台灣新北市自我傷害與自殺之時空群聚

Abstract

中文論文摘要 背景 自殺行為群聚的定義，是指在空間和時間上比預期更接近發生的自殺或自殺企圖。儘管先前已有對自殺群聚的研究，但對自我傷害群聚現象所知甚少。 目標 本研究旨在使用兩種不同的統計模型來檢測新北市之自我傷害行為的時空群聚。

方法 2012-2016年新北市到急診室求助的自我傷害與自殺死亡資料來自台灣自殺通報系統，包括自傷自殺日期與居住地址的資料，使用時空排列置換模型 (space-time permutation model) 來檢測自傷與自殺群聚。我們也使用離散卜瓦松模型 (discrete Poisson model) 來分析群集到月份與里層次的資料。我們同時分析年齡別與全年齡的資料。 結果 時空排列置換模型分析未發現有自傷群聚，但發現兩個自殺群聚 (分別在20-29歲和50-59歲年齡組)，以及一個在20-29歲的群聚，其中包括五個自傷事件與三個自殺死亡事件。卜瓦松模型分析則發現兩個自傷群聚 (分別在60歲以上和全年齡組)，但未發現自殺 死亡群聚。一個家庭集體自殺事件 (包含三個成人與一名青少年) 重覆出現在三組分析當中，包括針對全年齡層的自殺死亡與自傷/自殺合併資料時空排列置換模型分析，以及全年齡層的自殺資料卜瓦松模型分析。 結論 我們發現自殺行為的時空群聚，包括自殺死亡、自我傷害，以及兩者合併的群聚現象，但此現象並非常見。除了集體自殺事件之外，時空排列置換模型與卜瓦松模型的分析結果少有一致性。因此，統計模型的選擇會影響群聚的識別。未來的研究應考慮同時使用自傷與自殺資料，以及多種時空群聚識別方法來識別最可能出現的群聚。

Abstract Background A suicidal behavior cluster is defined as having more suicides or suicide attempts that occur closer together in space and time than expected. Although suicide clusters have been researched in previous studies, less is known about self-harm clusters. Aim This study aims to identify space-time self-harm and suicide clusters in New Taipei City using two different statistical models.

Methods Data for self-harm emergency room presentations and suicide in New Taipei City (2012-2016), extracted from Taiwan’s National Suicide Surveillance System, which included information on the date of self-harm or suicide events and residential address, were analyzed using the space-time permutation model to identify self-harm and suicide clusters. The discrete Poisson model was also used to analyze data aggregated to the level of month and Li (neighborhood). Analyses were conducted using both age-specific and all-age-group data. Results The space-time permutation model analyses found no self-harm clusters, two suicide clusters (one in individuals aged 20-29 years and another in those aged 50-59), and one cluster comprising five self-harm and three suicide events in the group aged 20-29. The Poisson model analyses identified two self-harm clusters, one in individuals aged 60+ and another in the all-age-group analysis, but no suicide clusters. One suicide pact of four individuals (three adults and one adolescent) from the same family was identified in three sets of analysis, including the all-age-group suicide only and suicide/self-harm combined data cluster analysis using the permutation model and the all-age-group suicide cluster analysis using the discrete Poisson model. Conclusions Space-time clusters of suicide, self-harm, and suicide and self-harm data combined were identified, but were rare. With the only exception of the suicide pact, there appeared to be little consistency in findings between the space-time permutation model and the discrete Poisson model analyses. Therefore, the statistical model choice may have an impact on cluster identification. Future research should consider combining self-harm and suicide data and using multiple space-time cluster identification methods to identify the most likely occurring clusters.