Modification of Temperature-Mortality Association in Summer by the Previous Winter Mortality in South Korea

Abstract

ISEE-0388 Background and Objective: Numerous studies in recent years have reported association between summer temperature and deaths in cities. However, in summer when a greater pool of vulnerable individuals leave, the heat-related mortality is little known. We aimed to explore how sensitive the heat-related mortality in summer is to mortality in the previous winter in 3 major cities, in South Korea. Methods: In the study period from 1992–2007, summer and winter were defined as June-August and December-February, respectively. The statistical analysis was conducted in two steps. First, to investigate thresholds of the temperature-mortality curves in summer, we fitted piecewise log-linear models using AIC for the model-fitting criteria. Second, to investigate dependency of heat-mortality effects above the threshold on previous winter mortality, we incorporated indicators of mortality in winter as effect modifiers in the first step model. Results: Estimated thresholds for ≥65 year old age group with two day (lag 0 and 1) mean temperature (27.4°C, 27.3°C and 23.2°C for Seoul, Daegu, and Incheon, respectively) were lower than those for all age groups (27.9°C, 28.1°C and 25.5°C, respectively). For summer mortality of all age group, a 1°C increment in temperature above the thresholds was associated with an increase of 10.57%, 8.55%, and 6.04% for low strata, 4.85%, 4.00%, and 2.63% for high strata of winter mortality in Seoul, Daegu, and Incheon, respectively (P-values for heterogeneity: 0.0081, 0.0142, and 0.0378, respectively). Those for ≥65 year old age group were 8.85%, 8.36%, and 3.88% for low strata, 7.75%, 5.14%, and 2.83% for high strata, respectively (P-values for heterogeneity: 0.5846, 0.0502, and 0.2651, respectively). Conclusion: Our results show that low mortality in winter leads to higher mortality in the following summer. It suggests that the yearly variation of heat effect in summer should be reconsidered in public health program, especially in heat warning system.