Abstract：
Long-term exposure to ambient PM2.5 is known associated with cardiovascular and respiratory health effects. However, the heterogeneous concentrationresponse function (CRF) between PM2.5 exposure across different concentration range and cardiopulmonary disease and diabetes mellitus (DM) incidence, and their implications on attributable years lived with disability (YLD) and regulation policy has not been well-studied. In this retrospective longitudinal cohort study, disease-free participants (approximately 170,000 individuals, aged ≥ 30 years) from the MJ Health Database were followed up (2007-2017) regarding incidents of coronary heart disease (CHD), ischemic stroke, chronic obstructive pulmonary disease (COPD), lower respiratory tract infections (LRIs), and DM. We used a time-dependent nonlinear weight-transformation Cox regression model for the CRF with an address-matched 3-year mean PM2.5 exposure estimate. Town/district-specific PM2.5-attributable YLD were calculated by multiplying the disease incidence rate, population attributable fraction, disability weight, and sex-age group specific subpopulation for each disease separately. The estimated CRFs for cardiopulmonary diseases were heterogeneously with the hazard ratios (HRs) increased rapidly for CHD and ischemic stroke at PM2.5 concentration lower than 10 μg/m3, whereas the HRs for DM (LRIs) increased with PM2.5 higher than 15 (20) μg/m3. Women had higher HRs for ischemic stroke and DM but not CHD. Relative to the lowest observed PM2.5 concentration of 6 μg/m3 of the study population, the PM2.5 level with an extra risk of 0.1 % (comparable to the disease incidence) for CHD, ischemic stroke, DM, and LRIs were 8.59, 11.85, 22.09, and 24.23 μg/m3, respectively. The associated attributable YLD decreased by 51.4 % with LRIs reduced most (83.6 %), followed by DM (63.7 %) as a result of PM2.5 concentration reduction from 26.10 to 16.82 μg/m3 during 2011-2019 in Taiwan. The proportion of YLD due to CHD and ischemic stroke remained dominant (56.4 %-69.9 %). The cost-benefit analysis for the tradeoff between avoidable YLD and mitigation cost suggested an optimal PM2.5 exposure level at 12 μg/m3. CRFs for cardiopulmonary diseases, attributable YLD, and regulation level, may vary depending on the national/regional background and spatial distribution of PM2.5 concentrations, as well as demographic characteristics.
摘要：
已知长期暴露于环境PM2.5与心血管和呼吸系统健康影响有关。然而,不同浓度范围的PM2.5暴露与心肺疾病和糖尿病（DM）发病率之间的浓度反应函数（CRF），它们对残疾归因年(YLD)和监管政策的影响尚未得到充分研究。在这项回顾性纵向队列研究中，无病参与者(约170,000人，年龄≥30岁）从MJ健康数据库中对冠心病（CHD）事件进行了随访（2007-2017），缺血性卒中,慢性阻塞性肺疾病(COPD),下呼吸道感染（LRIs），DM。我们使用时间相关的非线性权重变换Cox回归模型对CRF进行了地址匹配的3年平均PM2.5暴露估计值。通过乘以疾病发病率来计算城镇/地区特定的PM2.5归因YLD，人口归因分数，残疾体重,和性别年龄组的特定亚群分别为每个疾病。在PM2.5浓度低于10μg/m3时，冠心病和缺血性中风的危险比（HR）迅速增加，而在PM2.5浓度高于15（20）μg/m3时，DM的HR（LRI）增加。女性缺血性卒中和DM的患者有较高的HR，但不是CHD。相对于研究人群中观察到的最低PM2.5浓度6μg/m3，冠心病的PM2.5水平有0.1%的额外风险（与疾病发病率相当），缺血性卒中,DM,LRI分别为8.59、11.85、22.09和24.23μg/m3。相关的归因YLD下降了51.4%，LRI下降最多(83.6%)，其次是DM（63.7%），原因是台湾2011-2019年PM2.5浓度从26.10降至16.82μg/m3。冠心病和缺血性卒中导致的YLD比例仍然占主导地位（56.4％-69.9％）。可避免的YLD与缓解成本之间权衡的成本效益分析表明，PM2.5的最佳暴露水平为12μg/m3。心肺疾病的CRF，归因YLD，和监管水平，可能因国家/地区背景和PM2.5浓度的空间分布而异，以及人口特征。
