RESULTS

Section:

ChooseTop of pageAbstractMETHODSRESULTS <<DISCUSSIONReferencesCITING ARTICLES

A total of 54,439 children, ages 8 to 12 years old, had completed parental questionnaires, 31,759 underwent skin prick tests, and 8,951 participated in the measurement of IgE antibodies. Participation rates ranged from 36 to 100% for the questionnaire, from 24 to 99% for skin prick tests, and from 12.5 to 76% for IgE measurements (Table 1).

Association between Asthma Symptoms and Atopic Sensitization

The association between current wheeze and skin prick test reactivity was statistically significant in all centers in affluent countries, except for Reykjavik, Iceland, where the association did not reach formal statistical significance (Figure 1). The combined odds ratio (OR) for the affluent countries was 4.0 (95% confidence interval [CI], 3.5–4.6). The observed associations were substantially weaker in nonaffluent countries (combined OR, 2.2; 95% CI, 1.5–3.3). Guangzhou, China, was not included in the calculation of this estimate because the OR for this area was an extreme outlier (OR, 58.9) due to the virtual absence of nonatopic wheeze in the sample (Table 2). There was no significant evidence for heterogeneity among centers in affluent countries estimated from random effects meta-analysis (P = 0.21). There was a suggestion of heterogeneity among centers in nonaffluent countries only when the OR of Guangzhou was included in the meta-analysis (P = 0.052). A similar pattern emerged for elevated allergen-specific IgE levels where estimates for the combined OR were 3.5 (95% CI, 2.9–4.2) for centers in affluent countries and 1.9 (95% CI, 1.0–3.9) for centers in nonaffluent countries. Again, there was no apparent heterogeneity (P = 0.26 for affluent and P = 0.43 for nonaffluent countries).

Figure 1. Odds ratios (OR) with confidence intervals for the association of current wheeze with skin prick test reactivity (a) and elevated (⩾0.35 allergen-specific kilounits per liter [kU_A/L]) allergen-specific IgE (b) by study center. Centers are listed in alphabetical order (country, center name).

Download Figure | Download Powerpoint

[More] [Minimize]

Correlations at the Center Level

The prevalence of current wheeze was not correlated with the prevalence of skin prick test reactivity or elevated allergen-specific IgE (Figure 2). Prevalence rates of current wheeze (ρ = 0.39, P = 0.05) and skin test reactivity (ρ = 0.37, P = 0.05) were only weakly correlated with GNI (Figure 2). The prevalence of atopic wheeze (i.e., current wheeze in combination with a positive skin prick test) was strongly associated with GNI (ρ = 0.60, P = 0.002); rates of nonatopic wheeze were not (ρ = −0.02, P = 0.92; not shown in Figure 2). The PAFs, based on skin prick test reactivity and elevated IgE antibody levels, were both significantly correlated with GNI (Figure 3). The correlations between PAPs and GNI were even stronger (ρ = 0.74, P ≤ 0.0001 for skin test reactivity, and ρ = 0.74, P = 0.003, for IgE antibodies). Finally, the strength of the association between skin prick test reactivity and current wheeze (assessed by OR) also increased significantly with GNI. There was a suggestion for a similar correlation with IgE, but this correlation did not reach statistical significance.

Figure 2. (a–f) Correlations at center level of prevalence rates of current wheeze, skin prick test reactivity, elevated allergen-specific IgE (⩾0.35 kU_A/L), and gross national income per capita (GNI). ^#Spearman rank order correlation coefficient; ⁺participation > 60% for the written questionnaires.

Download Figure | Download Powerpoint

[More] [Minimize]

Figure 3. Attributable fractions (%), attributable prevalences (%), and odds ratios for the relation of current wheeze with skin prick test reactivity (a, c, e) and allergen-specific IgE (⩾0.35 kU_A/L) (b, d, f) by gross national income per capita (GNI). ^#Spearman rank order correlation coefficient; ⁺participation > 60% for the written questionnaires.

Download Figure | Download Powerpoint

[More] [Minimize]

DISCUSSION

Section:

ChooseTop of pageAbstractMETHODSRESULTSDISCUSSION <<ReferencesCITING ARTICLES

ISAAC II is the first large international study on children that can rely on methodologically comparable prevalence rates of skin prick test reactivity and elevated allergen-specific IgE levels in addition to questionnaire-based reports of asthma symptoms. We observed large variations in the prevalence of asthma symptoms and of atopic sensitization among populations. An intriguing finding is that the link between atopic sensitization and asthma symptoms increased with economic development.

