Predicting when we die : the design and testing of a tool to predict, at useful levels, the intent to commit suicide in the Hong Kong community and a review of suicide in China, Taiwan and Singapore
Abstract of thesis entitled 'Predicting when we die: The design and testing of a model
to predict, at useful levels, the intent to commit suicide in the Hong Kong community
(and a review of suicides in China, Taiwan and Singapore)
submitted by Peter E Halliday
for the degree of Doctor of Philosophy
at the University of Hong Kong in December, 1999
Chapter 1 discusses the main suicide theories. Durkheim was of the view that as a society weakens the individual depends less on the group, which leads to excessive individualism, or egoism. He felt that suicides in society is a function of the breakdown of religious adherence and the breakdown of society itself. The suicides rate is decided by society.
Douglas contests Durkheim's statistical-hypothetical approach. He regards suicides as "social facts." Baechler takes this further and considers a suicide to be a behaviour and a behaviour peculiar to an individual. Suicides are not ordained by the nature of society but by individual free will. It is suicides that should be considered not the phenomenon of suicides.
The Hong Kong literature is reviewed in Chapter 2. The only substantial study of suicides in Hong Kong is one undertaken by Yap Pow Ming after World War II. Yap demonstrates that, unlike western countries, suicides in Hong Kong increase with age and that the rate of suicides amongst females is more than half that of males. Comparisons are drawn with suicides in China, Taiwan and Singapore in Chapter 3 to the conclusion that Chinese societies, including Hong Kong, have a particular suicides profile, the main elements being the high rates of female suicides overall and the high rates amongst the elderly.
Chapter 5 reviews the literature on prediction of suicides. The consensus is that suicides cannot be predicted at useful levels and that accurate predictive tools cannot be assembled. Case-control methods appear to have the best potential for the successful design of suicides prediction tools. However, the high rates of false positives represent a major obstacle to accurate prediction.
In Chapter 6, it is concluded that the psychological autopsy approach to predicting suicides offers the best hope of success. However, most psychological autopsies reported in the literature involved relatively small samples. Also, many researchers have studied their samples pre-conceiving the factors which are associated with suicides, and only testing "negative" variables. No reliable predictive tool has ever been constructed. It was decided that the use of much larger samples and analysis of the suicides with no pre-conceptions might bring success in the construction of an accurate predictive instrument.
Suicides in Hong Kong in 1992 were subjected to special study (Chapter 8) and formed the researcher's database. The police investigation files were examined and variables of possible relevance to suicides identified. It was decided to attempt to build a predictive tool for suicides involving the elderly. Using logistic regression (Chapter 9) on the 285 cases of elderly suicides for 1992 and a large dataset (1,065) of controls, 11 variables were isolated as having significant association with suicide risk. Three variables were negatively associated with suicides. Discriminant and multivariate statistical analysis of the 11 significant variables produced a model which discriminated between high and low suicide risk to an accuracy of about 93 per cent. The false positives rate was 2.4 per cent.
Some thoughts on reducing the elderly suicides rare are presented.
School:The University of Hong Kong
School Location:China - Hong Kong SAR
Source Type:Master's Thesis
Keywords:suicide china hong kong taiwan singapore
Date of Publication:01/01/2000