【作者】 刘宝；

【导师】 胡善联；

【作者基本信息】 复旦大学 ， 社会医学与卫生事业管理， 2003， 博士

【摘要】 人群自报健康的水平与分布研究[背景] 卫生系统的首要目标是促进人群的健康。人群健康评价是卫生政策过程乃至公共政策过程的关键环节。伴随流行病学模式的转变，单纯依靠死亡信息的人群健康综合评价已不能适应卫生系统决策和管理的需要，人群非死亡性健康状况的综合测量是形成完整的人群健康评价的重要组成部分。通过患者或调查对象自报或自评包括身体和心理等各方面的健康状况实现对健康状况的评价，是人群非死亡性健康信息收集的一个主要手段。但人群自报健康信息的应用受到自报健康资料跨人群可比性问题的局限，同时由于缺乏对人群自报健康水平与分布综合测量研究，通过人群自报健康资料获得的非死亡性健康信息不仅难以与实际卫生系统的管理和决策有机地结合，还可能进一步影响卫生系统绩效的改善。[目的] 抓住目前国际上关于非死亡性健康评价的研究热点，针对自报健康研究中面临的一些主要问题，利用上海市嘉定区、宝山区、金山区和崇明县4区（县）的家庭卫生服务调查资料，在解决自报健康资料跨人群可比性的基础上，实现对样本地区人群自健康水平和分布的综合测量和评价。具体研究目的如下：（1）人群自报健康资料的评估及跨人群可比的自报健康资料的获得；（2）人群自报健康水平的综合测量；（3）人群自报健康分布的综合测量；（4）人群自报健康水平与分布综合测量指标的结合表达。[方法]资料来源于2001年在上海市嘉定区、宝山区、金山区和崇明县共10358户家庭开展的家庭卫生服务调查。调查收集了18岁以上家庭成员关于活动能力、疼痛、情感、认知、自理能力和日常活动等6个方面的自报健康资料，以及家庭成员的社会经济和人口学背景信息，包括年龄、性别、教育程度和家庭收入和支出状况等。应用描述性分析以及计量心理学中的经典测试理论和项目应答理论的相关模型评估人群自报健康资料的潜因子结构特征和内部一致性，并结合应用健康情景考察自报健康资料的切点位移问题。通过HOPIT模型获得跨人群可比的人群<WP=5>自报健康资料（伤残测度），结合寿命表资料测算人群伤残调整期望寿命；并通过伤残集中指数和伪基尼系数测算和分解分析实现对人群自报健康分布（包括收入相关健康不平等和纯粹健康不平等）的综合评价；最后在人群自报健康水平和分布综合测量的基础上，应用理论推导和实证研究的方式考察自报健康水平和分布综合测量指标的结合表达以及两种类型表达形式的数学联系。[结果]· 人群自报健康资料的评估研究结果显示，按照年龄和性别结构计算的各区（县）调查样本离差指数没有系统性地偏离1，表明本次调查样本具有良好的代表性。就调查样本的社会人口学特征而言，各区（县）之间性别结构没有显著统计意义的差别，但在年龄结构和教育程度的分布上存在差别。缺失率分析的结果表明，4区（县）合计，6个健康维度中仅自理能力维度中的一个项目（5d）以及日常活动维度一个项目（6b）的项目缺失率较高，其余项目缺失率均低于1%；行缺失率分析结果显示，嘉定、宝山、金山和崇明4区（县）的行缺失率高于1/24（即6个健康维度共计24个项目中缺失超过1项）的对象占全体对象的比例，除宝山区超过2%外，其余3个区（县）均低于1%，4区（县）合计结果亦低于1%。传统计量心理学测试理论（CTT）分析的结果表明，除情感健康维度的项目外，其余健康维度的项目则较突出地反映了潜变量并且单维性较好；项目内部一致性分析提示了类似结果，4区（县）合计，项目内部一致性指标（Cronbach’s Alpha）达到0.9412，远高于人群自报健康问卷的标准（0.7）。在CTT分析的基础上，进一步开展的关于自报健康资料的项目应答理论（IRT）模型分析结果提示，各健康维度的潜变量效应不同，宜针对不同的健康维度分别进行IRT模型分析。最后，研究针对自报健康资料中存在的切点位移问题，应用健康情景分别对6个健康维度进行了切点位移评价，结果显示，地区差别是造成切点位移的关键因素，而且在各健康维度都不同程度地存在切点位移甚至“整体性切点位移”，提示解决自报健康资料的切点位移问题是改善自报健康资料跨人群以及人群内不同亚组之间可比性的关键。· 人群自报健康水平的综合测量研究表明，6个方面健康维度的原始自报健康值的年龄分布存在明显的“顶效应”，不同区（县）、不同性别、不同健康维度都一致地表现出这个特点，提示原始自报健康资料的可比性差。而结合应用健康情景的HOPIT模型对切点位移进行校正的结果则有效地弱化了“顶效应”的影响，在很大程度上改善了自报健<WP=6>康资料的可比性。结合人群年龄组综合伤残测度与寿命表获得的4区（县）伤残调整期望寿命（DALE）结果提示，嘉定、宝山、金山和崇明4区（县）的男女合计20岁时DALE分别为39.9、42.1、41.2和41.7岁，而相应因伤残而损失的健康寿命（DLE）则分别为19.4、16.0、17.0和16.2岁。DALE指标值低于传统的期望寿命指标值，而且由于DALE综合考虑了死亡和伤残因素的影响，不同人群之间DALE的比较结果与应用期望寿命进行的比较的结果并不一致，如嘉定区20岁和60岁成年人期望寿命更多还原

