台灣死因別死亡率之社會經濟不平等（1971-2012）：生態研究

Abstract

背景與目的：健康不平等的議題近年來引發許多關注，減少或避免健康不平等成為重要的政策目標。在國家發展不同進程、流行病學轉型的過程當中，以及普遍性健康保險的實施，例如1995年台灣開始實施的全民健康保險，健康不平等的模式可能有重大的變化，但在這方面的研究較少，且多集中在西方國家。本研究分析1971-2012年期間台灣的全死因、死因別死亡以及醫療可避免死亡的社會經濟不平等趨勢變化。 研究方法：我們使用鄉鎮市區（n=354）作為分析單位，進行生態研究，使用五個時期（1971-1975，1978-1982，1988-1992，1998-2002，2008-2012）的原因別死亡資料，而鄉鎮地區的社經地位指標是由人口普查資料（1965,1980, 1990, 2000，2010）與所得稅資料（2000與2010）計算而得。由於與家戶所得關聯性最強的指標是低教育比率，我們因此使用低教育比率作為鄉鎮市區社經地位的主要指標，同時也使用農業人口比率，以及從主成分分析分析13個社經地位資料所得到第一主成分為敏感度分析的指標。我們計算鄉鎮市區的全死因、死因別以及醫療可避免死亡的當期期望生命年數損失（Period Expected Years of Life Lost），然後將鄉鎮市區依社經地位指標排序，以此繪製五個時期的集中度曲線（concentration curve）以及計算集中度指數（concentration index），以呈現1971-2012年間台灣健康不平等的趨勢變化。 研究結果：全死因集中度指標從1971-1975的-0.100（95%信賴區間-0.120，-0.080）下降到1998-2002的-0.123（-0.142，-0.104），然後略增到2008-2012的-0.120（-0.141，-0.100），表示生命年數損失在不同年代均傾向集中於低社經地區（集中度指標為負值），且不平等程度在研究期間1978-2002年期間擴大，之後穩定。不同死因的結果有明顯差異，例如，癌症在1971-1975期間無明顯不平等，但之後不平等持續擴大；心臟血管疾病的不平等情形在1971-2002年間擴大，但之後穩定；呼吸系統疾病及傳染性疾病的生命年數損失也傾向集中於低社經地區，但不平等有縮小的趨勢。意外傷害及中毒的不平等程度則在1971-2002期間逐年擴大後再略為縮小，然而其中自殺的健康不平等逐漸消失，而交通事故傷害和其他意外及中毒的健康不平等則相反地持續擴大。醫療可避免的死亡的集中度指數從1971-1975年的-0.126（95%信賴區間-0.151，-0.100）上升到1988-1992年的-0.087(-0.107，-0.066)，在1995年全民健保實施後，1998-2002的集中度指數為-0.090(-0.111，-0.070)，與健保實施前相近，但之後又於2008-2012年持續增加到-0.081(-0.102，-0.059)，表示健康不平等程度在研究期間縮小，但在實施健保前的健康不平等縮小程度大於健保實施後的。 結論：過去40年間，台灣大部份疾病的生命年數損失皆傾向集中於低社經地區，且全死因的健康不平等有擴大趨勢，然而醫療可避免死亡的不平等程度則持續下降，但全民健保改善健康不平等的效應並不顯著。

Background：Health inequalities have attracted much attention in recent years. Reducing or avoiding health inequalities are important policy goals. However, there could be significant changes in the patterns of health inequalities across different development stages and during the epidemiological transition. The implementation of universal health coverage, e.g. Taiwan’s National Health Insurance (NHI) may also play a role in the changes. Past research is limited in this area and mostly from Western countries. In this study we analyzed temporal changes in socioeconomic inequalities of all-cause and cause-specific mortality as well as causes of death considered amenable to health care in Taiwan during the period from 1971 to 2012. Methods：We conducted an ecological analysis using townships (n=354) as the area unit. Cause-of-death mortality data in five periods (1971-1975, 1978-1982, 1988-1992, 1998-2002, 2008-2012) were extracted from the Taiwan national mortality data files. Township-level socio-economic variables were derived from five censuses in 1965, 1980, 1990, 2000, and 2010 and National Taxes Statistics in 2000 and 2010. We use the proportion of people with below college education as the main indicator of township-level socioeconomic status across all periods as it showed the strongest association with household income from tax data. In sensitivity analyses we used the proportion of agricultural workers in the working population and the first component derived from principle component analysis based on 13 socioeconomic variables. We calculated township-level period expected years of life lost (PEYLL) for all-cause, cause-specific and amenable mortality. We ranked the townships based on the socioeconomic indicator and then plotted the concentration curve and calculated the concentration index to investigate trend in health inequalities in Taiwan in 1971-2012. Results：The concentration indices for all-cause mortality were -0.100 (95% Confidence Interval (CI) -0.120, -0.080) in 1971-1975 and declined to -0.123 (-0.142, -0.104) in 1998-2002, and then slightly increased to -0.120 (-0.141, -0.100) in 2008-2012, indicating that the PEYLL tended to concentrate in areas with low socioeconomic position and the level of inequalities increased in 1978-2002 and became stable afterwards. There were marked differences in the patterns of different causes of death. For example, there was no evidence for inequalities for cancer mortality in 1971-1975 but the level of inequalities increased over the study period; for cardiovascular diseases the level of inequalities expended during 1971-2002 and became stable toward the end of the study period. The PEYLL of respiratory diseases and infectious and parasitic diseases also tended to concentrate in areas of low socioeconomic position, whilst their inequalities levels decreased over the study period. The level of inequalities for injuries and poisonings rose during 1971-2002 and then slightly decreased; within this category of deaths, socioeconomic inequalities for suicide decreased and disappeared gradually, whilst those for motor vehicle accident and other injuries and poisonings increased in 1971-2012. The concentration indices of amenable mortality were -0.126 (95% CI: -0.151, -0.100) in 1971-1975 and rose to -0.087 (-0.107，-0.066) in 1988-1992, and, after the implementation of Taiwan’s NHI in 1995, the concentration index was -0.090 (-0.111, -0.070), similar to the pre-NHI level, although it later increased to -0.081 (-0.102, -0.059) in 2008-2012. The finding indicated that the level of inequalities for amenable mortality reduced over the study period but the reduction was greater in the periods before than after the NHI. Conclusion：Over the last four decades, mortality tended to concentrate in deprived areas for most causes of deaths, and the inequalities level for all-cause mortality showed a rise in Taiwan. However, the inequalities level of amenable mortality had reduced, but there was no strong evidence for an effect of universal health coverage.