Date of Award


Document Type



Edith Cowan University

Degree Name

Doctor of Philosophy


School of Science

First Supervisor

Professor Pierre Horwitz

Second Supervisor

Stacy Jupiter

Third Supervisor

Adam Jenney


Many current disease threats involve interactions within and between nested subsystems of biological organisation. Typhoid fever is a serious disease threat in the South Pacific region, with Fiji reporting the highest annual number of cases, yet risk factors in this setting have been poorly studied. While localised behaviours have dominated perspectives on typhoid transmission, interactions between distal ecological conditions, conditions of the residential environment and localised behaviour deserve greater attention for their potential to influence transmission. This thesis demonstrates a nested approach to typhoid epidemiology using a fivefold methodology to explore how regional, river basin, residential, socio-cultural and behavioural subsystems influence the risk of typhoid transmission in Central Division, Fiji, whereby I: (1) provide a regionally specific literature review examining health consequences of wetland ecosystem service interruption associated with common natural disasters; (2) use quantitative geospatial analysis to evaluate relationships between sub-catchment environmental characteristics and typhoid incidence and recurrence; (3) use a case-control design at a residential level to investigate bacterial contamination and chemical composition of water and soil as vehicles of exposure, complemented with observational analysis of living conditions, spatial analysis of household position and factor analysis to explore multivariate relationships influencing typhoid risk; (4) question 160 typhoid fever cases and 319 control subjects to reveal risky socio-cultural and behavioural practices and; (5) synthesize significant risk factors within and across nested subsystems and test several intervention scenarios using a Bayesian Network approach.

Regional typhoid burden is influenced by climate change induced warming, altered rainfall patterns, increased storm severity and rising seas, coupled with population growth, slow economic growth, urbanisation, environmental change and limited capacity for adaptive management. The most parsimonious models for incidence and recurrence at the sub-catchment scale included total high soil-erosion risk area, percentage area that was highly erodible, connectivity between road and river networks and riparian forest fragmentation as predictor variables. In the residential setting, five factors, related to drainage, housing and condition of water and sanitation were significant in predicting typhoid. Multivariate analysis of household questionnaires indicated the following significant risk factors for typhoid fever: using an unimproved pit latrine, not washing produce (i.e. fruit or vegetables) or hands before eating, bathing outside, water not always accessible, having sand/wood plank floors and attending mass gatherings. The above results suggest that anthropogenic alteration of land cover and hydrology in river basin and residential systems increases risk of exposure where sediment increases following runoff. Localised socio-cultural and behavioural subsystems interact with residential and river basin subsystems to enhance risk of typhoid transmission. Bayesian network analysis suggests combined interventions within a subsystem provides greater exposure reduction than the sum of individual interventions and simultaneous interventions on select risk factors, across multiple nested subsystems, provides greater exposure reduction than elimination of risk factors in any one subsystem. A nested epidemiological approach to studying and interrupting waterborne disease transmission extends the testing of causal assumptions beyond the domestic domain, enhances traditional case-control approaches and provides evidence for multi-scale interventions on both distal and proximal drivers of disease and environmental degradation.


Paper Location