Lee Mihye, Nordio Francesco, Zanobetti Antonella, Kinney Patrick, Vautard Robert, Schwartz Joel
Department of Environmental Health, Exposure, Epidemiology, and Risk Program, Harvard School of Public Health, 401 Park Drive, Landmark Center West 4th fl, Boston, MA 02215, USA.
Environ Health. 2014 Oct 28;13:89. doi: 10.1186/1476-069X-13-89.
Climate change has increased the days of unseasonal temperature. Although many studies have examined the association between temperature and mortality, few have examined the timing of exposure where whether this association varies depending on the exposure month even at the same temperature. Therefore, we investigated monthly differences in the effects of temperature on mortality in a study comprising a wide range of weather and years, and we also investigated heterogeneity among regions.
We analyzed 38,005,616 deaths from 148 cities in the U.S. from 1973 through 2006. We fit city specific Poisson regressions to examine the effect of temperature on mortality separately for each month of the year, using penalized splines. We used cluster analysis to group cities with similar weather patterns, and combined results across cities within clusters using meta-smoothing.
There was substantial variation in the effects of the same temperature by month. Heat effects were larger in the spring and early summer and cold effects were larger in late fall. In addition, heat effects were larger in clusters where high temperatures were less common, and vice versa for cold effects.
The effects of a given temperature on mortality vary spatially and temporally based on how unusual it is for that time and location. This suggests changes in variability of temperature may be more important for health as climate changes than changes of mean temperature. More emphasis should be placed on warnings targeted to early heat/cold temperature for the season or month rather than focusing only on the extremes.
气候变化增加了非季节性温度出现的天数。尽管许多研究已经考察了温度与死亡率之间的关联，但很少有研究考察暴露时间，即这种关联是否会因暴露月份不同而有所变化，即使温度相同。因此，我们在一项涵盖广泛天气情况和年份的研究中，调查了温度对死亡率影响的月度差异，并且我们还调查了不同地区之间的异质性。
我们分析了1973年至2006年美国148个城市的38005616例死亡病例。我们采用惩罚样条法，针对一年中的每个月份分别拟合特定城市的泊松回归模型，以检验温度对死亡率的影响。我们使用聚类分析对天气模式相似的城市进行分组，并通过元平滑法合并各聚类内城市的结果。
相同温度的影响在不同月份存在显著差异。春季和初夏的热效应更大，而深秋的冷效应更大。此外，在高温不太常见的聚类中热效应更大，反之冷效应在低温不太常见的聚类中更大。
给定温度对死亡率的影响会根据该时间和地点的异常程度在空间和时间上有所变化。这表明随着气候变化，温度变异性的变化可能对健康比平均温度的变化更为重要。应更加重视针对季节或月份早期炎热/寒冷温度的预警，而不仅仅关注极端情况。
