气温对中国五城市居民死亡率的滞后影响分析

Analysis of the lag-effects of temperature on the five cities＇ mortality in China

目的对中国5个城市气温与死亡率关系进行分析，探讨气温对不同城市死亡率影响的滞后效应特点。方法分别从中国CDC和气象网站获取北京、天津、上海、南京、长沙5个城市的人口和气象数据，采用R2．12．0软件的分布滞后非线性模型（OLNM）软件包将数据资料代人后进行分析。分析气象因素对死亡率的影响，累积效应，计算RR值。结果北京、天津属于温带城市，上海、南京和长沙均属于亚热带季风气候。在气候与死亡效应的关系中，在日平均气温达30．0℃、滞后0d时，RR值最高，南京（1．31，95％CI：1．21—1．41）和长沙市（1．25，95％CI：1．13～1．39）的死亡率的RR值要高于北京（1．18，95％CI：1．12～1．25）、天津（1．18，95％CI：1．10～1．26）和上海市（1．15，95％CI：1．06～1．24）。在总的滞后时间（30d）内，各个城市的最低日平均气温对天津、长沙、北京、南京和上海市的死亡率的RR值分别为3．41，95％CI：1．60～7．27、2．15，95％CI：1．11～4．15、2．24，95％C1：1．12～4．48、2．80，95％C1：1．75～4．48、1．53，95％C1：1．12～2．03。日平均温度对死亡率的累积RR值有较为明显的“U”形曲线。极端高温和最高日平均气温对死亡率的相对危险度在滞后0～1d时RR值均〉1；而低日平均气温在滞后2d后对死亡率有明显影响。结论高温对死亡率具有急性效应的影响，而低温影响的滞后时间较长。极端低温和最低日平均气温对北京、天津地区城市居民死亡率的影响较大，而极端高温和日最高平均气温则对上海、南京、长沙的影响较大。

Objective To study the characteristics of the effect of different temperatures on mortality of different cities through analyzing the relationship between mortality and meteorology of five Chinese cities. Methods We get the demography and climate data of Beijing, Tianjin, Shanghai, Nanjing and Changsha cities from National Center of Disease Control and Prevention and Climate net respectively. Then we applied the R software and Distributed Lag Non-linear Models （DLNM） package to analyze our data and find the nonlinear and lag effects on mortality using DLNM. Results The city of Beijing and Tianjin are located in the temperate zone. And the climate of Shanghai, Nanjing, Changsha belong to subtropical monsoon climate. When the daily mean temperature arrived 30 ~C and on lag 0 day, the values of relative risk of effect of high mean temperature on mortality in Nanjing （ l. 31,95 % CI： l. 21 - 1.41 ）and Changsha （ 1.25,95% CI： 1.13 - 1.39 ） are larger than that in Beijing（ 1.18,95% CI： 1.12 - 1.25 ） ,Tianjin （ 1.18,95 % CI： 1.10 - 1.26） and Shanghai （ 1.15,95 % CI： 1.06 - 1.24 ）. While the relative risk of effect of low mean temperature on mortality is lower and lasts for a longer lag time. During the whole lag time, the relative risk of effect of the lowest daily mean temperature of each city on mortality in Tianjin, Changsha, Beijing, Nanjing, and Shanghai is 3.41,95% CI： 1.60 - 7.27,2. 15,95% CI： 1.11 - 4. 15,2. 24,95% CI： 1.12 -4. 48,2.80,95% CI： 1.75 - 4.48,1.53,95% CI： 1.12 - 2. 03, respectively. The cumulative effect of mean temperature on mortality appears like a U-shape. When on lag 0 - 1 day, the value of relative risk of effect of extremely high temperature and the highest mean temperature on mortality is larger than 1. While the effect of low temperature on mortality becomes obvious after lag 2 days. Conclusion Depending on this research, extremely low temperature and the lowest mean temperature has a more obvious impact on mortality in the northern area than in the south. Extremely high temperature and the highest daily mean temperature is on the contrary. Meanwhile, different temperatures have different impacts on mortality in the same city：high temperature has an acute impact while there is a longer lag time in low temperature.