Prevalence and co-occurrence of parentally reported possible asthma and allergic manifestations in pre-school children

Sweden is one of the most sparsely populated countries in Europe, the median population density being 26 people per square kilometre, and 80% of municipalities have 82 people or fewer per square kilometre [16]. The corresponding numbers for the municipalities included in this study were 57 and 130 people. For administrative purposes, Sweden at the time of the data collection was divided into 25 regions and 290 municipalities, the smallest administrative unit.

All Swedish pre-school children are entitled by law to day-care (DC) organised by the local municipality. In 2002, 74% of all pre-school children attended DC, somewhat fewer among the youngest children and more than 80% of children of 3 or older [15]. The vast majority of DCs are run by the local municipal administration. The few privately operated DCs are all subcontracted to the municipal administration and follow the same set of rules as publicly operated DCs. A DC may have one to four sections. At the time of the study 15–20 children were cared for in each section. Many sections had children of all ages, but some were age stratified (1–3 years or 4–6 years). The day-care fees are heavily subsidised by the municipalities, parents usually paying about 10% of the real cost. A detailed description of choice of day care has been given elsewhere [17].

During the 1990s special DCs for children with asthma or allergies, ‘allergen avoidance day-care centres’ (AADCs) were established at the initiative of municipal school administrations, parents, local politicians, and local DC staffs. The operations, set of rules, and fees are the same as for ordinary day-care centres (ODC) with the exception that AADCs give priority to children with asthma or allergies, but accept other children as well, space permitting. As shown in a previous report all AADCs had strict regulations to avoid pet, smoking, perfume, and dust exposure [18]. The ODCs usually had no such regulations.

In the late 1990s all 72 AADCs in Sweden were identified as previously described in detail [18]. The two geographically closest ODCs to each AADC were chosen as control centres. Later, a few AADCs were closed and a few new ones were opened, leaving 70 AADCs with 84 sections and 140 ODCs with 440 sections for this study, in 62 municipalities, all over Sweden. One third of the AADCs were located in the same building as a control ODC. The indoor and outdoor environments at these day-care centres have been reported elsewhere [18].

The addresses of the 1,412 children attending the AADCs and the 7,345 children attending the ODCs were obtained from the local school authorities. A questionnaire was mailed to the parents of these children. Responses were obtained regarding 1,001 AADC children (70.9%) and 4,958 ODC children (67.5%), after two reminders when necessary. Of the respondents, 1,000 AADC children and 4,886 ODC children were 6 years old or younger. The ODC children constitute the study population for this report, except in the prevalence calculations, where also data from the AADC children was used.

The ISAAC written screening questionnaire with questions about asthma and wheezing, eczema, and rhinitis, extensively used all over the world and regarded as the gold standard for postal questionnaires on childhood asthma, rhinitis, and eczema, was used [1]. Although intended for children 6 years or older, it has been validated down to 3 years old with good results [19]. For this study, supplementary questions on medical treatment, physician assessed asthma, rhinitis and eczema diagnosis, food allergy, allergy to pollen or furred pets, parental education, smoking habits, and some additional variables not reported here, were added (see Additional file 1).

Regarding possible asthma, the following criteria were used: 1) ever had asthma and has current symptoms (wheezing in the last 12 months), or 2) wheezing 4 times or more during the last 12 months, or 3) physician diagnosis and current symptoms, or 4) current use of inhalation steroids. Regarding rhinitis, the following criteria were used: 1) ever had hay-fever and has current symptoms (sneezing or a runny or a blocked nose when the child did not have a cold or the flu), or 2) rhino-conjunctivitis without cold last 12 months, or 3) physician diagnosis and current symptoms, or 4) being on anti-allergy medication and having current symptoms.

The following eczema criteria were used: 1) itchy rash coming and going for at least 6 months in typical eczema localisations in the last 12 months, or 2) physician diagno-sis and current symptoms (itchy rash), or 3) current use of steroid ointments and current symptoms. The following food allergy criterion was used: Reported food allergy to milk, egg, fish, peanuts, nuts, soy, or stone fruits, but not including lactose or gluten intolerance.

Parental education was classified as a nine-year compulsory education (=1), a two-year upper secondary education (= 2), a three-four-year upper secondary education (= 3), or university or other tertiary education (= 4).

The study was approved on several occasions before and during the data collection process, first by the Research Ethics Committee at Uppsala University and later by the National Research Ethics Board.

The statistical analyses were conducted using the SAS software, version 9.3 [20]. Partial non-response (missing data in returned questionnaires) was on average 0.6% with a maximum for individual variables of 1.1%. Simple (crude) differences between groups in proportions were tested with the Mantel-Haenszel chi-square test.

A possible asthma variable (0 = no, 1 = yes) was created based on fulfilment of any of the criteria presented above. A rhinitis, eczema, food allergy, furred pet and pollen allergy variable was constructed accordingly. The variables selected were those of primary interest for the hypothesis. The analyses of co-occurrence between possible asthma, rhinitis, eczema, food, furred pet and pollen allergies were performed with multiple logistic regression analysis with presence of possible asthma, rhinitis and eczema (yes/no) as dependent variables, one at a time, and the others as independent variables, first by age (Figure 1) and then adjusted for the influence of age and sex (Table 1). The analytical technique provides odds ratios (OR), confidence intervals (95% CI), and Wald’s chi-square. The latter is the test variable on which the p-value is based. Consequently, Wald’s chi-square may be used to rank the impact of the independent variables on outcome.

