Magnus Wickman Department of Occupational and Environmental Health Norrbacka level 3 Karolinska Hospital SE-171 76 Stockholm Sweden
Background: It is well established that early diagnosis of allergic disease is warranted.
Methods: In a prospective birth cohort study (BAMSE) 3743 children at 4 years of age were included. Children were classified as having any allergic disease, e.g. asthma, suspected allergic rhinitis (suspAR), eczema or oro-gastro-intestinal symptoms with questionnaire. Blood was obtained from 2612 of these children and analysed for IgE antibodies (ab) towards 14 common food and airborne allergens.
Results: Positive IgE ab results were found in 38% of the children with any allergic disease, whereas such IgE ab results were found in 17% among those without any allergic disease. Furthermore, among children with any allergic disease the median summated IgE ab levels were 10.7 kUA/l compared with 1.5 kUA/l among those without such symptoms. The highest IgE ab levels were found to birch, peanut, cat and horse. When the sum of the IgE-ab levels towards the selected allergens was at least 34 kUA/l, or, alternatively, more than four allergen tests were positive, there was a 75% likelihood of identifying the individual with any allergic disease. To identify those with asthma, as well as those with suspAR, a significant interaction was found for the combination of the sum of IgE-ab levels and number of allergens positive at test. For eczema only, the number of positive allergens at test was associated to the likelihood of such disease.
Conclusions: In children, 4 years of age, allergic disease was frequently not associated with the presence of single positive IgE antibody results, whereas increased IgE ab levels were significantly more prevalent among those with allergic disease. Thus, testing a certain profile of airborne and food allergens, and utilizing the sum of the IgE-ab levels in combination with the number of allergens positive at tests, may represent a more efficient diagnostic tool then to use just single positive IgE-ab results.
For a quarter of a century the prevalence of allergic diseases has increased among children (1, 2). In Europe one out of four children suffer from any allergic disease and questions have been raised how this may develop in the future (3).
A number of studies have demonstrated that sensitization to food and airborne allergens may predict allergic airway disease, and young nonsensitized children with wheeze seem to go into remission more often than recurrent wheezers who are sensitized (4–8). Besides, and independent of age, sensitization to airborne allergens in asthmatic subjects is a risk factor for increased severity of disease (9–11). Consequently, there is a need for early diagnosis of allergic children and identification of offending allergens. This is a prerequisite for secondary preventive measures, e.g. specific treatment, allergen avoidance and family education to prevent exacerbation of disease (12).
There is a growing interest in the use of quantitative analysis of IgE antibody (ab) responses for allergy diagnostic purposes among children (13–15). We have recently described quantification of IgE-ab with mixes of inhalant and food allergens, Phadiatop® and fx5®, respectively (16). This was done in conjunction with the follow-up at 4 years of age in an ongoing large birth cohort study. We demonstrated that quantification of IgE-ab with such multi-allergen tests could facilitate identification of allergic disease, and could also give indications of asthma severity.
The aim of the present study was to investigate whether the sum of allergen-specific IgE-ab levels taken together, or the number of allergens with positive results, or a combination of these two ways to express sensitization would add information that could help to identify children with allergic disease in the same birth cohort of 4-year-old children.
A birth cohort study (BAMSE – Children, Allergy, Milieu, Stockholm, Epidemiological survey) of 4089 children was established during 1994–96, in Stockholm, Sweden. Study design, enrolment, criteria for inclusion, procedures for data collection, and definitions of exposure and outcome have been described recently (17–19). In brief, data on allergic heredity and various exposures were obtained by parental questionnaires when the children were 2–3 months of age. At child's age 1 and 2 years the parents answered questionnaires on symptoms of allergic disease, feeding and other key exposures. When the children were four the first degree relatives answered a symptom questionnaire and all those 3743 (92%) children were invited to a clinical examination. Of these, 2965 (73%) accepted and arrived for the examination. Permission for the study was obtained from the Ethics Committee of Karolinska Institutet. Blood samples were successfully taken from 2614 children of the 2965 (88%). One hundred and thirty-nine refused to give blood samples and in another 216 sampling failed or there was too little serum to be analysed.
