Background: A heatwave occurred in France in August 2003, with an accompanying excess of all-cause mortality. This study quantifies this excess mortality and investigates a possible harvesting effect in the few weeks after the heatwave.
Methods: A time-series study using a Poisson regression model with regression splines to control for nonlinear confounders was used to analyze the correlation between heatwave variable and mortality in 9 French cities.
Results: After controlling for long-term and seasonal time trends and the usual effects of temperature and air pollution, we estimated that 3,096 extra deaths resulted from the heatwave. The maximum daily relative risk of mortality during the heatwave (compared with expected deaths at that time of year) ranged from 1.16 in Le Havre to 5.00 in Paris. There was little evidence of mortality displacement in the few weeks after the heatwave, with an estimated deficit of 253 deaths at the end of the period.
Conclusions: The heatwave in France during August 2003 was associated with a large increase in the number of deaths. The impact estimated using a time-series design was consistent with crude previous estimates of the impact of the heatwave. This finding suggests that neither air pollution nor long-term and seasonal trends confounded previous estimates. There was no evidence to suggest that the extras deaths associated with the heatwave were simply brought forward in time.