新北市自我傷害與自殺之空間分布與相關因素

Abstract

研究背景：自殺及自我傷害都是公共衛生的重要優先事項。過去研究顯示自我傷害和自殺存在空間差異。然而相較於自殺身亡，過去對於自我傷害之地理分布與相關因素的研究相對稀少，並少有與自殺身亡之直接比較。 研究目的：探討新北市自我傷害與自殺之空間分布模式與地理層級相關因素，並關注到急診室旅行距離之關聯，因為較長的就醫距離可能是自我傷害後求醫的一大阻礙。 研究方法：本研究使用2012-2016年自殺防治通報系統中，新北市自我傷害與自殺資料。應用貝氏階層模型估計新北市1,032個里之平滑化標準化自我傷害發生比，以及平滑化標準化自殺死亡比，並分析里自傷率、里自殺率與一系列之區域特性變項，包含社會解離、社會經濟剝奪、經濟不平等、連接型社會資本、都市化程度、與最近醫院急診室距離指標之間的關聯性。 結果：自我傷害發生率及自殺率呈現微弱正相關 (斯皮爾曼相關係數=0.14；95%信賴區間0.07-0.20)。新北市自我傷害率與自殺率呈現不一樣的分布。自我傷害率高於整體平均值之地區主要位於人口密集區域，而低於平均值之區域主要在北部鄰海與東部、南部區域；人口集中之中心區域自殺率相對較低，高自殺區域分布在周邊鄰海或山區。在控制其他區域變項後，自我傷害率和自殺率兩者都和離婚率 (為社會解離指標) 呈現正相關；自我傷害率和自殺率兩者都和家庭收入中位數 (為社會經濟剝奪指標) 呈現負相關。距最近醫院急診室距離則只和自我傷害呈現負相關 (自我傷害率比值= 0.92，95%置信區間0.89-0.95)；相較於距最近醫院急診室距離最短的五等分位組，距最近醫院急診室之距離最遠的五等分位組別的自我傷害率較低 (自我傷害率比值= 0.72, 95%置信區間0.66-0.80)。距最近醫院急診室距離和不同自我傷害方法別之間的相關性，在控制其他區域變項後，只和自我切割、服藥過量呈現負相關，在其他方法 (燒炭、墜落) 則無證據顯示有關。 結論：新北市自我傷害及自殺的地理分布顯示兩者的空間分布模式不同。社會解離和社會經濟剝奪都和區域自我傷害率及自殺率有關。然而，以醫院就醫為主的自殺通報系統，可能低估離醫院較遠地區的自我傷害率。

Background: Suicide and self-harm are both public health priority. Past research showed spatial variations in the incidence of both self-harm and suicide but rarely compared the patterns and correlates of the geographical distributions between self-harm and suicide. Aim: This study is aimed to investigate the spatial distributions of self-harm and suicide and area-level characteristics associated with their patterns, focusing on the travel distance to hospital emergency departments as longer distance could be a major barrier for seeking treatment after self-harm. Methods: Data of self-harm and suicide in New Taipei City (2012-2016) were extracted from a nationwide surveillance system in Taiwan. Bayesian hierarchical models were used to estimate smoothed standardized incidence ratios of self-harm and smoothed standardized mortality ratios of suicide as well as their associations with area-level factors, including travel distance to the nearest hospital emergency department, socioeconomic deprivation, social fragmentation, income inequality, linkage social capital, and urbanization level across 1,032 neighborhoods (median population size = 3,426) in New Taipei City. Results: There was only a very weak correlation between neighborhoods’ self-harm and suicide rates (Spearman’s correlation coefficient = 0.14, 95% confidence interval 0.07-0.20). The spatial distributions of self-harm and suicide rates showed some dissimilar patterns. In adjusted models, both self-harm and suicide were positively associated with divorce rates and negatively associated with median household income. However, only self-harm rates showed a negative association with travel distance to the nearest hospital emergency department (rate ratio [RR] = 0.92, 95% credible interval [CrI] 0.89-0.95); low rates of self-harm hospital presentation were found particularly in the quintile of neighborhoods with the longest distance to the nearest emergency department compared to the quintile with the shortest distance (RR = 0.72, 95% CrI 0.66-0.80). The negative association between self-harm and travel distance to the nearest hospital emergency department was found in cutting and overdose only but not other methods of self-harm. Conclusions: Both social fragmentation and deprivation may influence the spatial distribution of self-harm and suicide incidence. However, surveillance system of suicidal behavior based on hospital presentations may underestimate self-harm incidence in areas that are distant to emergency departments.