Spatiotemporal analysis of associations between flood hydrometeorology
and gastroenteric infection: The Winter 2015-2016 flood event in the
Republic of Ireland
Abstract
While the infrastructural damage and subsequent costs associated with
flood events have, and will continue to receive widespread attention,
less attention is given to the adverse human health effects of these
events. This is particularly significant in the ROI, which is
characterised by the highest crude incidence rates of verotoxigenic
E. coli (VTEC) enteritis and cryptosporidiosis in Europe.
Accordingly, weekly infection incidence from July 2015 to June 2016 were
employed in concurrence with weekly time-series of antecedent
hydrometeorological parameters (rainfall, surface water discharge and
groundwater level), and high-resolution flood risk mapping. An ensemble
of statistical and time-series approaches were employed to quantify the
influence and timing of flood hydrometeorology on infections. Seasonal
decomposition identified a high residual infection peak (excluding
seasonal pattern) during April 2016, with space-timing scanning used to
identify the location, size and temporal extent of excess infection
clusters. Excess cases of VTEC enteritis were geographically associated
with the Shannon basin, while cryptosporidiosis excess was nationwide.
Generalised linear modelling indicates that areas with a surface water
body exhibited significantly higher incidence rates for both infections
(OR 1.225 - 1.363 p <0.001). Non-parametric ranking identified
a clear association between hydrometeorology and infection incidence,
with lagged associations from 16-20 weeks proving particularly strong,
thus indicating a link between infection peaks (April 2016) and the
flood event which began approximately 18 weeks earlier. Findings
demonstrate all three hydrometeorological variables could be used to
predict the increase in cryptosporidiosis during April 2016, while only
surface water discharge was associated with VTEC enteritis.