A systematic review of socioeconomic position in relation to asthma and allergic diseases

Spatial patterns in the global burden of disease

Prevalence rates of asthma and allergies have been on the rise in recent decades across the globe. The International Study of Asthma and Allergies in Childhood (ISAAC) phase III study conducted between 2001 and 2003 reported that the global prevalence rate of wheezing was 11.6% for 6–7-year-old children, and 13.7% for 13–14-year old children [3]. In Western countries, the prevalence rates of asthma, allergic diseases and associated symptoms are generally higher than in other countries. Pearce et al. [4] concluded that the prevalence rate of asthma symptoms is particularly high in English-speaking countries, Western Europe and some of the more affluent Asian countries. The prevalence rate of ever having had asthma among 13–14-year old children was 16.3% in Western Europe, 22.5% in North America, and 32.4% in New Zealand and Australia, while the global average was 13.8%. Apart from these high-income countries, Latin America is characterised by high prevalence rates. Other low- and middle-income countries, such as most of the African continent, Asia and Eastern Europe, have the lowest reported prevalence rates. Ever having had asthma or having a current wheeze was reported by 5.2% of 13–14-year-olds in Africa, 4.0% in the Asia-Pacific region and 3.1% on the Indian subcontinent, compared with the global prevalence rate of 6.2% [3]. These spatial patterns suggest a causal role for the environment, perhaps mediated through exposure to allergens, hygiene, air pollution and nutrition [5]. More recently, however, time trends have started to emerge, challenging the main theories on asthma causation.

Time trends in the global burden of disease

The global burden of asthma and allergic diseases is still increasing, at a time when the differences in the prevalence rates of asthma symptoms between countries seem to be decreasing [3]. This observation can partly be attributed to the rising prevalence rates of reported asthma and allergic diseases in many low- and middle-income countries. From 1994–1995 to 2001–2003, the upward trend in the prevalence rates of asthma and allergies has been particularly strong in Latin America, Northern and Eastern Europe, Africa, North America, and the Eastern Mediterranean [3]. In some high-income countries, however, and in particular those which are English speaking, the prevalence rates of reported asthma and allergies have now been reported to fall or at least plateau [2, 3, 6–8]. At this time, study findings contradict each other and not all researchers are convinced that the plateaus or decreases in prevalence rates observed in some studies indicate a trend [9–11]. Part of the change may be related to better awareness, improved diagnostic methods and changing diagnostic criteria that have uncovered undiagnosed cases in previous decades. However, this does not explain a decrease of prevalence rates in some populations. Even a stabilisation of new cases in Western countries challenges ideas about the causal role of factors such as obesity, hygienic environments and air pollution [3].

Socioeconomic position in relation to asthma and allergies

Spatial patterns in combination with changing trends over time may provide clues to understanding the causal pathways involved in the development of asthma and allergic diseases [12]. In relation to socioeconomic position (SEP), the “hygiene hypothesis” is the theory that features most prominently in the explanation of higher prevalence rates in more affluent Western countries: maternal exposure to microorganisms during pregnancy and early childhood exposure, which are thought to be higher in less “hygienic” and poorer environments such as farms and more crowded houses, would protect against the sensitisation to allergens later in life [13]. Despite being a popular theory, the hygiene hypothesis has been widely criticized, and results for various types of exposure and in different populations contradict each other [14, 15]. In addition, the evidence on the relationship between SEP and asthma and allergies has been conflicting, with both low and high SEP being reported as a risk factor.

A nonsystematic review published in 1996 summarised results from Australia, the USA, the UK and New Zealand, and found no consistent association between SEP and childhood asthma [16]. Although many factors associated with SEP, such as environmental tobacco smoke, housing, air pollution and obesity, have been studied in relation to asthma and allergic diseases, there has been no worldwide systematic review of the evidence of associations between SEP and asthma and allergies. We aim to review the evidence on the associations between SEP and the prevalence rates of asthma and allergic diseases by presenting an up-to-date overview of the international scientific literature.

