Exploratory Spatial Data Analysis of Motor Neurone Disease in North West England: Beyond the Address at Diagnosis

Abstract

This paper addresses the issues surrounding an individual’s exposure to potential environmental risk factors which can be implicated in the aetiology of a disease. We hope to further elucidate the ‘lag’ or latency period between the initial exposure to potential pathogens and the physical emergence of the disease, with specific reference to the rare neurological condition, Motor Neurone Disease (MND). A Space-Time approach is adopted, whereby patterns in both time and space are considered. No prior assumptions about the aetiology of MND are adopted. By using methods for the analysis of point processes, which preserve the continuous nature of the data, we resolve some of the problems of analysis based on often arbitrary areal units, such as postcode boundaries, or political boundaries, which diseases do not respect.

We use kernel estimation to model Space-Time patterns. Raised relative risk is assessed by adopting appropriate adjustments for the underlying population at risk, with the use of controls. Significance of the results are assessed using Monte-Carlo simulation.

Our results demonstrate the utility of kernel estimation as a visualisation tool. Small areas of elevated risk appear to have been identified, which need to be more closely examined before any firm conclusions can be drawn. We highlight a number of issues concerning the inadequacies of the data, and possibly of the techniques themselves. Various modifications have been proposed, with the most promising being adaptive kernel estimation.