Already in ancient Greece, Hippocrates postulated that disease showed a seasonal pattern characterised by excess winter mortality. Since then, several studies have confirmed this finding, and it was generally accepted that the increase in winter mortality was mostly due to respiratory infections and seasonal influenza. More recently, it was shown that cardiovascular disease (CVD) mortality also displayed such seasonality, and that the magnitude of the seasonal effect increased from the poles to the equator.1

The recent study by Yang et al2 assessed CVD mortality attributable to ambient temperature using daily data from 15 cities in China for years 2007–2013, including nearly two million CVD deaths. A high temperature variability between and within cities can be observed (figure 1). They used sophisticated statistical methodology to account for the complex temperature–mortality relationship; first, distributed lag non-linear models combined with quasi-Poisson regression to obtain city-specific estimates, taking into account temperature, relative humidity and atmospheric pressure; then, a meta-analysis to obtain the pooled estimates. The results confirm the winter excess mortality as reported by the Eurowinter3 and other4 groups, but they show that the magnitude of ambient temperature …