人種組成及海拔高度對自殺之影響：來自台灣地區的空間分析

Ethnicity Composition and Altitude on Suicide: A Spatial Analysis of Data from Taiwan

目的：目前自殺與空間特性關聯的研究不多，我們以空間統計分析方法來探索台灣本土自殺資料的空間特性。透過全國性的數據分析，探討在生態統計學上不同人種組成以及海拔高度與自殺率之相關性。方法：本研究收集自1997年1月1日至2003年12月31日期間15歲或以上的自殺死亡之國民，自殺死亡定義包括自殺（ICD-9碼：E950-959）及未確定是否意外或故意傷害(E980-989)。計算358個鄉鎮標準化死亡比，鄉鎮平均人口總數為59,490±74,932(mean±SD)，平均面積為100.9±182.6(mean±SD)。統計分析以「鄉鎮」為單位進行，透過控制社會人口學、經濟上、健康照護（包括：失業、婚姻狀況、失能人口、獨居人口、自有住宅、教育水平、人口密度、醫師總數、精神科病床數）等指標與空間自相關後，以空間迴歸模式分析人種組成及海拔高度與自殺率之關聯。結果：分析顯示，在多變數之標準化死亡比模型中，不論是原住民族群或是高海拔地區之居民，此二項因素皆與自殺率呈明顯的正相關，而且對自殺率的影響以在15-24歲及25-44歲這二個年齡層特別明顯。結論：本研究與前人研究同樣顯示高海拔為自殺風險相關因子；本研究的新發現為人種組成及海拔高度對自殺率之影響，主要出現在較年輕的族群。

Objective:Few studies have assessed the spatial association of suicide in the world. In this study, we intended to investigate ecological associations with ethnic composition and altitude on suicide rates using national data from Taiwan. Methods: All suicide deaths of residents aged 15 years and over between January 1, 1997 to December 31, 2003 in Taiwan (n = 23,380) were included. Definition of suicide deaths encompassed suicide (ICD-9 codes E950-959) and injury undeter-mined whether accidentally or purposely inflicted (E980-989). We calculated stan-dardized mortality ratios (SMRs) for all 358 townships, the residents was 59,490 ± 74,932 (mean ± SD) per township, and the average area of townships was 100.9 ± 182.6 (mean ± SD) square kilometers. Each township was treated as a unit in the analysis. We used spatial regression to test associations with ethnic composition and altitude, after adjusting for sociodemographic, economic and health care pro-vision indices (including unemployment, marital status, disability, single-person households, home ownership, educational attainment, population density, total number of medical doctors, psychiatric beds, and spatial autocorrelation). Results: Both proportion of aboriginal residents and altitude of residence were significantly and positively associated with SMRs in fully adjusted models (p < 0.001). The ef-fect of ethnic composition and altitude on suicide predominated in the age 15-24 (p < 0.001) and 25-44 (p < 0.001) subgroups. Conclusion:Our results are consis-tent with previous study finding that altitude is a risk factor for suicide. The novel finding was that the effect of ethnic composition and altitude on suicide was found to be predominated in the young age groups.