Space-Time Covariation of Mortality with Temperature: A Systematic Study of Deaths in France, 1968-2009

BACKGROUND: The temperature-mortality relationship has repeatedly been found, mostly in large cities, to be U/J-shaped, with higher minimum mortality temperature (MMT) at low latitudes being interpreted as indicating human adaptation to climate. OBJECTIVES: Our aim was to partition space with a high-resolution grid to assess the temperature-mortality relationship in a territory with wide climate diversity, over a period with notable climate warming. METHODS: The 16,487,668 death certificates of persons > 65 years of age who died of natural causes in continental France (1968-2009) were analyzed. A 30-km x 30-km grid was placed over the map of France. Generalized additive model regression was used to assess the temperature-mortality relationship for each grid square, and extract the MMT and the RM25 and RM25/18 (respectively, the ratios of mortality at 25 degrees C/MMT and 25 degrees C/18 degrees C). Three periods were considered: 1968-1981 (P1), 1982-1995 (P2), and 1996-2009 (P3). RESULTS: All temperature-mortality curves computed over the 42-year period were U/J-shaped. MMT and mean summer temperature were strongly correlated. Mean MMT increased from 17.5 degrees C for P1 to 17.8 degrees C for P2 and to 18.2 degrees C for P3 and paralleled the summer temperature increase observed between P1 and P3. The temporal MMT rise was below that expected from the geographic analysis. The RM25/18 ratio of mortality at 25 degrees C versus that at 18 degrees C declined significantly (p = 5 x 10(-5)) as warming increased: 18% for P1, 16% for P2, and 15% for P3. CONCLUSIONS: Results of this spatiotemporal analysis indicated some human adaptation to climate change, even in rural areas.