Our findings confirm the wide international variation in the prevalence and severity of asthma symptoms in children observed during ISAAC I (9, 10). Study centers that had ranked high (e.g., New Zealand, United Kingdom) or low (e.g., Albania, India) showed similar rankings in ISAAC II. The high prevalence of wheeze in the Brazilian center corroborates previous ISAAC observations of high asthma rates in urban centers in Latin America (9, 21). The ISAAC II data show that this high prevalence is mostly due to nonatopic wheeze, supporting similar findings from other areas in South America (22, 23).

The large variation in prevalence rates of current wheeze could not be explained by prevalence rates of skin test reactivity or elevated allergen-specific IgE levels as shown by the lack of correlation (Figures 2a and 2b). The prevalence of wheeze during the last year, however, increased with GNI, confirming previous analyses of the ISAAC I data (24). We also observed a significant correlation between the prevalence of skin test reactivity in children and GNI, a finding that has not been reported previously. The correlation between GNI and the prevalence of elevated allergen-specific IgE levels did not reach statistical significance, possibly due to the smaller number of centers. When restricting the skin test reactivity analysis to centers with data on IgE, there was also no correlation (ρ = 0.29, P = 0.25). The determinants of the large international variation in the prevalence of asthma symptoms and atopic sensitization are likely to be numerous and complex (25–27). A wide range of different factors, including nutrition, microbial and allergen exposure, housing conditions, exposure to pollutants, and so forth, may have played a role (28). Finally, it should to be kept in mind that we report an ecological correlation. A center-level correlation with GNI does not imply a similar relation at the individual level with personal wealth.

There was a wide variation in the prevalence of atopic sensitization across study centers. The variation was more pronounced for skin test reactivity than for elevated IgE antibodies, which could be partly due to the smaller number of centers with IgE data. Several study centers had tested additional allergens of local relevance, thereby enhancing the validity of findings on skin test reactivity. Inclusion of these additional allergens in the analyses, however, had only small effects on the parameters under investigation and mostly affected prevalence estimates (e.g., among centers that used additional allergens, the range of prevalence rates increased from 7.1 to 42.8% [using the six standard allergens] to 10.3 to 45.3%). For the same centers, the combined ORs for the association between wheezing and skin prick test reactivity changed from OR = 3.95 to OR = 4.06 in affluent countries and from OR = 2.62 to OR = 2.82 in nonaffluent countries after inclusion of the additional allergens. The correlation between GNI and ORs for the association between current wheeze and skin test reactivity became enhanced when additional allergens were included: ρ = 0.68, P = 0.003 without, and ρ = 0.77, P = 0.0003 with additional allergens. The ISAAC II findings are in line with previous reports of lower prevalence rates of skin test reactivity in less affluent countries (3, 4, 7, 22, 29–32). Our data also show large variations in the PAFs and attributable prevalence rates, particularly between centers in affluent and nonaffluent countries. The size of the observed PAFs corroborates and extends the findings by Pearce and colleagues for children (PAFs from 25 to 63%) (3) and by Sunyer and coworkers for adults (PAFs from 4 to 61%) (8).

We observed a strong correlation of the link between wheeze and atopic sensitization (measured by OR, PAF, and PAP) and GNI of the respective countries. Before discussion of potential interpretations, methodologic aspects need to be addressed. First, participation rates were low in some centers (e.g., <60% for questionnaires and <40% for skin prick or IgE measurements), raising concern about selective participation. In general, allergic symptoms (rhinoconjunctivitis and eczema) and parental allergies were slightly more often reported among participants with measurements of skin test reactivity or serum IgE than among children with completed questionnaires. For skin prick tests, statistically significant differences were only seen in one center (Tbilisi). For IgE measurements, significant differences (i.e., higher rates among those with IgE data) were observed for hay fever symptoms in Cartagena and Tbilisi, and for parental allergies in West Sussex and Tbilisi. However, although potential selection of children may have had some effect on prevalence estimates, it is less likely to introduce systematic bias for the assessment of associations within study centers (e.g., calculation of OR). When centers with low participation were excluded (i.e., <60% for questionnaires and <40% for skin prick or IgE measurements), most associations became substantially stronger (for skin prick test: PAF, ρ = 0.81, P < 0.0001; PAP, ρ = 0.82, P < 0.0001; OR, ρ = 0.76, P < 0.0001; and for allergen-specific IgE: PAF, ρ = 0.79, P < 0.0001; PAP, ρ = 0.83, P = 0.0008; OR, ρ = 0.33, P = 0.29). Exclusion of Tbilisi from our analyses had very little effect on the observed pattern of our findings. Furthermore, it could be speculated that the observed relations are due to differences in phenotype. Similar relations, however, were seen for reports of asthma-ever and frequent wheezing attacks during the past year (data not shown). In addition, when current wheeze was classified depending on whether it occurred in combination with symptoms of allergic rhinoconjunctivitis (i.e., therefore more likely to be related to atopic sensitization) or not, the relationship with GNI was enhanced for wheeze with rhinoconjunctivitis symptoms (e.g., correlation based on the OR: ρ = 0.58, P = 0.002 for prick test reactivity, and ρ = 0.67, P = 0.002 for allergen-specific IgE) and not present for wheeze without symptoms of allergic rhinitis (data not shown). Finally, GNI at the country level is not a precise measure of economic development in the study areas and the prevalence estimates from the study centers are not representative of the country as a whole. Thus, some misclassification cannot be ruled out. However, economic conditions and prevalence rates of asthma symptoms differ much more between countries than between regions within countries (9, 10). In addition, to explain the observed relations, misclassification would have to be differential—that is, dependent on the association between asthma symptoms and sensitization, which appears to be unlikely.

