基于大气污染累积健康风险的天津市空气质量健康指数研究

Establishment of air quality health index based on cumulative health risk of air pollution in Tianjin

[背景]累积风险指数(CRI)可综合考虑多种大气污染物对健康的联合作用,但目前罕见基于大气污染CRI构建空气质量健康指数(AQHI)的研究。[目的]基于大气污染物CRI构建天津市AQHI并对其有效性进行验证。[方法]收集天津市2015—2019年每日大气污染、气象因素、非意外死亡数据,并建立时间序列数据库,采用描述性统计分析方法对数据的基本分布特征进行分析。采用广义相加模型的单污染物模型和双污染物模型建立大气污染物与非意外死亡的暴露反应关系,筛选最佳滞后天数和指示污染物,然后采用多污染物模型计算大气污染的CRI值,并依据CRI值构建AQHI并分级。最后采用广义相加模型分别建立AQHI和空气质量指数(AQI)与非意外死亡的暴露反应关系,并对暴露反应关系系数进行比较,以广义交叉验证(GCV)值和模型的R^(2)对AQHI的有效性进行验证。[结果]本研究选择lag1为最佳滞后天数,在综合考虑超标情况和统计学模型结果的基础上,选择PM_(2.5)、SO_(2)、NO_(2)和O_(3)作为最终的指示污染物。4种指示污染物的浓度每升高1μg·m^(-3)对非意外死亡影响的效应值b,分别为-0.00002、0.00079、0.00015和0.00042。基于上述系数计算CRI,并构建AQHI。按照分级标准提示,63%的时间是处于低风险级别,有34%的时间处于中风险级别。分别建立了AQHI和AQI与全人群、女性人群和男性人群非意外死亡的暴露反应关系,结果显示,AQHI每增加一个四分位数间距(IQR),全人群、女性人群和男性人群非意外死亡的超额风险均高于AQI的相应指标值;且AQHI模型的GCV值(分别为2.694、1.819、1.938)均低于AQI模型的GCV值(分别为2.747、1.850、1.961),AQHI模型的R^(2)值(分别为0.849、0.780、0.820)均高于AQI模型的R^(2)值(分别为0.846、0.776、0.817)。[结论]与AQI相比,基于大气污染物的CRI构建的AQHI能够较好地预测天津市空气污染对人群的健康风险。

[Background]Cumulative risk index(CRI),as a commonly used approach to estimate the joint effects of multiple air pollutants on health,has been used by few studies to construct an air quality health index(AQHI).[Objective]To construct an AQHI based on the CRI of air pollution in Tianjin and evaluate the validity of the AQHI.[Methods]Daily data on air pollutants,meteorological factors,and non-accidental deaths during2015-2019 in Tianjin we re collected to create a time-se ries object.Descriptive statistical analyses were used to describe the characteristics of the data.To determine the best lag day and indicative pollutant,single-pollutant and two-pollutant generalized additive models were fitted to construct the exposure-response relationships between air pollutants and non-accidental deaths.After that we evaluated a CRI of air pollution using multi-pollutant models and constructed an AQHI and its classifications based on the CRI.Finally,we compared the exposure-response associations and coefficients of the AQHI and the conventional air quality index(AQI)with non-accidental deaths,and evaluated the health risk communication validity of the AQHI using generalized cross validation(GCV)values and R^(2)values.[Results]We selected lagl as the best lag day and PM_(2.5),SO_(2),NO_(2)and O_(3)as the appropriate pollutants according to the unqualified rates of pollutants and significant statistical results.Oneμg·m^(-3)increase of PM_(2.5),SO_(2),NO_(2),and O_(3)was associated with-0.00002,0.00079,0.00015,and 0.00042 increase in effect size b of the non-accidental mortality,respectively.Based on these coefficients,we calculated the CRI and AQHI.According to a pre-determined classification scheme of the AQHI,the air quality of 63%study days was low risks and that of 34%study days was median risks.The associations of AQHI and AQI with non-accidental deaths in different populations were evaluated.The results showed that the excess risks of non-accidental deaths in total,female,and male populations for per interquartile range(IQR)increase in AQHI were higher than the corresponding values of AQI.The GCV values of the AQHI model(2.694,1.819,and 1.938,respectively)were lower than those of the AQI model(2.747,1.850,and 1.961,respectively),and the R^(2)values of the AQHI model(0.849,0.780,and0.820,respectively)were higher than those of the AQI model(0.846,0.776,and 0.817,respectively).[Conclusion]Compared with AQI,the CRI-based AQHI may communicate the air pollution-related health risk to the public more effectively in Tianjin.