In this nationwide study there was considerable co-occurrence between prevalent possible asthma, rhinitis, eczema and food allergy. The prevalence of all conditions peaked at the age of 3 with the exception of rhinitis, where the prevalence increased throughout the whole age span 1–6. Eczema was by far the most common of the conditions. The mean prevalence across age of the other manifestations was approximately the same.
The strengths of the study include that it was based on a large sample of pre-school children covering all of Sweden. As shown in the Settings section, 74% of all Swedish preschool children, and more than 80% of children 3 years or older attend day-care centres. Children in Swedish day care centres may thereby be regarded as representative of all Swedish preschool children. One possible source of bias might have been that children attending DCs were more or less healthy than non-attending children. However, some studies have found that children attending DCs had more respiratory symptoms than children cared for at home , while others found no difference , and still others have found differences only in specific age groups . At any rate the differences found were small.
Moreover, the parental response rate to the postal questionnaire was 68-71%, which is higher than the approximately 50%- 60% response rate usually obtained in random population samples. The data obtained in this study are therefore equal to or more complete than corresponding data obtained from random population samples. The possible bias caused by non-response may be estimated based on the assumptions that non-respondents, for instance, had a possible asthma on average either five standard error units more often or five standard error units less often then the respondents, i.e., a considerable and highly significant difference. The overall possible asthma prevalence in respondents and non-respondents combined would then have been 9.0% if non-respondents had higher prevalence than respondents, and 8.8% if they had lower prevalence, as compared with the 8.9% we found among respondents. The potential bias owing to non-response is therefore small.
The limitations of the study include that postal questionnaire data was the only possible source of information in a study population of this size. It would have been desirable to have access to medical investigation data, however impractical with a study population of this size and dispersion across the country. The questionnaire used in this study is based on the ISAAC questionnaire, validated and used worldwide. All sources of information have inherent information bias to some extent. Medical history data usually have a low degree of such bias, and according to the validations made it is acceptable for clinical epidemiologic studies, like the present one.
The definitions of possible asthma and allergic manifestations were based on combinations of variables. Possible asthma, for instance, is in most prevalence studies defined only by wheezing during the last year. In a previous publication we were able to show that the prevalence of possible asthma criteria used in the present study was approximately the same per age and sex group, except for wheezing alone during the last year, where the prevalence was almost three times higher (approximately 35% among one-year-old boys and 19% among one-year old girls) . To improve precision and to arrive at a definition as similar as possible to that used in a clinical setting, the criteria used in this study were adopted.
In this study the total prevalence of possible asthma, rhinitis, eczema or food allergy across the age span 1–6 years was 35.7%. A previous Swedish study found a similar total prevalence, 32% . The 8.9% possible asthma prevalence is similar to what others have found [7, 9, 25]. We found a total rhinitis prevalence of 8.1%. Other western and northern European studies have reported similar prevalence levels [3, 9]. The eczema prevalence found in this study, 21.7%, is similar to results from other Swedish studies [3, 26], but higher than results reported from other European and non-European countries in the ISAAC study . The food allergy prevalence in the present study was 6.6%. Other Scandinavian and British studies reported 8%-15%, based on questionnaires [4, 27, 28], but only 2%-3.5% after oral provocation [5, 27–30]. The results from this study are thus similar to those from other Scandinavian studies, but they differ to some extent from data collected in other parts of the world.
We found a highly significant co-occurrence between the various atopic manifestations. Pet allergy and rhinitis had the largest impact effect of the allergy manifestations on the presence of possible asthma. Regarding rhinitis, pollen allergy had by far the largest impact. For eczema, food allergy had the unquestionably largest impact. These findings are well in line with clinical observations. Results from other studies are usually based on two-by-two manifestation comparisons, with a few exceptions [10–12, 30, 31] and with only partial adjustments, which means that the extent of the co-occurrence is only partially shown.
There are two interesting issues linked to co-occurrence: cause or causes and consequences. It is unlikely that the various atopic manifestations cause or trigger each other, it is much more likely that there are common underlying factors that trigger the onset of manifestations. These are presently incompletely known. It is a commonly held view that the manifestations become prevalent in some sort of ordered sequence, such as the ‘atopic march’ [12–14]. So far, no preventive action may be based on these circumstances.
However, in a German birth cohort study, children with moderate to severe early eczema often had early wheeze. Of these children, half had wheeze before the onset of eczema and half had eczema before or concomitant with wheezing . As early as at 2 years of age, 6.5% of the children in a Swedish birth cohort study had at least three of five atopic manifestations . In the present study co-occurrences were common as early as before 3 years of age. The impact of pollen allergy as compared with that of furred pet allergy on rhinitis was larger than expected in these young children and is particularly noteworthy.