To the Editor:

Previous studies have shown a sex-specific development of asthma; the underlying mechanisms are yet unknown (1–4). In childhood, boys develop asthma more often compared with girls; during adolescence this ratio inverts. Attempts to explain the sex-specific development of asthma include hormonal factors (2, 5, 6), lung growth (7), diagnostic practices (4, 8), and also environmental exposures (4, 7). Interestingly, under aspects of gender medicine, girls and boys differ in hygiene standards and practices, which may also contribute to a sex-specific development of asthma (9).

A prominent environmental influence on childhood asthma is the protective exposure to farming environments (10, 11). This “farm effect” has been documented in numerous studies, but evidence from longitudinal studies as well as on asthma in adolescence is scarce (10). In line with the hygiene hypothesis, early life microbial exposures activating and modulating innate and adaptive immunity have been suggested as mechanisms underlying the farm effect (11). Sex-specific hygiene practices may therefore lead to sex-specific patterns regarding the farm effect. However, to date no study has investigated sex differences in the farm effect, especially in adolescence.

The aim of this study is the detailed description of the sex-specific development of asthma in the context of the farm effect as well as the determination of potential causes of different courses of disease. Some of the results of this study have been previously reported in the form of an abstract (12).

A prospective, population-based cohort study was conducted in rural areas around Ulm, southern Germany, with baseline assessment among 6- to 10-year-old school children in 2006 as part of the GABRIELA study (13). Disproportionate random samples stratified for categories of exposure to farming environments were drawn (13). Of these 2,248 children with baseline parental questionnaires 92% completed at least one yearly follow-up in 2010–2013. Details on the methods and study population are shown in the online supplement. The ethics committee of Ulm University (Ulm, Germany) approved the study.

A doctor’s diagnosis of asthma was reported in parental questionnaires at each assessment including baseline; age at first diagnosis was reported at each follow-up. Farm children were reported to live on a farm at baseline, and nonfarm children did not.

Accounting for the complex survey design, the data were analyzed with stratified weighted Cox and logistic regression models using the Taylor series method to estimate variances. Subjects diagnosed with asthma before age 6 years were excluded from regression models because of the uncertainty of the diagnosis in this age group (14). Interactions were modeled with multiplicative terms. Putative confounders (all assessed at baseline) included as covariates in adjusted models were sex, age, number of siblings (categorically as 0, 1, 2, 3 or more), familial allergies (yes/no; asthma, hay fever, or atopic eczema in parents and/or siblings), parental smoking (maternal or paternal current or ex-smoking vs. never-smokers), and parental education (maternal or paternal qualification for university vs. all others). Furthermore, body mass index (BMI) was included as a time-dependent covariate in Cox regression models and BMI in the year of life preceding asthma diagnosis or end of observation was included as a covariate in logistic regression models (for details, see Methods in the online supplement). Onset of pubic hair growth was reported at each follow-up. Physical activity did not mediate or confound the association between living on a farm and asthma in a previous analysis of the Swiss GABRIELA cohort and was thus not considered a confounder (for details see Methods in the online supplement) (15). Statistical analyses were performed with SAS 9.3 (SAS Institute, Cary, NC).

In total, 2,064 children contributed to the analyses. Of note, because of the four consecutive school years included at baseline, in addition to those lost to follow-up, approximately 25% of the study population was censored with each year of age at the upper end of the age distribution. From age 6 years on, the data clearly show the protective “farm effect” on the onset of asthma with almost constant protection throughout childhood (Figure 1). The adjusted hazard ratio for doctor-diagnosed asthma comparing farm with nonfarm children was 0.58 (95% confidence interval: 0.36; 0.92).

Figure 1. Cumulative incidence of asthma across childhood, stratified by growing up on a farm.

Download Figure | Download Powerpoint

[More] [Minimize]

Further analyses suggest differential sex-specific development of asthma in comparing farm with nonfarm children (Figure 2). For nonfarm children, the sexes diverge at about age 5 years, with boys having a higher cumulative incidence of asthma, and rejoin at about age 13 years, whereupon the cumulative incidence is higher in girls. For farm children, the divergence of the sexes happens later, at about 8 years of age, and they keep drifting apart without any tendency to rejoin up to age 16 years.

Figure 2. Cumulative incidence of asthma across childhood, stratified by growing up on a farm and by sex.

Download Figure | Download Powerpoint

[More] [Minimize]

Analyzing the cumulative incidence of doctor-diagnosed asthma among 1,255 children already observed up to age 14 years revealed a marginally statistically significant interaction between living on a farm and sex (P = 0.066). Among 787 children already observed up to age 15 years this interaction became statistically significant (P = 0.049). After adjustment for the aforementioned confounders, these P values changed to P = 0.109 and P = 0.065, respectively.

Potential explanations for the observed sex-specific patterns include (1) delayed onset of puberty in farm children, (2) the loss of protective farm exposures or exposure to risk factors among farm boys more than among farm girls, or even (3) exposure to protective factors among nonfarm boys more than among nonfarm girls.

Among children observed up to age 15 years, farm children were slightly less likely to report onset of pubic hair growth (90 vs. 95% among nonfarm children, P_χ2 = 0.053). This difference was similar for girls and boys. The reported mean age at onset of pubic hair growth was about 12 months higher for boys than for girls (12.6 vs. 11.5 yr). The average delay in onset was 1.7 months for farm compared with nonfarm children (P_t test = 0.15). This was due mainly to a delay in onset among farm girls compared with nonfarm girls (on average 3 mo; P_t test = 0.03). These rather small differences make a delayed onset of puberty in farm children unlikely to fully explain the observed sex-specific patterns. Indeed, additional adjustment for the onset of pubic hair growth only attenuated the interaction odds ratio slightly (data not shown).

In line with the second explanation proposed previously, a study among Danish farming students found men to have longer occupational farming experience than women (16). It is conceivable that boys help out working on the farm to a greater extent than girls in adolescence and are thus exposed to hazardous rather than protective levels of farm exposures. Sample size and collected amount of data do not allow investigating this in the present study. However, further follow-ups are planned to enable analysis of the full cohort over puberty and of loss of protective or gain of hazardous exposures.

References

Section:

ChooseTop of pageReferences <<CITING ARTICLES