribution paid, and where co-payments are reasonable 38. Severity of disease could also (in part) explain the higher prevalence of bronchitis in children with high parental education in other countries; highly educated parents might take medical advice in mild cases of bronchitis whereas parents with a low level of education take medical advice in more severe cases only. Unfortunately, data on severity of bronchitis were not available for most of the present studies. Moreover, people with low SES are more likely to live in more highly polluted areas and/or to be more susceptible to the health effects of air pollution because of compromised health due to material deprivation and psychosocial stress 22. The current authors assessed the importance of exposure to air pollution defined as annual average PM10 exposure on the study area level, but it did not explain the SES effect. However, using this approach did not take into account within-study area variation of air pollution, which might also play a role.
Several studies conducted in Western European countries reported higher prevalence of eczema, hay fever and sensitisation to inhalant allergens among children, adolescents and adults of the most advantaged socio-economic group 11, 12, 14, 15, 39–43. To the best knowledge of the current authors, the present study is the first study to show that the positive association between SES and itchy rash and inhalant allergies is not restricted to Western Europe, but also exists in Eastern European countries (besides large differences between countries with regard to the prevalence of symptoms). Again, it cannot be ruled out completely that selective under-reporting or access to healthcare might explain at least part of the decrease in prevalence of allergies with decreasing SES. Furthermore, lifestyle and environmental factors are likely to play a role. A number of hypotheses have been proposed as potential partial explanations for between-country and within-country disparities in prevalence of asthma and allergies including differences in “hygiene” (family size, day care, viral exposure, endotoxin, microbial exposure of the infants' large bowel, domestic allergen exposure), diet, cigarette smoking, traffic pollution, antenatal exposures and physical activity/obesity 23, but it is unclear which lifestyle or environmental factor(s) account for the association between SES and prevalence of allergies. The present authors adjusted for a set of explanatory variables, but the associations remained. Other factors potentially associated with SES, such as helminth infections, early contact with other children and dietary factors, might explain the observed effect, but no data on these factors have been available.
East–West differences in the associations between children's health and parental education were found, for wheeze only: wheeze was positively associated with parental education in all Eastern European countries except (East) Germany along with Austria, but not in the remaining Western European countries and North America. The current authors speculate that atopy might play a role: atopy has been shown to be more prevalent among children and adults with high SES 11, 14–16; and considerably higher prevalence of atopy was reported for West Germany compared to East Germany 44. Thus, wheeze is perhaps more likely to be linked to atopy in Western Europe and North America compared with Eastern Europe, and therefore associations between parental social class and wheeze might differ between Eastern and Western Europe. However, this does not explain the Austrian results.
The lack of association between childhood asthma and SES in the present study is consistent with the result of a recent study from New Zealand 45 and a review of 24 studies on the association between childhood asthma and SES by Mielck et al. 20: the studies did not reveal a clear picture; positive associations were as frequent as negative ones, and most studies showed no association at all. Recent evidence for an inverse gradient between SES and asthma comes from the European Community Respiratory Health Survey 46. When Mielck et al. 20 analysed data of >4,000 German fourth grade schoolchildren by severity of asthma, they found an increased prevalence of severe asthma (defined as constant wheeze with >10 acute attacks per yr) among children with low SES 20. This is in line with the studies reviewed by Gold and Wright 23 that showed a greater disparity in asthma morbidity than in asthma prevalence for impoverished African Americans. However, a differentiation between mild and severe asthma was not possible for the present study. Moreover, associations between education and asthma may be different for atopic and nonatopic asthma 13, 46.
The present authors considered the possibility of a U-shaped association between SES and children's respiratory health with the very poor having more respiratory symptoms due to much higher exposures, and the rich reporting more symptoms due to better access to healthcare and better diagnoses. However, there was no indication of medium-SES children having less symptoms than low- and high-SES children in the PATY study. One potential explanation might be the current definition of SES resulting in sometimes rather large groups of children with low and high SES, respectively. It cannot be ruled out that a more extreme definition of low and high SES reveals a different pattern. However, given the information that is available from the original cross-sectional studies, no alternative definition is possible.
SES was defined by parental education, being aware of the fact that a person's SES is not only defined by his/her educational level but also, for example, by his/her occupation and income. However, parental education was available from all studies, only two studies asked for income and three studies for parental occupation. Moreover, the consistency of the association between parental education and crowding strengthens the current authors' confidence that parental education reflects the children's SES.
The maximum educational level achieved by either parent was used to assess the child's SES, assuming that the parent with the highest educational level is the determining factor for a family's SES. One might argue that maternal SES might be more relevant with regard to a child's health. However, the overall concordance between maternal and paternal education was fair to moderate (58–80%), and the association between maternal education and children's health was rather similar to the association between maximum parental education and children's health (data not shown).
Several studies of social conditions related to health inequalities, healthcare utilitisation and SES have shown that metropolitan area variation is greater than between-country variation 47, 48. Thus, it is possible that the level of aggregation (the country) may obscure the association between SES and the children's health. The current authors decided against an analysis on the study area level since the term “study area” has different meanings in the different studies: it represents school catchments in some of the studies and geographic areas (of different sizes) in others. Therefore, it cannot be ruled out that the results might have been different if the analysis had been performed on the community scale.
CONCLUSION
Children's respiratory and allergic symptoms depended on parental education in 13 diverse countries. The association could not be explained by established risk factors.
Distribution of explanatory variables by parental education: a) parental allergy; b) maternal smoking during pregnancy; c) breastfeeding; d) pet-ownership ever; e) crowding; f) mould/moisture in the child's home; g) use of gas for cooking) related to parental education (□: low; ░: medium; ▒: high) by country. BUL: Bulgaria; RUS: the Russian Federation; POL: Poland; HUN: Hungary; SLO: Slovakia; CZ: Czech Republic; I: Italy; NL: the Netherlands; GER: Germany; A: Austria; CAN: Canada; CH: Switzerland; USA: United States of America.
Multiple logistic regression results expressed as country-specific and mean odds ratios (95% confidence intervals) for the association between parental education and respiratory and allergic symptoms and diseases a) wheeze past 12 months; b) asthma ever; c) nocturnal dry cough; d) bronchitis past 12 months; e) inhalant allergy; f) itchy rash) in children adjusted for age, sex, study area parental allergy, smoking during pregnancy, breastfeeding, pet-ownership, crowding, mould/moisture in the child's home and use of gas for cooking. ○: medium versus high parental education; •: low versus high parental education. BUL: Bulgaria; RUS: the Russian Federation; POL: Poland; HUN: Hungary; SLO: Slovakia; CZ: Czech Republic; I: Italy; NL: The Netherlands; GER: Germany; A: Austria; CAN: Canada; CH: Switzerland; USA: United States of America. #: indicates that test for homogeneity p-value <0.10.
Description of the studies(ordered by per capita gross domestic product)
Definition of parental education
Description of the study population
Prevalence of symptoms and diseases
APPENDIX 1: DEFINITION OF HEALTH OUTCOMES⇑
Acknowledgments
The authors wish to thank B. Armstrong and A. Mielck for their advice, and V. Raghuyamshi Singh for his collaboration. Furthermore, the authors are indebted to all scientists and technical staff involved in the planning and conduction of the original cross-sectional studies.
The PATY study is funded by the EU 5th Framework Quality of Life Program (proposal no. QLRT-2001-02544). U. Gehring was supported by a post-doctoral fellowship of the German Academic Exchange Service (DAAD).
- Received February 15, 2005.
- Accepted August 25, 2005.
- © ERS Journals Ltd
References