The questions utilized in the forms were taken either from the ISAAC-questionnaire or from questionnaires used in studies previously published by our group (20–22). The questionnaires done at 4 years of age focused on symptoms of asthma, rhinitis, rhino-conjunctivitis and eczema. Reactions to food were also enquired about, but since such reactions may include skin and airway reactions, only symptoms from the oro-gastrointestinal tract were taken into consideration, to avoid misclassification with other allergic diseases.
All outcomes asked for dealt with the preceding 12 months only. Asthma in this study was defined as at least four episodes of wheezing (ISAAC) or one episode of wheezing in children being treated with inhaled glucocortico steroids, as this treatment could mask repeated episodes. Suspected allergic rhinitis (suspAR) was defined according to the ISAAC-questionnaire as runny, itchy or blocked nose without ongoing cold, and eczema as itchy rash for at least 2 weeks with typical distribution (face/distal limbs/folds of elbows or behind the knees/wrists or fronts of ankles) and dry skin (23). Oro-gastro-intestinal symptoms (OGIS) was defined as lip or oral oedema/itch, vomiting and diarrhoea at least once in relation to ingestion of food and was found. Symptoms of lip oedema also included oedema around the eyes as it was not possible to disentangle the two types of oedema from each other in the questionnaire. In this paper, the phrase ‘any allergic disease’ refers to any of asthma, suspAR, ECZEMA or OGIS according to the definitions given.
Measurements of sensitization
Levels of serum IgE-ab to inhalant allergens (cat, dog, horse, birch, timothy, mugwort, Dermatophagoides pteronyssinus and Cladosporium) were measured with the multi allergen tests Phadiatop® (Pharmacia CAP SystemTM). Levels of IgE-ab towards food allergens (cow's milk, egg white, soy bean, peanut, fish and wheat) were measured with fx5® (Pharmacia CAP SystemTM). Those 627 children who exhibited a positive test in Phadiatop® or fx5®, defined as IgE-ab levels ≥0.35 kUA/l, were analysed for IgE antibodies towards the selected eight airborne and six food allergens listed above. Between 2614 children tested for serum IgE-ab and 1129 children who were not, no statistically significant differences were found in the prevalence of asthma, susAR, ECZEMA or OGIS at 4 years of age. Nor was there any difference in heredity, or various environmental exposures between those who responded to the questionnaire and took part in the examination and/or blood test at age four, and those who did not.
For symptom related outcome data the results of the whole cohort were evaluated. For those of which serum samples had been obtained the levels of positive (≥0.35 kUA/l) IgE-ab towards the selected food and inhalant allergens were summated for each child, and the median level with a 95% CI was determined. Similarly, the number of allergens of ≥0.35 kUA/l towards the selected food and inhalant allergens were also summated in each child. This sum of the number of positive allergens was expressed as means with 95% CI interval. Confidence intervals were considered significant when they did not overlap. Chi square tests and OR with 95% confidence interval were performed when appropriate.
Logistic regression was used to evaluate the relationship between the sum of specific IgE antibody levels and the number of sensitizing allergens in relation to the likelihood of presence of allergic diseases. The two variables were first considered one by one and then analysed together, also with an interaction term. Appropriate tests and confidence intervals were according to Wald, using a P-value of 0.05 as significant (24). Computerized statistical analysis was performed with SAS System V 8.01.
Of the cohort of 3743, 1202 children (32%) at the age of four exhibited at least one manifestation of asthma, suspAR, eczema or OGIS. According to the definition used 260 children had asthma (7%) and 387 (10%) had suspAR. Five hundred and ninety-five children (16%) fulfilled the criteria of eczema, and OGIS was found in 356 (10%) children. Eight hundred and eighty-four children (24%) were classified as having only one of the allergic diseases. The remaining 318 (8%) exhibited at least two clinical manifestations of either asthma, suspAR, eczema or OGIS as illustrated in Fig. 1. Of the children with asthma or rhinitis as many as 48% were classified as having at least one additional allergic disease condition. For eczema the results were the opposite with almost 65% of the individuals classified as having eczema as the single symptom (P < 0.001). Of the 2614 children giving blood, 627 (24%) exhibited a positive test in Phadiatop® or fx5®, and of those, 59 children were negative to all of the selected allergens leaving 568 (22%) children sensitized to at least one allergen. Of those having positive IgE-abs 311 (55%) were classified to have at least one of the allergic symptoms, whereas 251 (44%) had no symptoms.