Overview of studies

Many studies were based on data from the ISAAC phase 1, 2 or 3 surveys, and others either adapted the ISAAC questionnaire for their own use or applied it without changes. The map in figure 3 shows the countries represented in these surveys. Excluded from the map are two studies using subsamples from international surveys without specifying which countries were included [20, 21]. The map shows that the Americas and Europe were well represented, while African countries were poorly represented. Ecuador and Brazil were overrepresented in Latin America, and data from the Middle East were disproportionally collected from Turkey. For some countries, the large number of studies is explained by national or regional birth cohorts, such as the Pelotas birth cohort study in Brazil, from which seven out of 17 studies drew data. In Turkey, however, all six studies were based on cross-sectional data collected from different parts of the country. A total of 32 countries not represented in the ISAAC surveys were covered by other surveys or cross-sectional studies.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3

Setting of research conducted on asthma/allergies and socioeconomic position. ISAAC: International Study of Asthma and Allergies in Childhood.

The majority of included studies applied a cross-sectional design (n=124), followed by 42 prospective longitudinal studies, eight case–control studies, five retrospective longitudinal studies and four time series designs. Table 2 presents an overview of study characteristics. Studies included in this review were largely published after 2000, which partly reflects the growing interest in social determinants of health and partly the limited availability of older papers. This synthesis of results shows a preference for parental education and income as measures of SEP. The main source of information on SEP tended to be self-reported or parent-reported data, with census data and medical records also being used to extract this information. The measurement of health outcomes varied from parent-reported asthma and allergies to doctor-diagnosed medical illnesses extracted from medical records. Allergic disease indicated by tests for allergen-specific IgE generally lead to higher prevalence rates than medically diagnosed or self-reported allergies. In the study by Hoffmann et al. [22], for example, the parent-reported prevalence of ever having had allergies was 12.5%, while tests showed allergic sensitisation in 58.4% of cases.

Quality of studies

Detailed information about the quality of each study can be found in the online supplementary material. 24 studies were classified as poor quality, 41 were considered good quality and the remaining 118 were rated as average quality. The transparency of publications was affected by main aspects of the study design and execution not being reported. Most common was the lack of a discussion of the generalisability of study results (n=9) and a statement of funding received (n=43). Furthermore, 42 studies did not adjust for any confounders and 12 studies did not describe clearly the rationale for certain measures used. Two studies, for example, did not specify how their SEP variables were defined or measured [23, 24]. Three studies used the UK system of classification of occupations but did not make this explicit [25–27]. In addition, 18 studies were underpowered for the analyses performed, resulting in very wide confidence intervals for the risk estimates.

Association between asthma and SEP

A positive relationship between SEP and the prevalence rate of asthma or wheeze was found in only 9% of the studies, whereas 63% reported lower SEP to be associated with higher odds of asthma. The association between lower SEP and a higher prevalence of asthma was found in studies with children and adults, and across world regions. The pooled unadjusted odds ratio for the prevalence of asthma for people of the lowest SEP compared with the highest SEP was 1.38 (95% CI 1.37–1.39) and the odds ratio adjusted for a wide range of covariates was 1.11 (95% CI 1.09–1.14) (online supplementary material). Some studies found that the relationship between asthma and SEP varied by measure of SEP. In a study among US children, for example, lower prevalence rates of asthma were associated with low family income (OR 0.33, 95% CI 0.20–0.54) but not with unemployment (OR 1.35, 95% CI 0.84–2.17) [28]. A study among children in Cuba found that those exposed to an economic crisis had lower odds of asthma than unexposed children (OR 0.40, 95% CI 0.17–0.95) [29]. The association between self-reported or doctor-diagnosed asthma and SEP was more evident than the association between symptoms of wheeze and SEP (unadjusted OR 1.02, 95% CI 0.99–1.06; adjusted OR 1.33, 95% CI 1.29–1.37) (online supplementary material).

