北京市大气污染与呼吸系统疾病死亡的相关性——基于卫星遥感数据的时空分析

目的：利用垂直湿度校正法校正气溶胶光学厚度（aerosol optical depth，AOD），并探索校正AOD作为衡量大气污染程度的指标与人群健康结局建立暴露反应关系的可行性。方法：先采用空间插值方法对AOD、地面PM2.5、相对湿度和边界层高度进行插值以获得不同地理位置的各项数据，按照地理位置与北京市呼吸系统疾病死亡数据进行匹配。利用大气边界层高度及由相对湿度计算得到的气溶胶吸湿增长因子分别对AOD进行垂直校正与湿度校正。为进行对比，将校正AOD和PM2.5采用标准化法进行无量纲化处理，使用广义相加混合模型，控制时间趋势、温湿度效应、星期几效应和节假日效应等，得到无量纲化后的校正AOD与PM2.5对北京市居民每日呼吸系统疾病总死亡和慢性阻塞性肺病（chronic obstructive pulmonary disease，COPD）死亡的急性效应。结果：经过垂直湿度校正后的AOD与PM2.5相关系数达到0.72。累积滞后0~3 d(lag 0~3)的标准化校正AOD对呼吸系统疾病总死亡数及COPD死亡数的累积滞后效应最强，每升高1个单位，超额危险度分别为3.64%（95%CI为0.58%~6.78%）和 4.92%（95%CI为1.81%~8.14%）。累积滞后0~1 d(lag 0~1)的标准化PM2.5对呼吸系统疾病死亡及COPD死亡的累积滞后效应最强，每升高1个单位（约155 μg/m3），超额危险度分别为3.96%（95%CI为0.82%~7.19%）及6.12%（95%CI为1.44%~11.02%）。与地面PM2.5的效应相比，校正AOD引起呼吸系统疾病死亡及COPD死亡效应的置信区间较窄，对于捕捉颗粒物对居民每日死亡的滞后效应更为敏感；且不同累积滞后天数的校正AOD对呼吸系统疾病死亡及COPD死亡的效应均具有统计学意义，对于捕捉颗粒物对于居民死亡的累积滞后效应更为敏感。结论：校正AOD可以反映地面颗粒物对北京市居民每日呼吸系统疾病死亡的影响，在缺乏地面监测的情况下，可以使用校正AOD作为衡量大气污染物的指标与人群健康结局建立暴露反应关系。

Correlation study between respiratory death and airborne particles in Beijing: Spa-tiotemporal analysis based on satellite remote sensing data

Objective: To use vertical and humidity correcting method to calibrate aerosol optical depth (AOD), and to explore the feasibility of calibrated-AOD as exposure index to measure the level of air pollutants from the ground and to establish the exposure-response relationship between calibrated-AOD and people's health outcomes.Methods: First of all, we interpolated AOD, PM2.5, relative humidity and planetary boundary layer height using Kriging method to obtain data at different locations and matched different data with respiratory death in Beijing by geographical coordinates.Then, the planetary boundary layer height and aerosol hygroscopic growth factor calculated based on relative humidity was used to calbrate the AOD.To compare the effects of calibrated-AOD and PM2.5, we used standardization method to get non dimensionless calibrated-AOD and PM2.5.At last, we used the generalized additive mixed model (GAMM) to estimate the acute effects of calibrated-AOD and PM2.5 on respiratory death and chronic obstructive pulmonary disease (COPD) death, after controlling the time trend, temperature and humidity effects, days of the week effect and holiday effects.Results: The correlation coefficient between calibrated-AOD and PM2.5 was 0.72.The effects of calibrated-AOD on respiratory death and COPD death was the strongest at lag 0-3, one unit of calibrated-AOD increases was associated with 3.64% (95%CI: 0.58%-6.78%) increase of respiratory death and 4.92% (95%CI: 1.81%-8.14%) increase of COPD death.As for PM2.5, the strongest effects appeared at lag 0-1, one unit of PM2.5 (about 155 μg/m3) increases was associated with excess risks of 3.96% (95%CI: 0.82%-7.19%) and 6.12% (95%CI: 1.44%-11.02%) for respiratory death and COPD death respectively.Compared with PM2.5, the effects of calibrated-AOD on respiratory death and COPD death had narrower confidence intervals.The calibrated-AOD was sensitive to capture the lag effects, and the cumulative lag effects of calibrated-AOD were all significant on multiple lag days which indicated that the calibrated-AOD was sensitive to capture cumulative lag effects of air pollutants on respiratory death and COPD death as well.Conclusion: We believe that calibrated-AOD can be used as an index to reflect the effects of air pollutants on respiratory death in Beijing.In the absence of ground monitoring, calibrated-AOD can be used to mea-sure the relationship between air pollutants and some health outcomes.