所得與道路交通事故死亡率關係之研究

Abstract

背景：跨國跨時研究顯示所得與道路交通事故死亡率的關係在低所得水平時是正相關，隨著所得水平提升正相關減弱，達到某一所得臨界值後，所得與死亡率開始呈現負相關。可惜並沒有研究以一國跨縣市跨時研究來探討兩者之關係並探討政策介入之影響。 目的：首先想瞭解以台灣為分析單位及以縣市別為分析單位分析所得與道路交通事故死亡率的關係在不同經濟發展階段是否不同？其次想瞭解造成這種不同階段關係改變的主要原因為何？第三想瞭解高低發展地區上述關係改變形態是否不同？第四想瞭解1997年強制戴安全帽立法政策介入對所得與死亡率關係的改變為何？ 方法：台灣為分析單位的所得指標是人均國內生產毛額，縣市別為分析單位的所得指標是家庭可支配所得，資料來源皆是主計處。道路交通事故死亡率資料由衛生署統計室獲得。所得與死亡率之關係分別以所得分組、朴瓦松廣義程式估計線性迴歸模型與廣義相加非線性迴歸模型進行估計。不同經濟發展階段主要以年代、所得水平與地區發展程度來區分。 結果：以台灣為單位的分析結果顯示所得與道路交通事故死亡率之關係的確在經濟發展初期呈現顯著正相關，隨著所得增加正相關減弱，當人均國內生產毛額超過6,223美元（1988年）後，兩者關係變成負相關，但是大於13,413美元（1997年）後負相關消失。以台灣縣市別為單位的分析結果也顯示類似的關係改變型態，當家庭可支配所得約47萬元（1989年）時開始出現轉折，但是在2001-2004年負相關強度也減弱。在1988年以前，家庭可支配所得是縣市別道路交通事故死亡率的最主要解釋變項，1988年之後是教育程度。影響所得與死亡率關係改變主要是因為所得與暴露率（機動化程度）關係改變造成。在低發展地區出現死亡率轉折的所得臨界值（43萬元）比高發展地區（47萬元）低，當高發展地區持續出現負相關的年代，低發展地區還是持續正相關。1997年政策介入對高低發展地區的死亡率都有顯著下降影響，但是低發展地區所得與死亡率的相關性在1997年以後開始不顯著。 結論：台灣做為一個經濟發展後進者，再加上全球化的影響，台灣與歐美先進國家（可能是三四十年前）達到相同所得水平時，所能消費的車輛安全性、道路環境與相關交通安全措施都比較佳。因此道路交通事故死亡率不會被所得命定，台灣可以在較低所得水平就出現死亡率下降趨勢。此外，台灣幅員小，道路安全行為與觀念在縣市間的傳散較快，道路建設與交通安全相關政策也都由中央推動。使得縣市別道路交通事故死亡率更不會被所得命定，尤其低發展縣市可以在較低所得水平就出現死亡率下降趨勢。

BACKGROUND: Cross-national time series studies reveal that the relationships between income and motor vehicle accident (MVA) mortality is non-linear: during initial stages of economic developments, a positive correlation is found. The correlation decreases when the income increases. When the income raises to a certain level, the association become negative. Nevertheless, no study examined the relationship using cross-county level time series data and assess the effects of road safety interventions on the relationships. OBJECTIVES: First, to determine the pattern of changes in relationships between income and MVA mortality in Taiwan at both country and county level. Second, to identify the main determinats resulting the changes in the relationships. Third, to reveal if the pattern of changes in the relationships differed between high and low developed areas. Fourth, evaluate the impacts of mandatory helmet law launched in 1997 on the relationships between income and MVA mortality. METHODS: Per capita gross domestic products (GDP) and disposable household income was used as the proxy of country and county income respectively. Both data were from Bureau of Budge and Accounting, Taiwan. MVA mortality data were extracted from Department of Health, Taiwan. Relationships between income and MVA mortality were assess by income group differences, coefficiency derived from Poisson generalized equation estimation (GEE) linear model and predicted mortality based on generalized added non-linear model (GAM). Stages of economic development were stratified by years, income level and areal development level. RESULTS: The association between income and MVA mortality at initial development stage was positive at country level. The level of association decreased when income level increased. The association became negative when the per capita GDP over 6,223 US$ (1988). The negative association, however, disappeared when per capita GDP over 13,413 US$ (1997). The pattern of changes in relationships between income and MVA mortality at county level was similar. The turningpoint of mortality changes occurred when household disposable income at 470 thousands NT$ (1989). Similarly, the negative relationship decreased in 2001-2004. Before 1988, household disposable income was the most significant determinant explaining the MVA mortality at county level. Level of education became the most important determinat after 1988. Changes in relationships between income and MVA mortality was mainly due to changes in relationships between income and exposure (motorization, vehicles per 1000 population). The income level of which the MVA mortality decline was lower in low developed areas (430 thousands NT$) than in high developed areas (470 thousands NT$). When the years of which the relationships were negative in high developed areas, the relationships were still positive in low developed areas. The MVA mortality significantly decreased after the launch of mandatory helmet law in 1997 in both high and low developed areas. Nevertheless, the relationship between income and MVA mortality became non-significant in low developed areas after 1997. CONCLUSIONS: Taiwan as a late comer of economic development in the age of globalization, when income level was the same between Taiwan and those highly developed countries (maybe thirty or forty years ago), the motor vehicle, road infrastructure and measures of transportation consumed by people in Taiwan would be safer than those in highly developed countries. Therefore, the MVA mortality was not fatally determined by income, the increasing MVA mortality trends could decrease at lower income level than expected. One of the features in Taiwan was that the road safety adaptation behavior and concepts could be diffused easily compared with many large scale countries. In addition, the most of construction of road infrastructure and road safety measures were centralized. The MVA mortality in county level was not fatally determined by income either. The increasing MVA mortality trends could decrease at lower income level than expected in low developed areas.