The association between precipitation and the spread of shiga toxin-producing Escherichia coli infection in Lombardy, North of Italy

The HUS (Hemolytic-Uremic Syndrome) is a disease that mainly affects children, it is a complication of an infection by shiga toxin produced by Eschirichia-coli (STEC) infection. It shows a seasonal nature with lower incidence in the cold period, it is endemic in central Europe and North America but fairly rare. In the high incidence period, from April to October, cases appear in time related clusters, usually not epidemic, in fact they are unrelated for spatially distance or caused by different serotypes. It suggests the role of the weather conditions  as a hot gun.  Attempting to predict by weather conditions the temporal and spatial distribution of the illness could prevent its severity and reduce the damage like kidney failure, disability or worse. 

Past climate study (Acquaotta, et al. 2017) suggests a correlation between the heat waves and infection clusters, for example three or more very hot consecutive days increase the risk of disease. Also the epidemiological study of Biggeri and his Collaborations (personal communication) have analyzed the spatial aggregation of illness in Lombardy (North Italian region) to show the areas with a higher or lower risk of infection. 

In this research the study area is the Lombardy plain with 249 selected cases from 2010 to 2020, 11 years. Two are the main goals: the first is to perform a climate characterisation of the high risk zone in Lombardy, highlighted by Biggeri and his Collaborations. The second is to identify the relationship between infection and extreme precipitation events. 

To analyse the climate peculiarities of Lombardy we used 88 weather series from ARPA (Regional Environmental Protection Agency) Lombardy with an homogeneous distribution in the area. On the daily weather series a quality control was carried out to identify some error in the data. Then a hierarchical cluster analysis by mean of the Ward.d2 method was applied to identify the climate areas. On the daily data, climate indices created by the Expert Team on Climate Change Detection and Indices (ETCCDI) were calculated: daily precipitation intensity (SDII), consecutive dry days (CDD), very wet days (R95p) and standard precipitation index (SPI) to show the drought period.

The distributed lag non-linear model (DLNM) was used to examine the relationship between precipitation series and daily illness during January 2010 to December 2020 (Bai et al. 2014;) with a maximum lag of 30 days in order to ensure a greater coverage. The infection dataset was also compared with four climate indices in order to highlight. The linear regression between indices and infections was then calculated. The trends were computed using the TheilSen approach (TSA) (Sen 1968).