A possible explanation for an increasing link between asthma and atopic sensitization with economic development could be that factors that protect children with atopic sensitization against the development of manifest disease get attenuated or lost in more affluent settings. Differences in the association of wheeze with atopy between populations have been found before, with weaker associations seen mainly in rural or semirural nonaffluent locations (4, 6, 33). In these cases, it is difficult to tease apart the effects of rural/urban versus affluent/nonaffluent, because rural is generally equivalent to less affluent. However, a lack of association was also observed in a deprived but urban environment in Lima, Peru (22). Others reported attenuated associations of circulating IgE antibodies with positive skin prick tests and allergic disease in former socialist countries (34, 35). In our data, ORs for the relation of wheeze with skin prick test reactivity tended to increase with the prevalence of skin test reactivity (ρ = 0.31, P = 0.10), not with the prevalence of elevated allergen-specific IgE (ρ = 0.07, P = 0.79), possibly indicating that the relevant factors may affect both the development of skin test reactivity and its link with clinical asthma symptoms.

Few studies have specifically investigated the factors that may influence the relation between allergen-specific IgE, skin prick test reactivity, and manifest atopic disease, and most of those relate to helminth infections (36, 37). There is evidence that increased production of IL-10 concomitant with helminth infections down-regulates allergic inflammatory responses (37). However, a cluster-randomized trial of geohelminth treatment in an endemic area showed no increase in atopic sensitization 1 year after the start of treatment (38). Other conditions that may modify the relation between atopic sensitization and asthma symptoms include overweight and obesity (39). A study in urban and rural South Africa observed that increasing body mass index was associated with a greater strength of association between allergen-specific IgE and the corresponding skin test (40).

It has also been hypothesized that commensal bacteria acquired very early in life play an important role in the induction of tolerance (41). Less or an altered microbial exposure (e.g., an altered gut flora) could therefore weaken normal immune regulatory function and tolerance. Differences in the gut flora have been reported between children with and without allergic disease (41) and between countries with different prevalence rates of allergic disease (41) (e.g., between Estonia and Sweden) (41, 42). Within affluent countries, special subgroups of children—for example, those living on farms or having an anthroposophic lifestyle—have been found to be at lower risk for asthma and allergies (43–45). The degree of urbanization is one factor that increases with higher GNI. In our study, using national data provided by the Food and Agriculture Organization of the United Nations (http://faostat.fao.org), urbanization was in fact correlated with the prevalence of wheeze attributable to skin test reactivity. However, after adjustment for economic development, only GNI remained significant in the model (see also the online supplement).

Our findings show that, at age 8 to 12 years, nonatopic wheeze represents a frequent form of wheeze. In many countries, it comprised the majority of cases, particularly in centers in nonaffluent countries. Studies from affluent settings described a phenotype of asthma that is linked to respiratory infections in early life and outgrown during later childhood (46, 47). The relatively high prevalence of nonatopic wheeze in some of our study centers may be related to increased susceptibility and/or high exposure to infections. The age at which the wheezing episodes are outgrown may also differ among countries, particularly between affluent and nonaffluent settings. The substantial proportion of nonatopic wheeze in many populations warrants further investigation of the determining factors. Several studies have shown that the patterns of risk factors for atopic versus nonatopic wheeze may differ (44, 48–50).

In conclusion, our study shows that prevalence rates of asthma symptoms and of atopic sensitization vary widely in 8- to 12-year-old children living under very different conditions worldwide. The fractions and prevalences of wheeze that were attributable to atopic sensitization, as well as the strength of the associations between wheeze and atopic sensitization, increased with economic development.