Modeling the Ageing Effect on Changes of Health Index: a Case Study

• Main Authors: Ting-Chun Chu, 朱庭君
• Other Authors: I-Feng Lin
• Format: Others
• Language: zh-TW
• Published: 2018
• Online Access: http://ndltd.ncl.edu.tw/handle/3ap28p

碩士 === 國立陽明大學 === 公共衛生研究所 === 106 === Backgrounds and Objective: Since the end of March 2017, Taiwan has entered the stage of an “aging society”. In the face of the rapid growth of aging population in Taiwan, it is urgently significant to understand the changes of health status for the elderly. The purpose of this study is to grasp a better picture of the long-term changes of the elderly’s health condition by tracking how those common health indicators of the elderly change over time. Materials and Methods: The data for this study were obtained from Taipei City Elderly Health Examination Database from 2005 to2010. Among the 90,949 participants aged from 65 to 90 years old, 63,446 who participated the health exams at least 4 times were analyzed using the mixed models (LMM) and GEE models, and 9,719 who participated the exams at all six continued years from 2005 to 2010 were analyzed using the grouped-based trajectory modeling (GBTM). Both cross-sectional and longitudinal ageing effects on the developmental trajectory of health indicators, such as BMI and total cholesterol, were explored. Results and Conclusions: The cross-sectional analysis showed that the older the examination age, the average BMI and total cholesterol were lower. Based on the longitudinal analysis with linear mixed and GEE models, there were significant interaction effects between the first examination age and the follow-up times (the time between the first examination age and the follow-up age). As older as the first examination age was, we found that the declined trend of total cholesterol tended to be steeper. On the other hands, the rate of change as age increased for BMI remained stable. In a subset analysis for those who continued to participate health exams for all six years, the group-based trajectory model could further divide the elderly into three different parallel trajectory groups: the healthy group, the slightly overweight group and the mild-obese group according to their BMI. Although the group-based trajectory model were widely used in panel data without missing values and can help researchers identify and profile the characteristics of each subgroup’s trajectory within the group, it is difficult to carry out the effective analysis and inference when the data consist of too many missing values. In the case of the elderly health examination, the information such as different initial age of each sample or his/her different health condition tracking frequency was not the same are all possible missing value to disturb the analysis. In this case, it would be better to apply linear mixed model or generalized estimating equation models with interactions between the first examination age and follow-up times for the analysis mentioned above.