Difference in temporal variation of temperature-related mortality risk in seven major South Korean cities spanning 1998-2013.

Temporal variation of temperature-related mortality risk is an important issue in climate change era. However, difference in this temporal variation across cities in South Korea remains unclear. The aim of this study was to explore whether temporal variation might differ spatially across seven metropolitan cities of Korea during the period of 1998-2013. We estimated cumulative associations between temperature (up to previous 14 days of exposure) and all-cause mortality, and compared cumulative associations between the first eight years (1998-2005) and the last eight years (2006-2013). Multivariate meta-regression analysis was performed to investigate what factors might be associated with spatial and temporal variation in cumulative associations. We found that Busan, Daegu, and Gwangju experienced decrease in heat effect from 1998-2005 to 2006-2013, while Incheon experienced increase in heat effect. By comparing mortality risk at 99th percentile of temperature to mortality risk at minimum mortality temperature, percentage increase of mortality risk changed from 4.8% (95% CI: -1.3, 11.3) to 0.4% (95% CI: -6.0, 7.4) in Busan, from 17.2% (95% CI: 10.2, 24.7) to 4.0% (95% CI: -1.4, 9.8) in Daegu, from 20.3% (95% CI: 11.5, 29.7) to 2.2% (95% CI: -3.5, 8.3) in Gwangju, and from 3.5% (95% CI: 0.2, 6.8) to 7.9% (95% CI: 5.0, 10.9) in Incheon, respectively. Change in average temperature from 1998-2005 to 2006-2013 was negatively associated with change in heat effect even though average temperature in most of the cities fluctuated over time. We also found that all seven cities had decrease in effect of moderate cold temperature from 1998-2005 to 2006-2013. Such decrease was associated with improvement in medical resources. Results of this study suggest that plans for adaptation to temperature-related risks should differ across populations because adaptation to temperature varies across populations and within the same population over different time.