Lee Won Kyung, Lee Hye Ah, Lim Youn Hee, Park Hyesook
Department of Social and Preventive Medicine, Inha University School of Medicine, 27 Inhang-Ro, Jung Gu, Incheon, Republic of Korea.
Department of Preventive Medicine, School of Medicine, Ewha Womans University, 1071, Anyangcheon-ro, Yangcheon-ku, Seoul, 158-710, Republic of Korea.
Int J Biometeorol. 2016 May;60(5):719-26. doi: 10.1007/s00484-015-1067-x. Epub 2015 Oct 1.
A heat wave could increase mortality owing to high temperature. However, little is known about the added (duration) effect of heat wave from the prolonged period of high temperature on mortality and different effect sizes depending on the definition of heat waves and models. A distributed lag non-linear model with a quasi-Poisson distribution was used to evaluate the added effect of heat wave on mortality after adjusting for long-term and intra-seasonal trends and apparent temperature. We evaluated the cumulative relative risk of the added wave effect on mortality on lag days 0-30. The models were constructed using nine definitions of heat wave and two relationships (cubic spline and linear threshold model) between temperature and mortality to leave out the high temperature effect. Further, we performed sensitivity analysis to evaluate the changes in the effect of heat wave on mortality according to the different degrees of freedom for time trend and cubic spline of temperature. We found that heat wave had the added effect from the prolonged period of high temperature on mortality and it was considerable in the aspect of cumulative risk because of the lagged influence. When heat wave was defined with a threshold of 98th percentile temperature and ≥2, 3, and 4 consecutive days, mortality increased by 14.8 % (7.5-22.6, 95 % confidence interval (CI)), 18.1 % (10.8-26.0, 95 % CI), 18.1 % (10.7-25.9, 95 % CI), respectively, in cubic spline model. When it came to the definitions of 90th and 95th percentile, the risk increase in mortality declined to 3.7-5.8 % and 8.6-11.3 %, respectively. This effect was robust to the flexibility of the model for temperature and time trend, while the definitions of a heat wave were critical in estimating its relationship with mortality. This finding could help deepen our understanding and quantifying of the relationship between heat wave and mortality and select an appropriate definition of heat wave and temperature model in the future studies.
热浪可能会因高温而增加死亡率。然而，对于热浪中高温持续时间对死亡率的附加（持续时间）影响以及根据热浪定义和模型的不同效应大小，我们所知甚少。我们使用具有拟泊松分布的分布滞后非线性模型，在调整长期和季节内趋势以及体感温度后，评估热浪对死亡率的附加影响。我们评估了滞后0至30天热浪对死亡率附加波效应的累积相对风险。使用九种热浪定义以及温度与死亡率之间的两种关系（三次样条和线性阈值模型）构建模型，以排除高温效应。此外，我们进行了敏感性分析，以评估根据时间趋势和温度三次样条的不同自由度，热浪对死亡率影响的变化。我们发现，热浪中高温持续时间对死亡率有附加影响，并且由于滞后影响，在累积风险方面相当可观。在三次样条模型中，当热浪定义为第98百分位数温度阈值且连续≥2、3和4天时，死亡率分别增加14.8%（7.5 - 22.6，95%置信区间（CI））、18.1%（10.8 - 26.0，95%CI）、18.1%（10.7 - 25.9，95%CI）。当涉及第90和第95百分位数定义时，死亡率风险增加分别降至3.7 - 5.8%和8.6 - 11.3%。这种效应对于温度和时间趋势模型的灵活性具有稳健性，而热浪的定义对于估计其与死亡率的关系至关重要。这一发现有助于加深我们对热浪与死亡率关系的理解和量化，并在未来研究中选择合适的热浪定义和温度模型。
