结构方程模型在中国居民自报健康状况评价中的应用

目的分析居民自报健康、真实健康和健康态度之间的潜在关系,解决自报健康跨人群不可比性问题,实现人群健康的综合评价。方法利用2007年中国居民自报健康抽样调查数据,将健康情景评价作为对健康态度的测量,应用结构方程模型的方法,构建自报健康、真实健康和健康态度之间相互关系的理论模型。结果居民自报健康结果除了受真实健康因子的支配外,还受个人健康态度的影响。调整健康态度的影响后,年龄、性别、文化程度、城乡、中部和西部、经济状况等因素的标准化回归系数分别为-0.0765（t=-6.3349）,0.0211（t=2.8827）,-0.0014（t=-0.7853）,0.0275（t=2.5759）,0.0491（t=5.4676）,-0.0092（t=-0.7778）,0.0701（t=7.7376）;而真实健康因子与年龄、性别、城乡、经济状况的标准化回归系数则分别为-0.2360（t=-12.3360）,0.0016（t=0.0551）,0.1269（t=8.3125）,0.0421（t=4.0852）,-0.0508（t=-4.3965）,-0.1653（t=-10.1801）,0.0760（t=7.2493）。结论结构方程模型可以应用于人群自报健康状况资料的分析,消除健康态度的影响后获得真实健康因子,后者比自报健康本身更具有可比性,从而客观评价不同人群真实健康状况。

Application of structural equation model in measurement of self-reported health status in Chinese residents

Objective To discuss and analyze the relationship among serf-reported health, true health status and health attitudes, and to explore the method of assessing the self-reported health status of Chinese residents comparably. Methods The data of the self-reported health survey in Chinese residents in 2007 was used to make the evaluation of health vignettes as measures of respondents＇ health attitudes. A theoretical structural equation model （SEM） was built to describe the relationship among self-reported health, true health status, and health attitudes. Results The data was fitted well, and the results were reasonable. Beside true health status, the self-reported health was exactly influenced by personal health attitudes. After adjusting the effect of attitudes, the standard estimates of β were - 0. 0765 （ t = - 6. 3349）, 0.0211 （t=2.8827）, -0.0014 （t=-0.7853）, 0.0275 （t=2.5759）, 0.0491 （t=5.4676）, -0.0092 （t=- 0. 7778）, and 0. 0701 （t =7. 7376） for effects of age, sex, education, rural vs city, central vs eastern provinces, western vs eastern provinces, income, respectively. However, correspondingly, for the true health which was subtracted the im- pacts of personal attitudes from the self- reported health, the standard estimates of β were -0. 2360 （t = - 12. 3360）, 0.0016 （t=0.0551）,0.1269 （t=8.3125）, 0.0421 （t=4.0852）, -0.0508 （t= -4.3965）, -0.1653 （t= - 10. 1801 ）, and 0. 0760 （t = 7. 2493）, respectively. Conclusion SEM can be applied in analyses and measurement of population self-reported health. It can adjust kinds of influencing factors, abstract the true health factor, and then evaluate self-reported health status among different population objectively.