The likelihood of the presence of the any allergic disease in relation to presence of IgE-abs to Phadiatop and fx5 (≥0.35 kUA/L) in 2614 4-year-old children was evaluated qualitatively. Among the 1202 children with any of asthma, suspAR, eczema or OGIS, 38% (95% CI 35–41%) exhibited such antibodies whereas in children with no such allergic disease, IgE-abs were found in 17% (15–19%) (P < 0.001).
The IgE antibody sensitization was most prevalent towards inhalant allergens such as birch, cat, timothy and horse, and also to some foods milk and peanut, with the highest prevalence in those individuals suffering from symptoms (Fig. 2A). Among those without symptoms the prevalence was generally lower except for milk. The highest median IgE-ab levels noted were towards birch followed by peanut, cat, dog and horse (Fig. 2B). With the exception of peanut, food allergens exhibited relatively low median levels of IgE-ab irrespective of symptoms. Taking both prevalence and concentration of IgE-ab into consideration, birch, cat and peanut seemed to be the most important allergens.
For children classified as having asthma, suspAR, eczema or OGIS, the levels of IgE-ab of ≥0.35 kUA/l) towards the selected food and inhalant allergens were summated and the number of allergens positive at test counted. The quantitative median levels in kUA/l and the number of sensitizing allergens in relation to disease are presented in Table 1. Those without any allergic disease had a significantly lower median sum compared with those with any of disease conditions, 1.5 and 10.7 kUA/L, respectively (P < 0.001). Apart from those without any allergic disease condition, the only significant differences in the median sum of IgE-ab levels were found between eczema and asthma and between eczema and suspAR. Moreover, for children with asthma and suspAR the median IgE-ab levels reached higher values than for children with eczema (P < 0.0001). For the number of positive allergens the mean number for individuals with any allergic symptoms was 3.4 only compared with 1.8 for those without any allergic disease (asthma, suspAR, eczema or OGIS) (P < 0.001) (Table 1).
Table 1. Quantitative median IgE antibody levels (kUA/l) and number of mean sensitizing allergens in relation to classification of allergic disease
Sum of spec IgE-ab (kUA/L)
95% confidence interval
Number of allergens
95% confidence interval
All median and mean values for symptoms of asthma, suspected allergic rhinitis (suspAR), eczema, oro-gastro-intestinal symptoms (OGIS) or any of the four disease conditions is separated symptoms of no disease with a P-value of <0.001.
Any of the four disease conditions
In order to further evaluate the value of quantitative analysis, i.e. the sum of IgE-ab levels towards the allergens that tested positive, and also the value of the total number of allergens that tested positive, logistic regression ere used to relate to the likelihood of identification of any allergic diseases (Fig. 3). As expected, no or low levels of IgE-ab to food and inhalant allergens were found to be the best discrimination for absence of any allergic disease (asthma, suspAR, eczema or OGIS). However, for the identification of an individual with any allergic disease, using only the information of sensitization, a sum of IgE-ab levels of at least 34 kUA/l was required to achieve a likelihood of 75%, or, using only the information of number of positive allergens, more than four allergens had to give positive test results (4.3 allergens to be precise). In terms of risk increase this corresponds to an increase in risk that is approximately 10-fold higher when being sensitized compared to a non-sensitized individual. To identify an individual with any allergic disease at a likelihood at 90%, using only the information of sensitization or the number of positive allergen, a sum of specific IgE-ab levels of more than 400 kUA/l was required, or seven or more positive allergen results (Fig. 3). This implies that being sensitized to a certain level of IgE ab can be associated with an increased risk compared with non-sensitized individual, Table 2.
