Multi-level Modelling of International Variations and Time Trends in Asthma and Allergic Diseases in Children and Adults

该项目将结合对哮喘病因和时间趋势的实质性研究与发展统计方法来处理复杂的多层次数据集。主要数据来源来自三个相互关联的研究：国际儿童哮喘和过敏研究（ISAAC）第一阶段（1992-1995年）通过问卷调查收集了来自56个国家155个合作中心的721,602名儿童的哮喘、鼻结膜炎和湿疹的症状和危险因素的信息。ISAAC第三阶段（2001-2002）从98个国家237个中心的1,187,496名儿童收集了类似的信息。全球哮喘网络（GAN）将在2016-2017年期间使用ISAAC方法收集类似数据，但会扩展关于父母的数据以及关于哮喘管理和环境的额外问题。迄今为止，来自34个国家的88个中心已登记为氮化镓，登记工作正在进行中。为研究收集的数据类型包括有关症状和严重程度的信息，以及三个不同年龄组的环境和人口数据，一个是儿童，一个是青少年，一个是成年人。这很符合MRC的战略研究目标，即通过生命过程调查健康状况，并了解生活方式和环境如何影响健康。将在每个国家内寻找至少一个城市和一个农村中心，以便探索一般环境对哮喘的影响。每个中心每个年龄组至少3000人的样本量将被用来提供足够的力量来检测哮喘严重程度的差异。在人口较少的岛屿，将选择该年龄组的所有学生及其父母。从现有的ISAAC调查开始，这些调查尚未充分利用最先进的统计建模技术，然后将其与GAN调查数据结合起来，将有可能对所有三项研究进行综合分析。这一复杂的多层次结构将包含3个年龄组、4个不同层面（个人、学校、中心和国家）、最多3个时间点的多重暴露。对于一些中心，将会遗漏年龄组和/或遗漏时间点，这些需要正确处理，以确保统计推断是可靠的。通过将大型、多层次数据集的新型统计方法的发展与哮喘和其他过敏性疾病的潜在原因或风险因素的流行病学问题的应用相结合，这与MRC定量和跨学科技能的技能优先级非常吻合。这个项目的结果应该对多个国家的公共卫生官员有用。了解不同层次的重要危险因素，可以通过改变卫生政策，提高对哮喘和其他过敏性疾病的患病率和严重程度产生积极影响的能力。结果将因国而异，这项工作将有助于估计潜在政策变化的影响。此外，方法学的发展将对从事其他卫生研究领域的统计学家有用，但这些统计学家使用的是类似构建的多层次数据集。

This project will combine substantive research into asthma causes and time trends with developing statistical methodologies to handle complex multi-level datasets.The main data sources are from three linked studies:Phase I (1992-1995) of the International Study of Asthma and Allergies in Childhood (ISAAC) collected information by questionnaire on symptoms and risk factors for asthma, rhinoconjunctivitis and eczema from 721,602 children in 155 collaborating centres in 56 countries. Phase III (2001-2002) of ISAAC collected similar information from 1,187,496 children in 237 centres in 98 countries. The Global Asthma Network (GAN) will use the ISAAC methodology to collect similar data during 2016-2017 but with the extension of data on parents and extra questions on asthma management and the environment. To date, 88 centres from 34 countries have registered for GAN and registration is ongoing.The type of data collected for the studies includes information on symptoms and severity along with environmental and demographic data for three different age groups, one in childhood, one teenage and one adult. This fits in well with the MRC strategic research aims of investigating health through the life course, and understanding how lifestyles and environment affect health. Within each country at least one urban and one rural centre will be sought so the influence of the general environment on asthma can be explored. A sample size of at least 3,000 people per age group per centre will be used to give sufficient power to detect differences in the severity of asthma. In smaller island populations, all pupils in the age group and their parents will be selected.Starting with the existing ISAAC surveys, which have yet to be fully exploited with state-of-the-art statistical modelling techniques, and later combining it with the GAN survey data it will be possible to run integrated analysis on all three studies. This complex multi-level structure will contain multiple exposures on 3 age groups, at 4 different levels (person, school, centre and country) across up to 3 time points. For some centres there will be missing age groups and/or missing time points which will need to be addressed correctly to ensure the statistical inference is robust. This aligns well with the MRC skills priorities of quantitative and interdisciplinary skills by combining the development of novel statistical methodologies for large, multi-level datasets with their application on epidemiological questions of potential causes or risk factors for asthma and other allergic diseases.The results of this project should be useful to public health officials within multiple countries. Understanding the important risk factors at different levels improves the ability to positively impact the prevalence and severity of asthma and other allergic diseases through changing health policies. Outcomes will vary widely from country to country and this work will help estimate the impact of potential policy changes. In addition, the methodological developments will be useful for statisticians working in other areas of health research but with similarly constructed multi-level datasets.