We analyzed data from the population-based birth cohort BAMSE (Barn/Children, Allergy/Asthma, Milieu, Stockholm, Epidemiology) of 4089 children born from 1994 to 1996 in Stockholm, Sweden. The study has been described in detail previously . In brief, baseline data such as background exposures and parental allergy were obtained shortly after birth. The children were followed up at 1, 2, 4, and 8 years of age using questionnaires asking about symptoms of allergy-related disorders. The response rate at the 8-year follow-up was 84% from baseline. Families that completed the questionnaire were invited for the clinical follow-up including blood sampling (n = 2480). Sera were analyzed using the Phadiatop® IgE test for allergic sensitization to pollen of birch, timothy, and mugwort, cat, dog, horse, Cladosporium herbarum, and Dermatophagoides pteronyssinus with ImmunoCAP System (Phadia AB; Uppsala, Sweden). A level of ≥0.35 kU/l was considered as positive. Included in our analyses were children with information on parental allergic diseases from the baseline questionnaire as well as rhinitis symptoms and results from the Phadiatop test at 8 years (n = 2413).
Permission for the study was obtained from the Ethics Committee of Karolinska Institutet, Stockholm, Sweden. The parents of all participating children gave informed consent.
Information on maternal and paternal hay fever, asthma, eczema, pollen, and pet allergy was collected at baseline (Table S1). As the definition of parental allergic rhinitis was self-reported and not based on sensitization, we chose to label it as ‘hay fever’.
Classification of outcomes
Allergic rhinitis (AR)
Reported symptoms of rhinitis (sneezing, runny or blocked nose in the absence of cold or flu within the last 12 months , or after exposure to pollen or furred animals), in combination with sensitization to the inhalant allergens tested (Phadiatop ≥ 0.35 kU/l) .
Nonallergic rhinitis (NAR)
Reported symptoms as above with no sensitization to any of the inhalant allergens tested (Phadiatop < 0.35 kU/l) .
At least 4 episodes of wheezing or at least 1 episode of wheezing among patients that had received a prescription for inhaled steroids, in the last 12 months .
Dry skin in combination with itchy rash for 2 or more weeks, at typical localizations, within the last 12 months and/or doctor's diagnosis of this disorder .
The odds ratios (ORs) and 95% confidence intervals (CIs) for AR and NAR were estimated using logistic regression models.
Firstly, we explored differences between maternal and paternal heredity in relation to AR and NAR (results presented in Table 1). We used categorical variables, where ‘reported hay fever’ was stratified into ‘mothers’, ‘fathers’, or ‘both’ and analyzed with ‘no heredity’ as reference category. To not reduce the number in the analysis, the other two diseases, asthma and eczema, were allowed. The same principle was used for asthma and eczema.
Table 1. Maternal versus paternal allergic disease, and the risk of allergic and nonallergic rhinitis at 8 years of age, with the different diseases analyzed as separate categorical variables stratified into mother, father, or both, among 2413 children, in the BAMSE birth cohort
|Reported hay fever|
Secondly, we compared the associations between parental hay fever, asthma, and eczema, and AR and NAR in the offspring (results presented in Table 2). One combined variable was used and analyzed with ‘no heredity’ as reference category. Hay fever only consisted of mothers, fathers, or both reporting only hay fever. The other diseases were excluded. The same yields for asthma only and eczema only.
Table 2. Patterns of parental allergic diseases, isolated or in combinations, and the odds for allergic and nonallergic rhinitis at 8 years of age, among 2413 children in the BAMSE birth cohort
|Hay fever onlyb||404||67||16.6||2.2||1.6–3.2||20||5.0||1.3||0.8–2.3|
|Hay feverb and asthmab||172||32||18.6||3.0||1.9–4.8||20||11.6||3.8||2.1–6.7|
|Hay feverb and eczemab||339||57||16.8||2.5||1.7–3.7||26||7.7||2.3||1.4–3.9|
|Asthmab and eczemab||45||8||17.8||2.9||1.3–6.8||7||15.6||5.2||2.1–12.7|
|Hay feverb, asthmab and eczemab||202||58||28.7||5.6||3.8–8.4||25||12.4||4.9||2.8–8.4|
To be included in the reference category, ‘no heredity’, both mothers and fathers had to have answered ‘no’ to the questions of hay fever, asthma, and eczema.
We tested for potential confounding by environmental and lifestyle factors (sex, socioeconomic status, maternal smoking, furred animals in the home, older siblings, early fish intake, birth month, mothers age, breastfeeding, home dampness). All these had little effect on the observed odds ratio (less than 2%), which is why we kept the final analyses unadjusted.
As co-occurrence between allergy-related diseases is common [1, 10] and may exist in one or both parents, there are many possible patterns of parental allergic disease. Thus, as a complement to the logistic regression analysis, we also performed a cluster analysis to search for latent phenotypes of heredity that are likely to be associated with AR and NAR. In a cluster analysis, it is not decided beforehand how to group the variables. Instead the data itself form the groups (clusters). For the cluster analysis, we used the k-means method, which is a frequently used method for clustering of large data sets [14-16]. We tested 3-, 4-, 5-, and 6-cluster models. The 10 variables clustered were reported hay fever, reported asthma, reported eczema, reported pollen allergy, and reported allergy to furred animals in mothers and in fathers, respectively. When a cluster model was found, proportions of the outcome variables AR and NAR were calculated for the different clusters. After repeated trials, we chose the cluster solution with the most commonly appearing clusters.
All analyses were conducted using STATA statistical software version 11 (College Station, Texas, USA).