Table 2. Odds ratio (OR) for different levels of IgE antibody levels (kUA/l) compared with a non-sensitized individual
95% confidence interval
With the aim to link the IgE-ab formation even stronger to the presence of disease the sum of positive (≥0.35 kUA/l) IgE-ab levels and the number of positive (≥0.35 kUA/l) allergens at test the results were also used in combination and analysed using logistic regression. In this combined analysis the number of sensitizing allergens was not sufficient to identify a 4-year-old child with any allergic disease (Fig. 4A), while the sum of positive IgE-ab levels was efficient for such identification (P < 0.005). However, although the number of sensitizing allergens did not obviously contributed to the model, there was an effect of the combination of the sum of positive IgE-ab levels and the number of sensitizing allergens (P for interaction 0.017). This effect was different for the different symptoms (Fig. 4A–D). When the same model was applied to the separate diseases, different patterns between the sum of positive IgE-ab levels and the number of positive allergens were found. For both asthma and suspAR there was a significant association between the sum of specific IgE concentrations and likelihood of disease (P = 0.03, P < 0.001 respectively) and for asthma also an effect from the numbers of positive allergens (P = 0.0017). This was further accentuated when using the interacting effect of the sum and number of allergens (P = 0.004). In contrast, the number of allergens positive at test for suspAR, was not useful in this respect, (Fig. 4B,C). For children classified with eczema the opposite was found (Fig. 4D); a significant association could be demonstrated between the number of sensitizing allergens and the likelihood of disease (P = 0.0003), but no effect could be seen neither for the sum of specific IgE concentrations nor by a combination of the sum and the number of positive allergens.
For children with eczema the sensitizing allergens were predominantly foods at 38% followed by grass, animals and other allergens, 29, 27 and 6%, respectively. For children with OGIS there was no significant increase in likelihood for identification by the use of specific IgE-ab neither expressed as the sum of IgE-ab levels nor as the number of allergens that tested positive (data not shown).
Using both the information of sensitization and the number of positive allergens, a likelihood of 90% could be reached for different combinations of number of positive allergens and level of sensitizations, e.g. six positive allergens and a sum of 86 kUA/l, or eight positive allergens and a sum of 32 kUA/l (Fig. 5).
In this prospective birth cohort study of 4-year-old children we have shown that sensitization to inhalant outdoor and indoor allergens, such as allergens from birch and timothy pollen, and cat dander, together with the food allergen cow's milk protein and peanut, was most prevalent. In addition, birch, peanut, cat, dog and horse gave the highest IgE-ab levels, indicating a clinical importance of these allergens for the sensitization process as well as triggering symptoms. In particular, using quantitation of IgE-antibodies 7.2-fold higher median values were recorded than among symptomatic individuals. For the number of positive allergens the difference was twofold for the symptomatic individuals, similar to the prevalence of sensitization between the children with any symptom and the children without symptoms.
In cases of uncertainty to evaluate the symptoms as an allergy, a sum of IgE ab levels against common sensitizing food and airborne allergens of at least 34 kUA/l could correctly identify any allergic disease (asthma, suspAR, eczema and OGIS) in three of four children. The presence of positive test reactions to more than four allergens, revealed allergic disease with a similar likelihood of 75%, and if seven allergens were positive at test the likelihood increased to 90% for such recognition of disease. For identification of asthma, suspAR or eczema in the children, the importance differed of the sum of allergen-specific IgE-ab levels and of the number of allergens positive at test. This probably reflected the natural history of the allergic conditions in this young age group of 4-year-old children living in Sweden. For an appropriate identification of asthma a significant interaction between the sum of specific IgE levels and number of sensitizing allergens was required. This points to that for identification of symptoms as allergic, not only the IgE-ab levels, but also the number of allergens are involved in the process and need to be known. For the likelihood to identify a child with mainly the sum of allergen specific IgE-ab levels contributed. In contrast, for classification of eczema in the children it appeared that the number of allergens positive at tests was related to the likelihood of disease, whereas the summated levels of allergen IgE-ab had little influence. This influence of number of allergens positive at test could not be explained by that the important allergens were only foods, since rather low IgE-ab levels were found against the six common food allergens tested, with peanut as the exception. Furthermore, only 38% of the food allergens were associated to eczema. Hence, for correct identification of eczema in young children, inhalant allergens also need to be considered.
The association between sensitization and allergic disease has been known for a long time. With a qualitative approach to our data we found that 17% children were sensitized to at least one of the 14 food and airborne allergens studied despite being classified as having no allergic disease whereas twice as many (38%) of the children with asthma, suspAR, eczema or OGIS were sensitized to the allergens studied. As such a considerable number of children were sensitized without symptoms of any allergic disease, such information has limited value in the clinical diagnosis of symptomatic individuals. This emphasized the need for evaluation of not only the presence of any positive IgE ab result but also the absolute IgE-ab levels, as well as the number of allergens positive at test for the most accurate identification of an allergic aetiology of symptoms in paediatric allergy care and research. One likely reason for the observations in this age group may be that at the age of four, sensitization to food is gradually losing its importance whereas the sensitization to inhalant allergens is gradually increasing its importance. Thus, although sensitization to food was as common as sensitization to airborne allergens, the median IgE-ab levels were considerably lower.
