Share this article on:


Biggeri, A*; Baccini, M*; Michelozzi, P; Kirchmayer, U; Analitis, A; Katsouyanni, K; Bisanti, L§; Cadum, E; Tiittanen, P; Medina, S**; Paldy, A††

Program and Abstracts: The Seventeenth Conference of the International Society for Environmental Epidemiology (ISEE): Abstracts

*Department of Statistics, University of Florence, Italy; †Department of Epidemiology, Local Health Authority Asl/RmE; ‡Department of Hygiene & Epidemiology, University of Athens Medical School; §Department of Epidemiology, Local Health Authority Milan; ¶Regional Environmental Protection Agency, Piedmont; ∥Unit of Environmental Epidemiology, National Public Health Institute; **Departement Sante-Environnement Department, Institut de Veille Sanitaire, Paris, France; ††Dep. of Biological Monitoring Jozsef Fodor National Center of Public Health, Hungary.


Back to Top | Article Outline


Epidemiological studies indicate that exposure to extreme meteorological conditions is associated with an increase in mortality. We evaluated the health effects of apparent temperature (a combination of the dry bulb and dew point temperatures) during warm season in 15 European cities participating to the PHEWE project (Athens, Barcelona, Budapest, Helsinki, Dublin, Ljubljana, London, Milan, Paris, Prague, Rome, Stockholm, Turin, Valencia, Zurich). The aim of this work was to provide city-specific summary measures of the effect of high apparent temperatures on mortality in order to make straightforward comparison among cities.

Back to Top | Article Outline


All cities provided daily counts of deaths for all causes and for cardiovascular and respiratory causes and 3-hourly meteorological data (namely temperature and humidity) retrieved from the nearest airport weather station. 3-hourly apparent temperature time series were computed for each city and daily maximum value was used as exposure variable. Data on several confounders including other meteorological variables and air pollution variables were also provided. A large variability in climatic characteristics and in mortality was observed in the cities involved.

The analysis considered only data from warm seasons (April to September). Following the PHEWE protocol, a common GEE model was specified on daily mortality data from each city according to a Poisson probability model. Independence was assumed among separate summers, while an AR(1) covariance structure was specified within summer. We adjusted for long term time trend, month, day of the week, holiday, air pollution concentration, wind speed and sea level pressure, including in the model appropriate parametric terms. We modelled the apparent temperature effect by a “V” shaped function and we used the estimated slope above the breakpoint as an indicator of high temperature exposure effect. Different definitions of breakpoints were compared: unique pre-defined breakpoint for all the cities; city-specific breakpoints estimated by flexible regression spline approach; city-specific breakpoints estimated by maximum likelihood.

Back to Top | Article Outline

Results and discussion:

Apparent Temperature (AT: maximum, mean, minimum) Linear Effects above threshold were very consistent among cities (for maximum AT and lag 0–3 days, the rate ratio for one degree in AT range between 0.02; 0.09). Distributed lag analysis (up to 40 days) showed an excess risk concentrated in the first days and limited evidence of harvesting at days 5–20. The AT effect at lag-0 range between 0.01 ; 0.04 (rate ratios for one degree in AT above threshold). Breakpoints values appeared to vary among cities, ranging approximately from 20 (north European cities) to 30 (Athens, Rome and Milan) units of maximum apparent temperature. We found that the slope estimate increased as the turning point estimate increased, indicating possible non linearity of the effect.

Back to Top | Article Outline


The PHEWE project is supported by the European Commission, DG Research within the 5th Framework programme, contract N°QLK4-CT-2001-00152.

© 2005 Lippincott Williams & Wilkins, Inc.