【Abstract】 A Study on the Level and Distribution of Population Self-reported Health [BACKGROUND]The primary purpose of a health system is to improve the population health. Population health assessment, so, is the key to inform the health policy process or even to the public policy process. The traditional population health measure relying heavily on vital statistics is becoming increasingly uninformative because of the epidemiological transition. In this context, non-fatal health outcomes, as the complement to the previous mortality information, should be include to form a summary measure of population health. Self-reported health survey is one of key methods to collect data on the population non-fatal health status. However, the use of self-reported health information is quite limited due to its problem in cross-population comparability. In addition, rare study on summary measure of level and distribution of self-reported health also prevents its integration into the health policy making and its possible benefits to improve the health system performance. [OBJECTIVE]To capture the international research focus on non-fatal health assessment and some main issues facing self-reported health data, the study, by using the household survey data collected from 4 districts (Jiading, Baoshan, Jinshan and Chongming) of Shanghai municipality, studied the summary measure of level and distribution of population self-reported health on the basis of solving the problem of cross-population comparability of self-reported health data. The specific objectives are as follows:(1) To assess the population self-reported health data and to obtain the cross-population comparable self-reported health data;(2) Summary measure of the level of population self-reported health;(3) Summary measure of the distribution of population self-reported health;(4) Combination of the summary measure of the level and distribution of population self-reported health.<WP=9>[METHODS]The data came from the 2001 household survey conducted in the 4 districts (Jiading, Baoshan, Jinshan and Chongming) of Shanghai municipality. 10358 households were interviewed by trained interviewer. The content of the data includes the self-reported health and vignette information covering 6 core domains of health (mobility, pain, affection, cognition, self-care and usual activities) of each household member aged over 18. Socioeconomic and demographic information such as age, sex, education and household income and expenditure were also collected. Descriptive analysis and Classical Test Theory (CTT) and Item Response Theory (IRT) in psychometrics were used to assess the latent factor structure and internal consistency of the self-reported health data. The cut-point shift issue was also investigated by using health vignettes. Hierarchical Ordered Probit (HOPIT) model was used to calculate the severity-weighted prevalence of disability of the sample population, which was then combined with the period life table to obtain the Disability-Adjusted Life Expectancy (DALE); concentration index, pseudo-Gini coefficient, and their decomposition were used for obtaining the summary measure of the distribution of population self-reported health; and finally, based on the summary measure of the level and distribution of population self-reported health, the thesis discussed the theoretical construction of the combination of the summary measure of level and distribution, and also, mathematical relationship between the two kinds of combination methods were investigated and demonstrated with empirical evidence. [RESULTS]· Assessment of population self-reported health dataThe results of the study suggest that，the sample deviation indices calculated according to age and sex for each of the 4 sample districts are not systematically deviate from 1, indicating the good representation of the sample. The descriptive analysis of the socio-demographic information does not show the significant difference of sex, but with significant difference of age and更多还原