Sensitization in early childhood to food or airborne allergens has been reported to be predictive for asthma later in childhood (25–27). Furthermore, wheezy infants with IgE-abs to food or airborne allergens seem to have an increased risk for continuation of problems with respiratory symptoms (28). Several publications have addressed the issue of quantification of allergen-specific IgE, focusing mainly on concentrations of IgE-ab to certain foods and the probability of reacting to a food at a given IgE-ab level (13, 29). Recently data have been published on quantification of IgE to hen's egg allergen in infants and young children in relation to prospective duration of clinical reactions to egg (14).
In an earlier article we showed that the probability of correctly classifying 4-year-old children with asthma, rhinitis, eczema or food allergy increased significantly with increasing IgE-ab levels as measured with Phadiatop® or fx5® (16). In that publication we suggested that quantitative analysis of IgE seemed promising for the diagnosis of any allergic diseases in young children. However, because we had used multi-allergens tests, we were not able to disentangle the importance of certain allergens for classification. Besides, the children had been classified as ‘food-allergic’ if the parents reported asthma, rhinitis or eczema after ingestion of a certain food. The present study allows us to draw more accurate conclusions on the use of quantification of IgE-ab to identify allergic diseases and to study the importance of individual allergens.
When allergen specific IgE sensitization is used in the context of presence or absence of IgE, either in vivo or in vitro methods can be used. If, on the contrary, the interest is focused on whether allergy related symptoms like wheezing, eczema, rhintis, urticaria, etc, are related to the presence of IgE ab to allergens, it would be unlikely that there is a specific discrimination point for each of these that would give a high sensitivity and specificity. In an individual with an increased level of IgE ab a significant risk would be more likely compared with an individual with no or little such antibodies.
For both asthma and suspAR an interaction between the summated IgE-ab levels and number of allergens positive at test was observed for the likelihood of correct classification of disease. This raises the question of which allergens should be included in a test profile, so that one includes not only the allergens that are likely to cause symptoms, but also allergens that will help to identify children with allergic diseases. This will be further investigated in a planned follow-up study of the cohort. It needs to be borne in mind that the examined children for this study were only 4 years of age and had been raised in a region where sensitization to house dust mites and moulds is uncommon. The allergens that need to be in such a test profile for diagnosis of allergic disease will probably vary depending on both subject's age and place of residence. Used in this way sensitization can be used, not only as a crude risk estimate, but also for estimating an adjusted risk estimate, or the risk increase when sensitized in conjunction with other factors. In the case where the property that is measured is likely to more serious with high values, the measuring range can either be divided into several groups and the risk compared with absence or low values can be estimated for increasingly higher groups, or the risk is estimated using the continuous scale. In the dichotomous analysis of the presence or absence of a particular factor like IgE antibodies in relation to a particular disease, Receiver Operating Characteristics (ROC) has been used in making a medical decision since the beginning of the 1960s the tests utilized as supporting clinical diagnosis of allergy have in most cases already been evaluated for efficacy in determining the presence of IgE ab above a certain threshold in relation to disease. An alternative is to use the threshold and express the risk for the disease depending on positive ore negative test results as an odds ratio (OR). The traditional performance characteristics can be employed to describe the ability to detect the disease. However, when there is an interest whether such IgE ab can be regarded as risk factor for the presence or absence of disease, the ROC analysis has limitations.
Based on the results in this study we suggest that the summated levels of IgE-ab to common food and inhalant allergens in combination with the number of allergens tested positive can be used as one of several tools for diagnosis of the allergic child. Eczema may be easier to diagnose on the basis of objective and visible signs of disease. However, the way asthma and allergic rhinitis present themselves, in particular in young children, sometimes make diagnosis difficult and often diagnosis is delayed. The described procedure in this study may allow an earlier identification and diagnosis of young, symptomatic children.
The study was financially supported by the Vårdal Foundation, The Swedish Asthma and Allergy Foundation, The Heart and Lung Foundation and the County Council of Stockholm. The reagents costs were supported by Pharmacia Diagnostics, Uppsala, Sweden.