Has the impact of temperature on mortality really decreased over time?

Many studies have reported that the temperature effect on mortality has decreased over time. However, most of those studies did not consider lag times longer than 10 days, which is frequently used to explore its effect net out compensatory effect (harvesting) and lag effects. We sought to examine the temporal variation of the temperature effect on mortality, considering both a lag effect and mortality displacement. Time-series analysis was conducted with lag of temperature up to 21 days on all-cause, cardiovascular, cerebrovascular, and respiratory deaths. We applied a series of time-windows, 8 years long, with which we compared the oldest to more recent intervals and took consecutive annual variation, excluding an interannual harvesting effect. At the 99th percentile (29°C), relative to the 90th percentile (25°C), we found a decreasing trend of heat effect on concurrent days whereas the risk of cardiovascular deaths increased over time. Cumulative risks of deaths increased recently except for respiratory disease. At the 10th percentile (-1°C) relative to the 25th percentile (4°C), cumulative cold effects on cardiovascular and respiratory mortality have emerged recently. Our study showed differences in the temporal variation in the temperature effect on mortality at concurrent day and in cumulative term. It is suggested that the time-varying nature of the temperature-mortality relationship depends not only on suggested factors, such as improvements in technology and infrastructure, and human physiological acclimatization, but also mortality displacement and lagged effects. Further studies on its complex nature are needed to provide relevant evidence for public health policy making.