A Psychological Autopsy Study On Rural Suicides Using Conditional Logistic Regression Model

Background: Suicide is a self-destructive behavior caused by psychological conflict, which is an intentional means to end one's own life. The study of the suicide risk factors has become one of the hotspots of scholars all over the world. At present, there are two methods to study the causes of suicides: the epidemiology method and the psychological autopsy method. Psychological autopsy is better to understand the social, psychological and physical conditions of suicide, so it is widely used. While in China, most of the studies on suicides still use the epidemiology method except one research by Phillips using the psychological autopsy. There are few studies on the suicide risk factors using case-control conditional logistic regression in the world, and in China, there is no such study.Objective: To analyse the risk factors of suicide with the psychological autopsy method, in order to find out the social, psychological and physical risk factors of suicide and provide reliable evidences for suicide interference.Method: 66 complete suicides in JinZhou and ZhuangHe, DaLian, LiaoNing province from 2000 till 2001 were investigated in the study. Each suicide case was matched with a normal living individual on gender, age (±5years) and location. Case-control psychological autopsy method was used in this study. The protocol is a set of questionnaires including the self-structured basic status questionnaire of suicide, Beck's Suicide Intent Scale (SIS), Paykel's Interview for Recent Life Events (IRLE), Duke Social Support Index (DSSI), The Structured Clinical Interview for DSM-Ⅲ-R (SCID), Hamilton Depression Rating Scale (HAM-D), Personality Measurement (NEO). SPSS11.5 and SAS8.2 were used to analyse the data. Since the suicides and controls were respectively matched, conditional logistic regression was applied for data analysis. Backward LR statistic was used in the conditional logistic regression model. Result: In univariate analysis there were significent differences between the suicide group and the control group on education, family income, marital status, belief in life after death, the score of social communication, the score of perceived social support, the score of instrumental social support, the total score of social support, the score of Hamilton depression rating scale, neuroticism, extroversion, agreeableness, conscientiousness of NEO-FFI, mental disease, schizophrenia, depression, neurosis, the frequency of life events. Marital status, family income, education status, belief in life after death, living condition, employment status, the total score of social support, the score of Hamilton, the five factors of NEO-FFI, schizophrenia, neurosis, the frequence of life events were entered into the conditional logistic regression model. Three factors of suicide remained in the final conditional logistic regression model. They were social support ( OR = 0.663 ,CI = 0.481 ~ 0.913 ), depression(OR=1.181,CI=1.031~1.354) and neurosis(OR=2594.590,CI=6.294~1069550). Conclusion: In this study, we confirm that depression, neurosis and social support are closely related to suicide. Our results lend strong support to a multifactorial approach to the understanding and prevention of suicide-ie, that the combined effect of multiple factors increases the risk of suicide, not the unique effect of single factors. The preventive efforts concentrated on single factors are not likely to reduce suicide rates substantially. Suicide intervention should focus on the combined effect of multiple factors.