|
[Print-friendly version]
 Margaret Donald
Faculty of Science, Quaternary Earth and Water Systems
Thesis Title: Health risks associated with recycled water use (a Bayesian analysis)
Aim: With climate change and increased prolonged droughts affecting most states in Australia, interest has developed in reusing waste water.
In Western Australia recycled water will be introduced in stages. Prior to undertaking the managed aquifer recharge scheduled for 2015,
it is planned to first use
recycled water for watering ovals, for industrial processes and for agriculture, then to
introduce third pipe systems for new residential areas, and finally to use high quality
recycled water for recharging the aquifers. The School of Population Health, UWA,
has won a contract with the Premier's Water Foundation to undertake \integrated
health risk assessments for all modes of recycled water use in urban and rural Western Australian
communities using cutting-edge epidemiological and quantitative risk
assessment techniques".
The PhD project described here aims to contribute to this research endeavour by
developing and applying a range of Bayesian statistical methods to the problem of
identifying the health risks associated with exposure to recycled water. Three main
approaches will be adopted: Bayesian networks, Bayesian spatio-temporal models
and Bayesian time series changepoint models.
A number of specific disease risks, recycled water sources and data types will be considered using these approaches.
The Bayesian networks will focus on expert-elicited
risk of disease, in particular rotavirus, cryptosporidiosis, and campylobacteriosis,
associated with the use of recycled water for irrigation for on wastewater treatment
plant in WA. Data-based risk assessment for particular sources related to a specific site, MacGillivray Oval in Perth, will also be developed.
The spatio-temporal
and time series models will focus on notifiable diseases such as cryptosporidiosis,
rotavirus and giardiasis which have a component of water-borne transmission, although in developed countries the majority of cases arise from faecal-oral spread.
The intention is to use the national hospital morbidity database ? and hence ICD-9,
ICD-10 de¯nitions. These data will be analysed in conjunction with weather data,
and spatially identified recycling schemes.
Many diarrhoeal diseases show marked temporal variability, often in relation to
seasonal factors. Rotavirus, for example, in south-eastern Australia shows a marked
peak in September, and this sort of pattern can be observed with campylobacterosis
also 1. There is a possibility of linking patterns in these diseases with the patterns
of recycled water introduction, adjusting for pre-existing time and spatially varying
factors such as climate.
The research will be in a Bayesian framework as it provides possibly the most °exible
and transparent form of statistical modelling.
1Associated in the US with excess spring rain run-off, contaminating surface-water and wells
from infections in animal herds.
|
