Kovats, S; Edwards, S; Hajat, S; Ebi, K
Epidemiology: September 2003

Objective: To describe and quantify the relationship between environmental temperature and reported cases of disease caused by Salmonella in 7 European countries.

Data: Laboratory confirmed cases of salmonella were obtained from surveillance centres in England and Wales (years 1989-97), Scotland (90-97), Netherlands (84-2001), Czech Republic (93-01), Poland (2000-02), Denmark (91-01), Switzerland (90-00) and Spain (94-00). Cases where infection was acquired abroad were removed, validated temperature series was constructed for each country.

Methods: Series of the total weekly number of disease cases and the weekly average of mean daily temperature were generated. If data were available, series were also created for 3 age groups, and for different salmonella species. Public holidays and any trend in disease were controlled for. Fourier terms (up to the sixth harmonic) were added to each model to control for any mid- to long-term annual and seasonal patterns (i.e. cycles greater than 2 months). A threshold for each country was identified by comparison of maximum likelihood of models with different threshold values of the temperature measure. The temperature-disease relationship was quantified by adding to the model a linear term for temperature above the threshold value. To take into account the long lag time between temperature exposures and reported cases, the temperature measure used in the final model was an average value of temperatures in the 0-9 weeks before the case. After allowance for overdispersion, Poisson regression was used to estimate the percentage change in the number of cases associated with a one degree increase in temperature above the threshold value identified for each country.

Results: We found a significant and consistent linear association between environmental temperature and the number of reported cases of salmonella. The adult age group (15-64) and S. Enteritidis showed the greatest sensitivity to temperature effects. The greatest effect of temperature is observed one week before the onset of illness.

Conclusions: These results are consistent with previous studies and the current understanding of the effect of temperature on salmonella transmission.

Threshold value and percentage increase (with confidence intervals) per degree centigrade above the threshold

(C) 2003 Lippincott Williams & Wilkins, Inc.