Predicting suicide in older adults - a community-based cohort study in Taipei City, Taiwan

Background: Older adults worldwide are at a greater risk of suicide than other age groups. There is a scarcity of prospective studies exploring risk factors for suicide in older people and their discriminative ability to identify future suicide. Methods: We examined a prospective cohort of senior Taipei City residents between 2005 and 2009 (N=101,764). Cox proportional hazards regression analysis was used to determine significant risk factors and to construct a predictive score. The accuracy of the derived score in the prediction was tested by Receiver Operating Characteristic analysis. Results: Male sex (Hazard Ratio [HR]=3.41, p<0.001), lower education (HR=3.31, p<0.001) and lower income (HR=2.52, p=0.01) were associated with an increased risk of suicide, as well as depressed mood (HR=1.44, p=0.02; per unit increase in a 4-point. scale) and insomnia (HR=1.30, p=0.03; per unit increase in a 4-point scale). The derived prediction score yielded a sensitivity of 0.63 a specificity of 0.73 and an area under curve of 0.73. Removing depressed mood from the prediction model did not significantly alter suicide predictability (P=0.11). Limitations: The dataset examined did not contain information regarding to important risk factors such as substance misuse and prescribed medications and the measures of mental health were relatively limited. Conclusion: Prediction of suicide based on factors recorded in a routine health screen of elderly people was unsatisfactory; the strongest predictors were factors that cannot be easily altered. Further understanding of how the socioeconomic condition of seniors contributes to suicide may provide valuable insights for intervention targeting this growing population-at-risk. (C) 2014 Elsevier B.V. All rights reserved