| BackgroundSuicide, particularly youth suicide in rural areas, is a major public health issue and social problem which our country is currently confronted. Effective suicide prevention and intervention strategies require in-depth studies of completed suicide as a reliable scientific foundation. However, a methodological problem, how to effectively collect the deceased's full information, is in the way of researches for completed suicide. The development and application of Psychological Autopsy, offers an effective way to solve this difficulty. Through the interviews with "the other people who know the information of the dead", etc., Psychological Autopsy, to be effective in the reconstruction of death and suicide-related psychological and social factors, a better understanding of suicide causes, is the most scientific and operational approach for suicide research.The application of Psychological Autopsy in suicide researches in our country is still in its infancy. Moreover, so far, there are no Psychological Autopsy studies specific to rural youth suicide. Effective suicide intervention requires strategies custom-tailored to the characteristics of different populations accordingly. In order to better implement rural youth suicide prevention and intervention in our country, employing Psychological Autopsy method to further explore the characteristics of youth suicide in rural areas is very necessary.The informant is the main information source of Psychological Autopsy, while a single source of information often can't comprehensively reflect the whole situation of the target, so several informants can be served as Psychological Autopsy information sources when human resources, money and other conditions allows. When two informant or more are selected in Psychological Autopsy, different information sources may provide inconsistent data. To date, there is no uniform protocol on how to synthesize various data from several information sources in the worldwide field of suicide research, and convincing evidences to prove which information synthesis technique is relatively better are also not available now. Therefore, how to utilize proxy data gained from several informants to reconstruct the decedents or other targets, and whether these proxy data provided by these techniques can effectively represent exact information of targets, have been on the agenda of development of Psychological Autopsy.Application of Psychological Autopsy should not merely focus on theoretical study, and it should also be applied to the practice of suicide research so that it can better serve for suicide prevention and intervention. Previous Psychological Autopsy studies mainly target to identify single feature, rather than characteristics cluster. Analysis of characteristics cluster, usually explored by Classification and Regression Trees, has its unique advantages in exploration of suicide risk factors and other suicide studies. Furthermore, a large number of studies have consistently demonstrated that, suicide doesn't result from a single risk factor and suicide decedents usually have multiple risk factors with several protective factors turned on at the same time. Hence, the approach of characteristics cluster analysis in Psychological Autopsy study will be much more instructive and operable for suicide prevention and intervention than identification of single feature.In response to those multiple questions mentioned above, we chose two informants for each target in case and control groups to conduct Psychological Autopsy, and it is the first time in the international to fully explore validities of various information reconstruction techniques, then apply these techniques which have been testified to be valid to indentify characteristics clusters of young suicide decedents in rural China, with a view of providing a scientific basis for suicide prevention and intervention in our country.Objectives1. To explore validities of various information reconstruction techniques to provide a scientific foundation for information collection in Psychological Autopsy and information rebuilding for targets; 2. To explore high-risk characteristics clusters of completed suicides among rural youth in China to serve as an effective reference to frame strategies for suicide prevention and intervention.MethodsTo randomly select 16 counties or county-level cities in Shandong, Liaoning and Hunan provinces, then consecutively sample rural residents aged 15-34 who died by suicide during the period from 1st October,2005 to 30th June,2008, in those 16 counties or cities, meanwhile randomly sample rural living residents at the same age group in the same quantity as controls of those suicide cases. The list of suicide originates from monitor systems of death causes of Centers for Disease Control and Prevention at the county levels. Totally 392 suicide cases and 416 controls were included in this study.Data of control group were used to evaluate the validities of different Psychological Autopsy information reconstruction techniques, with information provided by controls themselves as gold standard. Paired t-test and Intra-class Correlation Coefficient were employed to evaluate validities of type lst,2nd, A, B, C, D and E, a total of seven different types of information reconstruction techniques on numerical data, such as hopelessness, social support, impulsivity, trait anxiety, coping skills and so on. McNemar test in combination with several parameters such as Sensitivity, Specificity, Youden Index and Kappa Value were adopted to evaluate validities of type 1 st,2nd, D, E, G and H, a total of six different types of information reconstruction techniques for dichotomous data such as occurrence of life events.Calculate the absolute differences between proxy data and golden standard on hopeless, social support, impulsivity, trait anxiety and subscale scores of Coping Response Inventory respectively, then add them up as dependent variable to screen variables of informant's characteristics which affect the validity of proxy data, with "relationship with target, familiarity to target, sex, age, marital status, occupation, schooling, religion, family annual income, physical health and CES-D depression score" as independent variables. Generalized Linear Model was employed for analyses, with Scaled Deviance and Pearson Chi-Square as modeling fitness parameters.The best or relatively better Psychological Autopsy information reconstruction techniques were correspondingly used to rebuild different aspects of information for 392 suicide cases and 416 controls. Then explore influencing factors of youth suicide in rural China by Case-Control study, with informants'characteristics which affect the validity of proxy data adjusted, and screen their high-risk characteristics clusters by Classification Tree as well.Main Results1. Validity Evaluations of Different Information Reconstruction Techniques in Psychologcial AutopsyIntra-class Correlation Coefficients or Kappa Values of the information built by nine information reconstruction techniques including type 1 st,2nd, A, B, C, D, E, G and H were less than 0.6, which indicated consistencies between proxy data and information provided by control himself/herself were not good enough, that is, accuracies of proxy data information were not so good.Information built by the five information reconstruction