臺北市432個里的自殺空間分布、相關變項與不平等

Abstract

背景：超過半數的世界人口居住於都市地區，瞭解都市地區的自殺空間分布能應用於防治。過去針對城市內部的自殺空間分布進行分析的多為西方國家的研究。本研究的目標為探討臺北市的自殺空間模式、以及與其相關的因子。 方法：我們使用貝氏階層模型估計臺北市2004-2010年432個里 (平均人口數：5,500人) 的整體、性別年齡別、與自殺方法 (上吊、燒炭、墜落、溺水、與其他) 別標準化自殺死亡比的平滑估計值。我們並分析一系列地區層級變項，包含社會解離、社會經濟剝奪、連接型社會資本、所得不平等、與方法可近性指標等等，與里自殺率之間的關聯性。 結果：臺北市中心地區的整體自殺率低於全市平均值，但部分都市的邊陲地區的自殺率則高於全市平均值。不同年齡層的男性自殺率呈現相似的地理分布，但不同年齡層女性自殺率的地理模式則有明顯差異。在控制其他變項之後，兩個地區變項與里自殺率相關：離婚/分居人口比例 (為社會解離指標；每增加一個標準差的自殺率比值=1.08，95%置信區間1.01-1.16) 與家庭所得中位數 (為社會經濟剝奪指標；自殺率比值=0.80，95%置信區間0.73-0.86)。除了墜落自殺沒有明顯的地理分布，不同方法別自殺率與整體自殺率的空間分布相似。在控制其他變項後，除了墜落自殺外，不同方法別的自殺率都與家庭所得中位數 (自殺率比值介於0.64-0.84) 相關。燒炭自殺額外與離婚/分居人口比例 (為社會解離指標；自殺率比值=1.12，95%置信區間1.03-1.23) 與獨居家戶比例 (為在居住單位燒炭的容易度指標；自殺率比值= 1.14，95%置信區間1.03-1.26) 有關。墜落自殺僅與居住在六樓以上家戶比例 (為高樓可近性指標；自殺率比值=1.18，95%置信區間1.05-1.31) 有關。溺水與居住地區是否鄰近河川 (為河川可近性指標；自殺率比值=1.24，95%置信區間0.90-1.67) 無顯著相關。在最低與最高的家庭所得中位數地區間的整體自殺率差異達1.8倍，且中年男性有最大的梯度存在 (3.2倍的差異)。 結論：在一個人口密集的亞洲城市，與西方國家的城市相比，自殺率的地理分布顯示獨特的空間分布模式，以及與社會經濟變項和方法可近性的相關性。研究發現對於未來在城市內辨識自殺的特定因子與預防策略有所啟發。

Introduction: More than half of the world’s population now lives in urban areas. Understanding the spatial distribution of suicide in these settings may inform prevention. Previous analyses of the spatial distribution of suicide in cities were mostly restricted to Western nations. We investigated the spatial patterns of suicide, and factors associated with it’s spatial distribution in Taipei City, Taiwan. Methods: We estimated smoothed standardized mortality ratios for overall suicide and suicide by sex/age group and method (hanging, charcoal burning, jumping, drowning and other methods) across 432 neighborhoods (‘Li’; mean population size: 5,500) in Taipei City, Taiwan (2004–2010) using Bayesian hierarchical models. A range of area-level characteristics including social fragmentation, socioeconomic deprivation, linking social capital, income inequality, and means accessibility indicators were investigated for their associations with neighborhood suicide rates. Results: Overall suicide rates were below average in the city center, whereas above average rates were found in some suburbs. Male suicides of different age groups showed similar geographic patterns, while the geographic distribution of female suicides differed across age groups. After adjusting for all other variables, only two area characteristics were found to be associated with area suicide rates: the proportion of divorced/separated adults (rate ratio [RR] per one standard deviation increase =1.08, 95% confidence interval 1.01-1.16), an indicator of social fragmentation; and median household income (RR=0.80, 95% CrI=0.73-0.86), an indicator of socioeconomic deprivation. Method-specific suicide rates showed similar spatial patterning to that of overall suicide with the only exception of jumping suicide rates, which showed no spatial patterning. In adjusted analyses, neighborhood suicide rates of different methods, except jumping, were associated with median household income (RR ranged 0.64-0.84). Charcoal-burning suicide rates were additionally associated with the proportions of divorced/separated adults (RR=1.12, 95% CrI=1.03-1.23), an indicator of social fragmentation, and single-person households (RR=1.14, 95% CrI=1.03-1.26), an indicator of ease with burning charcoal in the living units. Jumping suicide rates were only associated with the proportion of households living on sixth floor or above (RR= 1.18, 95% CrI=1.05-1.31), an indicator of access to high places. Drowning suicide rates were non-significantly associated with neighborhoods’ adjacency to rivers (RR=1.24, 95% CrI=0.90-1.67). There was a 1.8-fold difference in suicide rates between neighborhood quintiles with the lowest and the highest median household income, with middle-aged males showing the largest gradient (3.2-fold difference). Conclusions: In a densely populated city in Asia, the geography of suicide showed distinct patterns of spatial distribution and associations with socioeconomic and means accessibility factors compared to cities in Western nations. Findings have implications for identifying specific determinants and prevention strategies for suicide in cities.