目標：探討台灣不同世代老人存活趨勢變化差異及影響老年人口存活相關因子。方法：本
研究利用國民健康局「台灣地區中老年人身心社會狀況長期追蹤研究調查」自1989年至2007年
之六次訪查資料，以Cox脆弱加權模式分析探討老年人口存活相關因素並比較不同出生A、B世
代(1915-1929、1929-1943)分別於1989-1993年及2003-2007年存活趨勢變化差異。結果：就兩世
代樣本特性言，以性別及族群差異較大，其他影響世代存活相關因子間差異，A世代抽菸人數
與體能、自評健康狀況及自認經濟狀況好者較B世代為多，顯現世代間之健康行為及其對身心
狀況的感受亦有所差異；罹患糖尿病與高血壓的比例B世代較A世代為多，具呼吸疾病的比例
則以A世代老人較高。其餘疾病如中風、心臟疾病及胃潰瘍或胃病等差異不大。另經統計模式
選擇結果發現年齡、住地類型、自評健康、體能狀況、有無糖尿病及抽菸為影響台灣不同世代
老人存活趨勢變化差異之重要指標。結論：在不同世代存活4年趨勢變化差異上，較為年輕B世
代老人，由於經濟情況的改善及全民健保的實施，相較於A世代老人，生活條件較好並享有較
為優質的醫療照顧；經相關變項調整後，B世代死亡風險為A世代的0.601倍；兩世代老人的存
活情況，受性別、族群、嚼檳榔及社團參與的影響不大，糖尿病則為影響不同世代60-74歲老
人存活之主要慢性疾病，然未來時日增長，增加較多調查數據時，情況亦可能有所變化，有待
進一步探討。(台灣衛誌 2012；31(6)：597-611)

Objectives: To investigate the factors related to the survival rates of the elderly in two
different cohorts in Taiwan. Methods: Data derived from six waves of the survey of Health and
Living Status of the Elderly in Taiwan from 1989 to 2007 were used to compare the survival
rates of the elderly during 1989-1993 and 2003-2007 between two different birth cohorts
(1915-1929 – Cohort A and 1929-1943 – Cohort B). The weighted Cox frailty model with time-
dependent covariates was used for data analysis. Results: There were differences in gender
and ethnicity between two cohorts. The rates of smoking, good physical condition, self-rated
health and economic situation were higher in cohort A than in cohort B. The rates of diabetes
and hypertension were higher in cohort B than in cohort A. Conversely, the rate of respiratory
system diseases was higher in cohort A. For other chronic diseases, e.g. stroke, heart disease,
gastric ulcers, and stomach ailments, there were slight difference between the two cohorts. Based
on Cox model selection, age, residential location, self-rated health, physical function, diabetes,
and smoking were significant factors related to the survival of the elderly in both cohorts.
Conclusions: By the results of Cox frailty model, four-year survival rates differ a lot between
cohort A and cohort B. This difference might be due to improved economic conditions and the
implementation of national health insurance, which resulted in a better living standard and medical
care for the elderly in cohort B compared to cohort A. In addition, after adjusting for related
covariates, the risk ratio of cohort B to cohort A was about 0.601. Gender, ethnicity, chewing betel
nut, and social activity were insignifcant for both cohorts. Diabetes was the main disease related to
survival for both cohorts. Some factors may become signifcant if more data is included in a future
study. This issue requires further investigation. (Taiwan J Public Health. 2012;31(6):597-611)