Abstract：
Previous studies of air pollution and respiratory disease often relied on aggregated or lagged acute respiratory disease outcome measures, such as emergency department (ED) visits or hospitalizations, which may lack temporal and spatial resolution. This study investigated the association between daily air pollution exposure and respiratory symptoms among participants with asthma and chronic obstructive pulmonary disease (COPD), using a unique dataset passively collected by digital sensors monitoring inhaled medication use. The aggregated dataset comprised 456,779 short-acting beta-agonist (SABA) puffs across 3,386 people with asthma or COPD, between 2012 and 2019, across the state of California. Each rescue use was assigned space-time air pollution values of nitrogen dioxide (NO2), fine particulate matter with diameter ≤ 2.5 µm (PM2.5) and ozone (O3), derived from highly spatially resolved air pollution surfaces generated for the state of California. Statistical analyses were conducted using linear mixed models and random forest machine learning. Results indicate that daily air pollution exposure is positively associated with an increase in daily SABA use, for individual pollutants and simultaneous exposure to multiple pollutants. The advanced linear mixed model found that a 10-ppb increase in NO2, a 10 μg m-3 increase in PM2.5, and a 30-ppb increase in O3 were respectively associated with incidence rate ratios of SABA use of 1.025 (95 % CI: 1.013-1.038), 1.054 (95 % CI: 1.041-1.068), and 1.161 (95 % CI: 1.127-1.233), equivalent to a respective 2.5 %, 5.4 % and 16 % increase in SABA puffs over the mean. The random forest machine learning approach showed similar results. This study highlights the potential of digital health sensors to provide valuable insights into the daily health impacts of environmental exposures, offering a novel approach to epidemiological research that goes beyond residential address. Further investigation is warranted to explore potential causal relationships and to inform public health strategies for respiratory disease management.
摘要：
以前对空气污染和呼吸系统疾病的研究通常依赖于汇总或滞后的急性呼吸系统疾病结局指标，如急诊科(ED)就诊或住院，这可能缺乏时间和空间分辨率。这项研究调查了哮喘和慢性阻塞性肺疾病（COPD）参与者的每日空气污染暴露与呼吸道症状之间的关系。使用数字传感器被动收集的独特数据集，监测吸入药物的使用。汇总的数据集包括针对3,386名哮喘或COPD患者的456,779名短效β-激动剂（SABA）粉扑，在2012年至2019年之间，遍及加利福尼亚州。每次救援使用都分配了二氧化氮（NO2）的时空空气污染值，直径≤2.5µm的细颗粒物（PM2.5）和臭氧（O3），源自加利福尼亚州产生的高度空间分辨率的空气污染表面。使用线性混合模型和随机森林机器学习进行统计分析。结果表明，每日空气污染暴露与每日SABA使用量的增加呈正相关，对于单个污染物和同时暴露于多种污染物。高级线性混合模型发现，NO2的10-ppb增加，PM2.5的10μgm-3增加和O3的30-ppb增加分别与SABA使用的发生率比率为1.025（95％CI：1.013-1.038）有关，1.054（95%CI：1.041-1.068），和1.161（95%CI：1.127-1.233），相当于相应的2.5%，SABA粉扑比平均值增加5.4%和16%。随机森林机器学习方法显示出相似的结果。这项研究强调了数字健康传感器的潜力，可以为环境暴露对日常健康的影响提供有价值的见解。提供了一种超越居住地址的流行病学研究的新方法。需要进一步调查以探索潜在的因果关系，并为呼吸系统疾病管理的公共卫生策略提供信息。
