Increasing suicide death in the taiwanese elderly from 1999-2007: A time-spatial multilevel approach

uicide rose dramatically in Taiwan during the last decade, especially among elders. This study was to determine whether there are statistically significant geographical clusters of suicide among those aged above 65 years and whether these clustering patterns persist over time. Method: Suicide data was obtained from Death Certification data (ICD-9 code from 950-959) were obtained from 1999 to 2007. The Time-Spatial Scan Statistic (SaTScan) in combination with multilevel models was used to identify potential time-spatial clusters of suicide. Within different contexts of the time-spatial clusters, multilevel logistic regression was used to examine individual and neighborhood variables associated with the chance of elderly suicide death. Two significant (p < 0.001) time-spatial clusters were identified in southern (RR=1.226) and northern (RR=1.661) Taiwan between 2003 and 2006. The odds of elderly suicide death is stronger in Southern than in Northern Taiwan. The time-spatial clusters are mainly socio-economic disadvantaged areas characterized by low level of urbanization, low level of high education, high level of single-elderly family, high-level of single-parent family. Within time-spatial clusters, male vs. female, younger (aged 65-74 year), divorced were significant predictor of elderly suicide death. The seasonality of suicide death (especially spring and summer) is found significant in time-spatial clusters. The findings of the present study provide evidence for the existence of two time-spatial clusters, and show the geographic variations with regards to suicide death in Taiwanese elderly. The public health interventions should focus on local patterns of suicide death in elderly.