DISCUSSION
Among the five obesity-related factors examined for causality with childhood asthma, we obtained consistent evidence that obesity, SDB, and sleep quality are observationally and genetically associated with asthma. Prospective survival analysis also confirmed the temporal relationship. When stratified by asthma subphenotypes, SDB was associated with atopic incident asthma. Detection and treatment of sleep-related symptoms may reduce the risk of asthma.
In our data, poor sleep quality and SDB were the strongest risk factors contributing to asthma in the observational and MR analyses. Their predicted prevalence of asthma was also the highest in the Q5 GRS. Two previous prospective studies support our findings that insomnia increases the risk of asthma in adults7, 8. Brumpton et al. reported an increased risk of asthma in those having problems initiating sleep, waking up too early, and experiencing nonrestorative sleep8. Moreover, Lin et al. conducted a population-based study in Taiwan and found that patients with insomnia had a 2-fold risk of having new-onset asthma during the 4-year follow-up7. On the other hand, our study is the first to investigate this association among school-age children. Although the prevalence of insomnia among school-age children is not high, the global pandemic of sleep loss among children has gained the attention of public health and school officials24. According to a large-scale meta-analysis25, European adolescents sleep for an average of 500 minutes per day, whereas Asian adolescents sleep for only 400 minutes per day on weekdays. In our study, we used the PSQI questionnaire, representing a combination of various components of sleep parameters, to assess sleep quality in children. Poor sleep quality was a robust risk factor for asthma in the three models. The mechanism behind poor sleep quality leading to asthma may be explained by the shifting of T helper (Th) cell activity toward a Th2 cytokine profile under chronic sleep deprivation26. Moreover, protracted upregulation of inflammatory markers, such as increased interleukin (IL)-6, was also observed among patients with poor sleep27. IL-6 production induced by poor sleep quality might exacerbate airway hypersensitivity. Other potential mechanisms, such as poor sleep–induced chronic inflammatory status, manifesting as increased NF-kB level28 and decreased IFN-γ29, may also explain why poor sleep quality triggers new-onset asthma.
SDB was another strong risk factor for asthma in the present study. Although researchers have postulated that SDB and asthma exhibit similar airway inflammation, the temporal relationship between the two remains unclear30. The current study is the first to confirm the causal association using both MR analysis and a prospective cohort. SDB is associated with excessive daytime sleepiness, the release of proinflammatory cytokines, and increased leptin level, contributing to the development of asthma31. Our finding that SDB predicts atopic incident asthma is supported by one study32 but contradicted by another related study, which found that SDB is associated with neutrophilic airway inflammation33. Further studies are warranted to explain relevant mechanisms linking SDB and asthma.
The strength of the current analysis centers on the use of three methods (observational, MR, prospective) to investigate the causal associations between risk factors and asthma. Our review of the literature suggests that no related study has employed three methods (observational, MR, and survival analysis) to investigate modifiable obesity-related factors for childhood asthma. The unique, population-based children cohort with prospective follow-up surveys enabled this analysis. Our study used self-reported physical activity and sleep quality, which represents a potential limitation. However, the TCHS employed globally used validated questionnaires to assess physical activity and sleep quality, limiting the reporting bias. Moreover, F-statistic suggests that the GRS of sleep quality is a weak instrument, and we might be underpowered to detect small-moderate effect sizes, as compared to large sample GWAS studies.