Most of the municipalities represented in this study had only one AADC and two ODCs sampled, irrespective of population size, resulting in under-representation of large municipalities in the prevalence calculations. To adjust for this circumstance, an analytical model was employed in which the total number of children, by age and sex, in each municipality represented in the study was downloaded from Statistics Sweden [21]. The number of children with possible asthma, rhinitis, eczema and food allergy by age and sex in the municipality was then computed, based on the prevalence of these conditions in the municipality-specific ODC study populations, after which the number with these diseases in the municipality-specific AADC study population was added. National prevalence was obtained by pooling the municipality-specific data on number of cases in relation to exposed population size. The procedure has been described in detail elsewhere [17].

Age was used as a continuous variable and defined as age attained at last birthday. The age-specific prevalence functions were computed with logistic regression technique. The fit between the crude age-specific prevalence and prevalence data generated from the analysis model was tested with a logistic regression technique. To check for non-linearity a model including a second and a third degree age polynomial, and interaction between sex and age were tested. The best fit between observed and expected outcome was obtained with a second degree (possible asthma and rhinitis) or a third degree (eczema and food allergy) age polynomial and an interaction term between sex and age, explaining 50% of the prevalence variation. On scrutiny, the model fit appeared excellent. All tests were two-tailed. The level of significance was set at p < 0.05.

Of the 4,886 ODC children, 2,476 (51%) were boys. The number of 1-year-old children was 203 (4%), 2-year-olds 916 (19%), 3-year-olds 976 (20%), 4-year-olds 1,017 (21%), 5-year-olds 1,168 (24%), and 6-year-olds 606 (12%). The age distribution was similar in the AADC children. Regarding parental education, 268 (5%) children had parents with compulsory education only, 1,032 (21%) had at least one parent with two-year upper secondary education, 1,459 (30%) had at least one parent with three- or four-year upper secondary education, and 2,127 (44%) had at least one parent with college or university education.

Possible asthma, rhinitis and eczema manifestations are presented by age groups in Table 2. The prevalence of possible asthma manifestations was stable or decreased by age, while the prevalence of rhinitis manifestations was stable or increased, and the prevalence of eczema manifestations was generally stable, except that ointment treatment became less prevalent by age.

The prevalence of food, furred pet and pollen allergies is shown in Table 3. The prevalence of egg allergy decreased by age, while the prevalence of peanut and nut allergy increased. All other food allergies had a stable prevalence. The prevalence of all allergies to furred pets increase by age and so did the pollen allergy prevalence.

The age-specific prevalence of possible asthma, rhinitis, eczema and food allergy adjusted for sampling municipality size is presented in Figure 2. The mean prevalence across age was 21.7% for eczema, 8.9% for possible asthma, 8.1% for rhinitis, and 6.6% for food allergy. The prevalence of the most common condition, eczema, increased during the early years of life from 21.3% at age 1 to 24.1% at age 3 and then decreased to 18.2% at age 6. The possible asthma prevalence increased from 9.3% at age 1 to 10.6% at age 3 and then decreased to 5.7% at age 6. The rhinitis prevalence increased from 5.2% at age 1 to 9.9% at age 6. Food allergy prevalence increased from 3.2% at age 1 to 8.4% at age 3, and was then fairly stable.

Crude co-occurrence among the ODC children is illustrated in the Venn diagram in Figure 3, showing that 35.7% had any of the manifestations. The most common manifestations or manifestation combinations were eczema only, possible asthma, rhinitis, food allergy, and various combinations of eczema, possible asthma and rhinitis, and food allergy.

Statistical measures of co-occurrence between prevalent possible asthma, rhinitis, eczema, and allergies to food, furred pets and pollen in multiple logistic regression analyses are displayed by age in Figure 1a-c and summarised across age in Table 1. Presence of possible asthma (dependent variable) was highly significantly related to the presence of rhinitis, eczema and all the allergy variables (independent variables), after adjustment for the influence on possible asthma prevalence of age and sex (co-variates), indicating a relationship between the presence of possible asthma and presence of all the other variables. The variables with the largest impact on possible asthma prevalence, as indicated by Wald’s chi-square, were in ranked order furred pet allergy, rhinitis, food allergy, eczema and pollen allergy. As shown in Figure 1a the co-occurrence between possible asthma on the one hand and rhinitis, furred pet, pollen and food allergy tended to increase by age, while possible asthma-eczema co-occurrence appeared to be more stable across age.

A corresponding analysis (Table 1, mid-section) showed that presence of possible asthma, eczema, food, and pollen allergy were all highly significantly related to presence of rhinitis. Pollen allergy was the far most important determinant, with a Wald’s chi-square value more than four times that of all other variables combined. All co-occurrence by age tended to increase except for eczema, Figure 1b.

For eczema presence (Table 1, right hand section), the most important determinant was food allergy, Wald’s chi-square measure being larger than that of all other variables combined. The co-occurrence increased by age for all variables, with the largest increase at age 2 and age 6, Figure 1c.