Wyper, Grant M.A.
An evaluation of small-area statistical methods for detecting excess risk: with applications in breast and colon cancer mortality in Scotland 1986-1995.
MSc(R) thesis, University of Glasgow.
Full text available as:
The need to report data at small-area level is constantly increasingly. In a society which is both health-conscious and environmentally aware, statistics at small-area level have a high degree of political significance. This type of data is required to plan and implement regional policies and apportion health care in accordance to the differing needs of the population. Recent advances in computer power has brought many advances to this area of study. For all the advances in technology and methodology, the problem of small numbers consistently appears. Is there an excess risk or is it down to chance? This is a question which is paramount in small-area statistics and will be addressed in this thesis.
An overview of the thesis is provided below:
Chapter 1 introduces the concept of small-area statistics and some of the social and political issues connected with this topic. There is a discussion of the analysis of small-area health data and the principal ideas that need to be considered in a statistical, political and social sense in this area of work. The aims of ISD Scotland are introduced and how they can be linked to this field of study.
Chapter 2 describes an overview of the methods used in small-area statistics. The chapter begins by firstly considering the Standardised Incidence Ratio (SIR) which is the technique mainly used in the basic analysis done by ISD Scotland. Other techniques are then considered, however not all of these techniques are directly comparable to each other. The strengths and weaknesses of these techniques in previous research are discussed to give an idea of how the techniques perform in different scenarios.
Chapter 3 is a simulation study of three of the techniques discussed in Chapter 2, these being the SIR, Circular Spatial Scan and Flexibly-Shaped Spatial Scan. The reason for this simulation study is to evaluate these techniques on simulated data arising from real scenarios. The strengths and weaknesses of these techniques are then highlighted which will prove helpful when analysing the data in Chapter 4.
Chapter 4 provides an analysis of the mortality of breast and colon cancer in Scotland in the ten-year time period from 1986 to 1995. Using data provided by ISD Scotland, the analysis is carried out to identity any potential mortality clusters in both diseases.
Chapter 5 provides a conclusion to this research by providing a summary of findings of the thesis and gives recommendations based upon these findings. A discussion is also given for potential further study in this field that could provide some value to ISD Scotland as they look to other ways of analysing their small-area data.
||Small area, epidemiology, breast cancer, colon cancer, risk, disease, incidence, mortality, SMR, SIR, scan statistic, clusters, simulation
||H Social Sciences > HA Statistics
||College of Science and Engineering > School of Mathematics and Statistics > Statistics
||Scott, Prof. E. Marian
|Date of Award:
Mr Grant M A Wyper
||Copyright of this thesis is held by the author.
||24 Nov 2009
||10 Dec 2012 13:37
Actions (login required)