Association between allergic diseases and SEP

In contrast to asthma, of the studies that used allergies as an outcome measure, 66% found higher individual SEP to be associated with a higher prevalence rate and 12% reported that lower SEP was associated with higher rates of allergies (table 3). The pooled unadjusted odds ratio for the prevalence of allergies or atopy in general, for people of the lowest SEP compared with the highest SEP, was 0.67 (95% CI 0.62–0.72), and the odds ratio adjusted for a wide range of covariates was 0.69 (95% CI 0.60–0.79) (online supplementary material). A lower SEP was also associated with a lower prevalence rate of atopic dermatitis (unadjusted OR 0.72, 95% CI 0.61–0.83; adjusted OR 0.79, 95% CI 0.73–0.85) and allergic rhinoconjunctivitis (unadjusted OR 0.52, 95% CI 0.46–0.59; adjusted OR 0.79, 95% CI 0.78–0.80) (online supplementary material). Reported associations between higher prevalence rates of allergic diseases and lower SES were not consistent, and varied according to the measures of SEP or the health outcomes used in the analysis.

Area-level SEP in relation to asthma and allergic diseases

A total of 19 studies reported on the association between area-level measures of SEP and either individual- or area-level health outcomes. 11 of these, most of which used measures of area deprivation or average household income, found that lower area SEP was associated with hospital admissions for asthma (n=6) [30–35], admission for anaphylactic shock [36] and the prevalence of asthma/wheeze (n=4) [37–40]. Only one study of the associations between asthma and area-level income found no significant result [41], and a study among UK adults showed an association between higher area deprivation and asthma for women but not for men [42]. Two studies using data on gross national income concluded that more affluent countries had higher prevalence rates of asthma, wheeze and atopic sensitisation [43, 44]. An ecological study making use of the ISAAC study data found a weak linear association between gross national product and wheeze and eczema, although this association was most prominent in non-European countries [45]. The remaining three studies reported mixed results [46–48].

Sensitivity analysis with high-quality studies

Including only high-quality studies confirms the relationships found between lower SEP and higher prevalence rates of asthma, and reproduces the evidence of the association between higher SEP and allergic diseases. 36 studies rated as good quality included asthma or wheezing as a health outcome (online supplementary material). Of these studies, 17 suggested asthma or wheezing were associated with lower SEP and only two studies found the opposite association. The remaining 17 studies either did not study this particular association (n=6) or reported mixed results depending on measures of health and subgroups within the population under study (n=11).

Aligne et al. [49] studied a sample of black and white Americans, and found that the prevalence of asthma was not related to poverty or ethnicity but to living in an urban setting. Another US study concluded that higher odds of asthma were associated with a lower SEP for African-American children and a higher SEP for Mexican-American children, while results for the “other Latino” group were not significant [50].

Eight studies on the relationship between individual SEP and allergies were rated high quality. Together, these studies reported on 11 relationships between SEP and different allergies, of which nine results in seven studies pointed towards a higher prevalence for people of higher SEP [51–57]. One study found no relationship between SEP and doctor-diagnosed atopic eczema in Japanese children [58].

Study limitations

The findings of this systematic review are likely to be affected by bias within studies as well as bias in the pooled estimates of studies. Some studies did not provide a rationale for their choice of measures, and it is likely that measures of SEP chosen for reasons of convenience and availability may not always be the most suitable measures to detect social inequalities in health. Most studies used income or education of the parents as a proxy for SEP (table 2). Although these are often used and therefore facilitate comparison between studies, education of the parents in particular does not necessarily reflect the SEP of children in the household. In studies focussing on asthma or wheezing in children, measures of parental SEP may account for only part of the effects of social disadvantage on child health. This life-course perspective on childhood asthma was rarely considered in any of the studies. Due to the suboptimal quality of many studies, we analysed the sample rated as high quality separately. This did not lead to different conclusions on the relationships between SEP and asthma and allergies.

The generalisability of our findings may be affected by the study settings in which the data were collected. The results draw heavily upon data from Europe and the Americas, and the regions of Africa and Oceania were included in only 20% of the articles (table 2). The majority of African countries did not participate in the ISAAC surveys upon which many studies relied for their data collection (fig. 3).

