基于死亡数据用DLNM构建气象健康指数

Based on Death Information and Using DLNM to Develop a Meteorological Health Index

目的探索建立气象健康指数,预警提示气温导致的健康风险。方法基于2011—2015年广州市居民死亡数据,用分布滞后非线性模型(DLNM)分析气温与居民非意外日死亡数的关系;对气温导致的健康风险进行分级,用不同风险等级对应的温度范围开展健康风险预报。结果 2011—2015年广州市的日平均气温为21.7℃,日相对湿度为78.6%,日均居民死亡为132例。日平均温度与居民死亡之间暴露—反应曲线呈U形,是非线性关系,低温和高温都会引起死亡风险,高温对人群死亡的影响是急性效应,低温产生的效应会比较延迟,但持续时间长。舒适温度区间为19.4℃～25.4℃,其所对应的RR值为1.02～1.03。冷效应RR值低中高三个风险等级分别为1.03～1.17,1.17～1.31,1.32～1.46,其对应的温度区间依次为12.9℃～19.3℃,8.5℃～12.8℃,4.8℃～8.4℃;热效应RR值低中高三个风险等级分别为1.02～1.08,1.08～1.13,1.13～1.19,其对应的温度区间依次为25.5℃～27.8℃,27.9℃～29.5℃,29.6℃～31.1℃。结论由DLNM评估气象条件与居民日死亡人数的关系可反映气温引起的健康风险,将健康风险等间隔分级预警较合理,可对开展气象健康指数预报服务提供理论支持。

Objectives Exploring and establishing a meteorological health index to warn the risk of temperature on human health. Methods Using a Distributed Lag Non-linear Model（DLNM） to estimate the effect of temperature on non-accidental mortality in Guangzhou during 2011—2015. Classifying human health risks caused by temperature and estimating the risk level correspond to different temperature ranges. Results During the period of study（1 January 2011-31 December 2015）, the mean temperature was 21.7℃, relative humidity was 78.6%, and the average number of daily non-accidental death was 132 cases. There was a U-shaped relationship between temperature and mortality in Guangzhou. Both low and high temperatures could cause an increase of death number, the influence of hot temperature was acute while the influence of cold was delayed by 3 days and persisted for longer days. The range of comfortable temperature was 19.4℃～25.4℃, its corresponding RR was 1.02～1.03. The relative risk（RR） of cold effect could be divided into three grades, 1.03～1.17, 1.17～1.31, and 1.32～1.46, and their corresponding temperature ranges were 12.9℃～19.3℃, 8.5℃～12.8℃ and 4.8℃～8.4℃, respectively. The relative risk（RR） of hot effect could be divided into three grades, 1.02～1.08, 1.08～1.13 and 1.13～1.19, and their corresponding temperature ranges were 25.5℃～27.8℃, 27.9℃～29.5℃ and 29.6℃～31.1℃, respectively. Conclusions The association between meteorological condition and daily mortality assessed by DLNM could reflect the health risk caused by temperature. It is reasonable to classify the health risk at equal intervals for early warning, which could provide a theoretical support for developing a meteorological health index forecasting service.