Childhood Wheezing, Asthma, Allergy, Atopy, and Lung Function: Different Socioeconomic Patterns for Different Phenotypes

Identifying preventable exposures that lead to asthma and associated allergies has proved challenging, partly because of the difficulty in differentiating phenotypes that define homogeneous disease groups. Understanding the socioeconomic patterns of disease phenotypes can help distinguish which exposures are preventable. In the present study, we identified disease phenotypes that are susceptible to socioeconomic variation, and we determined which life-course exposures were associated with these inequalities in a contemporary birth cohort. Participants included children from the Avon Longitudinal Study of Parents and Children, a population-based birth cohort in England, who were born in 1991 and 1992 and attended the clinic at 7–8 years of age (n = 6,378). Disease phenotypes included asthma, atopy, wheezing, altered lung function, and bronchial reactivity phenotypes. Combining atopy with a diagnosis of asthma from a doctor captured the greatest socioeconomic variation, including opposing patterns between phenotype groups: Children with a low socioeconomic position (SEP) had more asthma alone (adjusted multinomial odds ratio = 1.50, 95% confidence interval: 1.21, 1.87) but less atopy alone (adjusted multinomial odds ratio = 0.80, 95% confidence interval: 0.66, 0.98) than did children with high SEP. Adjustment for maternal exposure to tobacco smoke during pregnancy and childhood exposure to tobacco smoke reduced the odds of asthma alone in children with a low SEP. Current inequalities among children who have asthma but not atopy can be prevented by eliminating exposure to tobacco smoke. Other disease phenotypes were not socially patterned or had SEP patterns that were not related to smoke exposure.

reported by the mother, was categorized as smoker or non-smoker using the number of cigarettes smoked during week and weekend days at 73 months. Environmental tobacco smoke (ETS) exposure was ascertained in gestation (32 weeks) and childhood (77 months) based on the following question "how often during the day you are in a room or enclosed place where other people are smoking?" and categorised as "never or less than 1 hour" and "1 hour or greater". Finally, cotinine levels were measured from blood samples provided at age 7-8 years. Each sample was run in duplicate and the average of these two measures was taken. In order to determine cotinine levels the absorbance at a wavelength of 450nm was measured in ng/ml. The assay uses a cubic spline curve to extrapolate and calculate cotinine concentration from the absorbance.
Use of bleach or hair dye was measured at 8 weeks of the pregnancy and was categorized as "Daily or most days" or "About once a week or less" (reference group). A crowding index was calculated by dividing the number of people in the household by the number of rooms, both reported by questionnaire. These were measured at the beginning of the pregnancy (8 weeks) and during early childhood (21 months). An ordinal variable was derived with groupings of <0.5 (reference group), 0.5-<0.75, 0.75-<1 and 1 or more. Day care attendance was assessed at 15 months by questionnaire and categorized as "not being in childcare" (reference group), "childcare at someone else home" or "being at nursery". At 85 months the mother reported about the child's health. A positive response to chest infections in the last 12 months is used in these analyses.
Bedroom temperature in winter (8 weeks pregnancy) or coldest time of the year (85 months childhood ) were categorised as "cold or very cold" or "very warm, warm or about right" (reference group). Presence and level of mould in the house was categorized as "fairly or very serious" or "none or not serious" (reference level) through questionnaires at 8 weeks gestation and 85 months in childhood. Pet ownership was calculated based on positive responses to owning either cats, dogs, rabbits, rodents, birds or other pets at pregnancy (8 weeks) and childhood (85 months). Exposure to pests in the home (8 week pregnancy and 85 month childhood) was determined if the mother reported that the home was invaded or there was dirt in the garden/balcony/yard related to rats, mice, pigeons or cockroaches. Gas cooking was ascertained if the mother positive response to "Do you use gas for cooking (rings and/or oven)?" at 8 week pregnancy at 85 months childhood.
Finally, maternal and paternal history of asthma, eczema and hay fever were reported by questionnaire administered at 12 weeks gestation.

Multivariable multiple imputation
There was a varying degree of missing values due to the long term follow-up. In order to increase efficiency and minimise selection bias we used multivariable multiple imputation to impute missing variables for participants considered to be eligible. We used chained equations(1), where a separate regression model was specified for each missing variable which included all life course exposures and characteristics, outcomes and potential predictors of missing data (2). We generated 25 imputed datasets and combined the multiple results into one multiple-imputation inference using Rubin's combining rules (3). Supporting Table S1 provides the percentage imputed for each variable and the mean value/distribution for continuous/categorical variables in the imputed and the original dataset. Compared with "No asthma and no atopy" group. 3 Tests equality of coefficients across outcome groups Web Table 3. Adjusted association of low paternal education (compared to high education) for asthma and eczema in the last 12 months, and persistent wheezing, before and after simultaneously adjusting for groups of exposures. ALSPAC, 1991-1999.

Model
Asthma * "Persistent wheezing" of the 6 wheezing phenotypes, compared with "Never/infrequent" wheezing group and using a child's phenotype probability as