Despite the large number of studies reviewed, by limiting the literature search to one electronic database, we might have not included all relevant literature. Furthermore, funnel plots indicate that not all relevant studies might have been published. There was a lack of smaller studies reporting results that contradict the main findings; in other words, a higher SEP being associated with higher prevalence rates of asthma or a lower SEP being associated with lower prevalence rates of allergies (online supplementary material).

We calculated pooled measures of association to aid the interpretation of the vast amount of data collected from the studies included in this review, despite large differences between studies in measurement of SEP, and health, setting, focus and statistical analyses. Higgins' I²-tests indicated considerable heterogeneity in studies of all groups of health outcomes [59] (online supplementary material). We stress that this systematic review is no more than a summary of previous findings and ask that results are interpreted with caution. Although we are confident that the pooled estimates give a reliable indication of the nature of the associations between SEP and allergic diseases for the studies included in this review, this information by no means quantifies the relationship between SEP and allergic diseases in the general population.

Implications for research and policy

The evidence for an association between lower SEP and a higher prevalence rate of asthma suggests that social and economic determinants of health associated with deprivation may play a role in the aetiology of asthma. Examples of such determinants from recent research include obesity, tobacco smoking, and indoor and outdoor pollution [14, 60, 61]. People with a lower SEP may be more exposed to these factors than those with a higher SEP. However, this does not explain why the prevalence of asthma is generally lower in low-income countries, where especially residents in urban areas in tropical climates may be more exposed to indoor and outdoor pollution than in higher income countries. Research on migrants has shown that the prevalence of asthma in people migrating to high-income countries rises over time, with increased length of exposure to determinants of health in the host country [62]. More research is needed to determine which environmental factors in westernised societies drive this change in susceptibility to asthma.

Lower SEP is associated with poorer health for almost every health outcome, and our results indicate that the association between higher SEP and allergies is a notable exception. For other health outcomes, the inverse social gradient can usually be explained by demographic characteristics and health behaviour (breast cancer and skin cancer). Higher prevalence rates of allergies in higher social groups have been considered consistent with the hygiene hypothesis, which has also been linked to the lack of a social gradient or inverse social gradient for other immune disorders (e.g. type 1 diabetes) [63, 64]. In contrast to evidence supporting the hygiene hypothesis, a Korean study found higher rates of sensitisation to most allergies in children with a higher household income, while sensitisation to cockroach allergen was more common in children of lower SEP [65]. Evidence on cockroach allergy not included in this review confirms that higher exposure to cockroach allergens in people with a lower SEP is associated with higher, rather than lower, prevalence rates of allergy [66, 67]. We cannot draw final conclusions on the hygiene hypothesis in relation to allergies, as most studies included in this review did not measure exposure to allergens in addition to the diagnosis of allergy.

Another explanation for an inverse social gradient in allergies is that people of higher SEP would be more likely to report allergy symptoms. One study found that the prevalence of self-reported allergies was significantly higher in people of higher SEP, while laboratory results on allergic sensitisation did not reach statistical significance [22]. Two studies showed a significant inverse social gradient in parent-reported as well as clinically diagnosed atopic dermatitis, with symptoms of illness being more common among children of higher SEP [68, 69].

Although we do not provide causal evidence, this review suggests it could be useful to treat asthma and different types of allergy as separate health outcomes, particularly when investigating the social determinants of health. In addition to a well-considered health outcome, using a combination of SEP measures increases the likelihood that the effects of SEP are captured accurately for children and adults, men and women, and different ethnic groups. This nuanced approach to the examination of determinants of asthma and allergies is likely to improve our understanding of causal pathways, in order to inform clinical practice and evidence-based policy on the prevention of asthma and allergies.

Conclusion

The evidence from this systematic review suggests that the association with SEP differs for asthma and allergic diseases. Asthma is associated with lower SEP, whereas allergies are more often associated with higher SEP.