Figure 1. Temperatures recorded during the 2003 summer in France

Source: Cité des sciences et de l’industrie from Météo France

Of the meteorological stations compared by Météo France [2], it appears that in Madrid, maximal temperatures were close to those seen in Paris, but that minimal temperatures remained five to six degrees Centigrade lower at the peak of the heat wave (Figure 2).

Figure 2. Minimal and maximal temperatures recorded from 1 to 15 August 2003 in eight European cities

In Spain, only Andalusia experienced a more severe heat wave than in France. However, different populations in Europe have different thresholds for high temperatures: Andalusians, for example, have a biocritical threshold of 41°C, while Belgians’ begin to feel ill at 27.5°C [3]. In Lisbon, the minimal temperatures observed in the first fortnight of August were close to those seen in Paris, despite a higher variability. However, maximal temperatures were slightly lower, but also with a higher variability. As the difference between minimal and maximal temperatures is often attributable to a drop in temperatures during the night, it seems that the persistence of high night temperatures is more responsible for the deadly nature of a heat wave than maximal temperatures, although the interference of pollution factors cannot be discarded.

Calculation of excess mortality
Excess mortality attributed to the heat wave was measured by the author by comparing the number of deaths corrected for seasonal variations that occurred in the whole of August to the moving average number of deaths, measured over 25 months and focused on the month of August 2003. This produces an over-estimate, since the total variation is attributed to the heat wave. To take into account the fact that the duration of the heat wave varied widely from one country to another [4], the same analysis was performed for the July-September 2003 period in 27 European countries.

Results
This method produced estimates close to those presented in the special issue of Eurosurveillance [5] (Table 1).

Table 1. Total excess deaths from the heat wave of August 2003 across Europe

Only positive values exceeding 50 deaths are presented in the table. For nine other countries in the analysis (Denmark, Greece, Ireland, Norway, Sweden, Hungary, Macedonia, Poland, Slovakia, the Czech Republic and Romania), the calculation yields a slightly positive value and even widely negative ones like in Poland (-805) or in Greece (-425), certainly a sign of the importance random variations can have. Of the 27 countries in question, only seven seem to have been significantly affected by the heat wave in the first two weeks of August, with an excess mortality above 9% (Table 2): Switzerland, Germany, Italy, Luxembourg, Spain, Portugal and France. With the exception of Luxembourg, these countries also experienced the highest mortality rates. Other countries were also affected at a lesser level: Slovenia (6%), Croatia, England and Wales, Belgium and the Netherlands (4%). Everywhere else, the effect of the heat wave on mortality was very low to non-existent.

Table 2. Average excess mortality recorded in Europe during the summer 2003 heat wave (in %)

Applying calculations to the whole summer period (July-September), excess mortality yields identical results. Although France remains the most affected country, the level of excess mortality is nevertheless considerably reduced: a similar difference is seen with Portugal’s figures. However, the excess mortality is not decreased by nearly as much in Italy when applied to the longer period, suggesting that that country had a longer heat wave or one that had an impact over a longer period. The geographical distribution of excess mortality covers the regions that experienced the most days with temperatures above 35° (Figure 3).

Figure 3. Number of days with maximal temperatures above 35°c and 40°C

Source: Météo France

This map illustrates the clear difference between the situations in Luxembourg and Belgium: Brussels was not affected by the heat wave, whereas Luxembourg experienced temperatures above 34°C for 10 days. The 2003 heat wave, with its fatal outcomes, therefore mainly affected southern Europe (Iberian Peninsula, France and Italy). Luxembourg, as Germany and Switzerland, was affected through small extensions of the meteorological phenomenon that centered over France. In Luxembourg, the effects were quite severe since the whole county was concerned, due mainly to its small surface. In the east, Slovenia, Croatia and Bosnia were also partly affected by the heat wave.

This letter draws heavily on the chapter dedicated to the 2003 heat wave in Sardon JP, Evolution démographique récente des pays développés, Population, n°3, 2006, pp.225-300.