techniques including type 1st,2nd, A, D and E, had no statistical differences from the information provided by control on Hopelessness Scale and its three sub-scales, Social Interaction and Instrumental Support which were two sub-scales of Duke Social Support Index, Functional Impulsiveness of Impulsiveness Scale, Trait Anxiety Inventory, Positive Coping Dimension of Coping Response Inventory and its four sub-scales, and Negative Coping Dimension and its two subscales including Cognitive Avoiding and Compensation Seeking. Scores of Perceived Social Support Subscale of Duke Social Support Index were all larger than those of controls (P?0.05), which indicated information of this subscale was not good. Scores of Non-functional Impulsiveness sub-scale of Impulsiveness Scale and Emotion Discharge of Negative Coping Dimension were also smaller than those of controls themselves (P?0.05). Discrepancies between information provided by the five information reconstruction techniques mentioned above and that of control himself/herself on Scale of Impulsiveness and "Accept or Give Up" sub-scale of Negative Coping Dimension of Coping Response Inventory were not the same:(1) as far as Scale of Impulsiveness was concerned, proxy data of type 1st,2nd and D, had no difference from golden standard, but those of type A and E were smaller than those of controls themselves (P?0.05); (2) regarding "Accept or Give Up" sub-scale, information provided by 1st and D type information reconstruction techniques had no significant difference from control, other three proxy data of 2nd, A and E type were smaller than that of control (P?0.05).In addition, information built by B and C type information reconstruction techniques was significant statistical different from that of control in all the above-mentioned scales. Information of the former one was larger than that of control, while the latter one was smaller than control. Overall, validities of 1st and D type information reconstruction techniques were a little bit better than other five information Reconstruction techniques (2nd, A, B, C and E) on the above scales.Information concerning four categories of live events including marriage related issues, work and study related issues, health related issues, and law related issues and others, offered by G-type information reconstruction technique which was featured by "taking positive value", was not significant different from that of controls themselves, and their P values were 0.810,0.363,0.534 and 0.477 respectively. Type 1st proxy information had no significant difference from control on family life related life event (P=0.828). Information built by other four information reconstruction techniques (2nd, D, E and H type) in the above life events was significant different. In addition, information provided by the six information reconstruction techniques was significant different on Sensitivity, Specificity and Youden Index of Life Event Scale; among which, the information provided by G-type information reconstruction approach which was featured by taking positive value, had the best Sensitivity and Youden Index, but the worst Specificity.2. Characteristic Cluster Analyses of Completed Suicides among Young Residents in Rural ChinaAmong rural young people in our country, those who were with more schooling, better physical health status, better ability of seeking guidance and support, less marriage-related negative life events happened to them in the past year, had lower risk for suicide. Having unaccomplished desire and suffering from mental disorders were suicide risk factors, and their OR values were 4.19 and 9.95 respectively. Higher scores of Feeling for Future and Expectation for Future in Hopelessness Scale indicated higher suicide risk. Higher score of Emotion Discharge of Negative Coping Dimension and positive rate of Family History for Suicide, showed the trend to increase suicide risk but there were no statistical significances (OR=1.09,4.5; P=0.183,0.157). Social support, impulsivity, trait anxiety are not important influencing factors for completed suicide among rural young residents in our country.High-risk characteristic clusters of China's rural youth suicide mainly were:(1) with score of Feeling for Future larger than 17.5 and Expectation for Future larger than 12.5; (2) with score of Feeling for Future equal to or smaller than 17.5 and suffering from mental diseases; (3) with score of Feeling for Future equal to or smaller than 17.5, no mental diseases, relatively adequate ability of Seeking Guidance and Support, less schooling, and score of Emotion Discharge larger than 5.5; (4) with score of Feeling for Future equal to or smaller than 17.5, without mental diseases, and worse ability of Seeking Guidance and Support; (5) with score of Expectation for Future smaller than or equal to 12.5 and Feeling for Future larger than 17.5. Sensitivity of the suicide Classification Tree in training sample was 84.34%, while the specificity was 92.08%. This Classification Tree was validated by test sample, and its Sensitivity was 76.47% while the specificity was 85.92%. The Sensitivity and Specificity of training and test sample consistently showed that the suicide Classification Tree discriminated well.Conclusions and SignificanceAlthough the accuracy of proxy information is not so good, proper information reconstruction technique can still be an effective tool for estimating targets' information. The first informants who are mainly composed of target's relatives are informative enough to build majority of the target's information in Psychological Autopsy studies. But the presences of the second informants who are mainly target's friends and neighbors, will not contribute much to enhance validities of most proxy data. It is better to interview two informants to collect life event information for better validity, but proper information reconstruction technique should be adopted. When the second informants are interviewed, inappropriate information synthesis method, for example, taking the maximum or minimum of numerical information, will lead agent information to a serious deviation from the targets'.There are a variety of high-risk suicide characteristic clusters among China's rural youth, not only including high-risk suicide factors, but may also accompanied by other protective factors for suicide, or just an aggregate of several low-risk suicide factors. Potential high risk for suicide should not be ignored only because some individuals have protective factors for suicide or do not have any common high-risk characteristics of suicide. Effective suicide prevention depends on identification of suicide high-risk characteristics clusters, development of corresponding screening tools, and comprehensive assessment of suicide risk for individuals.It's the first time in the international to fully explore validities of different information reconstruction techniques, with a view of providing a scientific basis for data collection and synthesis in Psychological Autopsy, partly establishing a foundation for standardization of information construction in Psychological Autopsy. Besides, this study tried to explore high-risk characteristics clusters among suicide of rural youth in China as a pioneer, providing an effective reference for making suicide intervention strategies for our country. |
PDF Full Text Request