Effect of temperature and precipitation on salmonellosis cases in South-East Queensland, Australia: an observational study.

OBJECTIVE

Foodborne illnesses in Australia, including salmonellosis, are estimated to cost over $A1.25 billion annually. The weather has been identified as being influential on salmonellosis incidence, as cases increase during summer, however time series modelling of salmonellosis is challenging because outbreaks cause strong autocorrelation. This study assesses whether switching models is an improved method of estimating weather-salmonellosis associations.

DESIGN

We analysed weather and salmonellosis in South-East Queensland between 2004 and 2013 using 2 common regression models and a switching model, each with 21-day lags for temperature and precipitation.

RESULTS

The switching model best fit the data, as judged by its substantial improvement in deviance information criterion over the regression models, less autocorrelated residuals and control of seasonality. The switching model estimated a 5 °C increase in mean temperature and 10 mm precipitation were associated with increases in salmonellosis cases of 45.4% (95% CrI 40.4%, 50.5%) and 24.1% (95% CrI 17.0%, 31.6%), respectively.

CONCLUSIONS

Switching models improve on traditional time series models in quantifying weather-salmonellosis associations. A better understanding of how temperature and precipitation influence salmonellosis may identify where interventions can be made to lower the health and economic costs of salmonellosis.