The study analysed over 74 million (74225200) deaths between 1985 and 2012 in 13 countries with a wide range of climates, from cold to subtropical. Data on daily average temperature, death rates, and confounding variables (eg, humidity and air pollution) were used to calculate the temperature of minimum mortality (the optimal temperature), and to quantify total deaths due to non-optimal ambient temperature in each location. The researchers then estimated the relative contributions of heat and cold, from moderate to extreme temperatures.

Around 7.71% of all deaths were caused by non-optimal temperatures, with substantial differences between countries, ranging from around 3% in Thailand, Brazil, and Sweden to about 11% in China, Italy, and Japan. Cold was responsible for the majority of these deaths (7.29% of all deaths), while just 0.42% of all deaths were attributable to heat.

The study also found that extreme temperatures were responsible for less than 1% of all deaths, while mildly sub-optimal temperatures accounted for around 7% of all deaths–with most (6.66% of all deaths) related to moderate cold.

According to Dr Gasparrini, "Current public-health policies focus almost exclusively on minimising the health consequences of heat waves. Our findings suggest that these measures need to be refocused and extended to take account of a whole range of effects associated with temperature." 

Writing in a linked Comment, Keith Dear and Zhan Wang from Duke Kunshan University, Kunshan, Jiangsu, China say, "Factors such as susceptibility or resilience have not been included in the analysis, including socioeconomic status, age, and confounding air pollutants…Since high or low temperatures affect susceptible groups such as unwell, young, and elderly people the most, attempts to mitigate the risk associated with temperature would benefit from in-depth studies of the interaction between attributable mortality and socioeconomic factors, to avoid adverse policy outcomes and achieve effective adaptation."