不同湿度下日均气温对居民非意外死亡的滞后影响

Lag-effect of Daily Mean Temperature on Daily Deaths under Different Relative Humidity

目的通过构建分布滞后非线性模型(DLNM)探究不同湿度分层下日均气温对成都市居民每日非意外死亡人数的影响。方法收集2011~2014年成都市每日居民死亡数、气象和大气污染物数据,采用quasiPoisson回归结合DLNM分析相对湿度对日均气温与每日居民非意外死亡数关系的修饰作用,并比较不同湿度下冷、热效应的不同。结果成都市日均气温-死亡效应在不同湿度下冷、热效应的差异均有所不同。高湿度下的具有明显的冷效应且当天达到最大值,持续时间约为15天;未观察到明显的热效应。中湿度下的冷效应在滞后0~1天出现,2天左右达到最大值,效应持续时间约为10天;而热效应显现出较弱的急性效应和收获效应。低湿度下的冷效应在滞后1天出现,3天达到最大值,效应持续时间约为5天,此时热效应表现较为急骤,当天最高,然后迅速下降并表现出一定的收获效应。结论成都市的气温-死亡效应主要表现为冷效应,且湿度越高冷效应达到峰值时间越短,效应持续时间越长;而热效应不明显,但随着湿度的降低,也可以表现出较为明显的急性效应和收获效应。

Objective To estimate how relative humidity modify the effect of daily mean temperature on mortality in Chengdu by using distributed lag non-linear model（DLNM）. Methods The data of daily mortality, meteorology and air pollution in Chengdu during 2011-2014 were collected. The relationship of daily mean temperature and daily death counts in different relative humidity（RH） layers was analyzed with quasiPoisson regression combining DLNM approach. Results The difference of cold and heat effect varies with variation of RH. Cold effect with high RH was strong touching the top at the first day and could last for about 15 days while heat effect was not observed obviously; in the medium RH, cold effect appeared at lag 0-1 day touching the top at about 2 nd day and lasted for approximately 10 days while heat effect was weak and showed quickly; in the layer of low RH, cold effect appeared at lag 1 day getting the peak at 3 rd day and lasted for about 5 days while heat effect appeared strongly touching the maximum at the first day and decreased quickly. Conclusion Temperature-death effect showed mainly as cold effect in Chengdu and with RH increasing, the RR of cold effect on daily death counts was higher and lasted longer. Heat effect was not obvious but still appeared strongly and quickly when RH was low.