【作者】 赵倩倩；

【导师】 施学忠；

【作者基本信息】 郑州大学 ， 流行病与卫生统计学， 2011， 硕士

【摘要】 卫生系统反应性是指卫生系统对公众非医疗因素方面普遍、合理期望的认知和适当反应,主要包括尊严、保密性、自主性、及时性、选择性、社会支持和基础设施质量等7个方面。卫生系统反应性数据往往具有地区等层次结构,当采用以往的传统统计分析方法进行反应性影响因素研究时,可能导致参数估计和标准误出现偏倚。为了进一步分析地区等因素在卫生系统反应性中的作用,获得更为准确的影响因素估计,该研究先采用因子分析方法计算河南省卫生系统反应性评分,然后采用二水平模型分析反应性的影响因素,为同类研究提供方法学参考。目的通过对河南省卫生系统反应性资料的分析,探讨反应性中地区因素的作用大小及其主要影响因素,为卫生系统建设提供基本信息和科学依据,也为卫生系统反应性研究提供方法学参考。研究对象与方法采用2000年世界卫生组织《健康和卫生系统反应性调查》问卷,使用多阶段分层整群随机抽样的方法,在2000年河南省16个城市调查年满18周岁的居民3000人,选取过去的12个月中在医疗机构接受过医疗服务,且完成卫生系统反应性问卷调查的1311人进行分析。采用SPSS16.0进行研究对象一般情况的统计描述和卫生系统反应性评分的因子分析；采用MLwiN2.02进行反应性影响因素的二水平模型拟合,参数估计采用限制性迭代广义最小二乘法,残差方差与回归系数的假设检验采用Wald检验,不同模型效果的比较采用-2对数似然函数；检验水准α=0.05。结果1.河南省卫生系统反应性的7个方面中,评价最好的是尊严,“很好”和“好”的比例为70.8%；评价最差的是基础设施质量,“很好”和“好”比例为32.9%。2.反应性数据分布存在层次结构,组内相关系数ICC (intra-class correlation)为14.2%,水平2(地区)残差的方差σu02的假设检验χ2=5.76,P=0.016,σu02≠0。3.二水平方差成分模型组内相关系数为13.97%,比零模型有所降低,水平2(地区)残差的方差σu02的假设检验χ2=5.661,P=0.017,σu02≠0.4.年龄较大(60-69岁)、饮酒频率越高、参与社会活动有轻度困难、拥有门诊保险相对于拥有住院保险的人群对反应性评价较低,回归系数β分别为0.231、0.086、0.119.0.488, Wald检验P<0.05,有统计学意义。5.二水平方差成分模型比一水平线性回归模型的各参数估计的标准误更小,参数估计更加精确；前者的残差方差比后者减小,-2对数似然函数值降低了137.436,根据χ2分布,P<0.001,差异有统计学意义,二水平方差成分模型的拟合结果更优。结论1.河南省卫生系统反应性的分布存在地区聚集性,数据具有层次结构特征。2.采用多水平模型进行反应性影响因素分析比传统模型效果更优。3.年龄较大、饮酒频率越高、拥有门诊保险、参与社会活动有轻度困难的人群对河南省卫生系统反应性评价较低,应引起卫生工作者的关注。更多还原

【Abstract】 Health system responsiveness was defined as the responsiveness of health system to general legitimate non-health expectations in the population, including 7 aspects:dignity, confidentiality, autonomy, prompt attention, choice of providers, social support networks and quality of basic amenities. The data of health system responsiveness usually has regional hierarchical structures, and if we use traditional statistical methods to investigate its influencing factors, it might lead to bias of parameter estimation and standard error. For further analysis of the region effect in health system responsiveness and accurate estimation of influencing factors, the study calculated scores of health system responsiveness in Henan province by factor analysis, and then fitted two level models for influencing factors so as to provide methodological reference for similar research.ObjectiveTo investigate the influencing factors of health system responsiveness in Henan province, and to explore the region effect in the responsiveness and its main influencing factors, so as to provide basic information and scientific basis for the construction of health system, also to provide a methodological reference for the study of health system responsiveness. Subjects and methodsBy multi-stage stratified cluster random sampling technique,3000 residents aged over 18 years old were sampled from 16 cities of Henan province.1311 residents, who received service in medical clinics during the past 12 months and fulfilled questionnaire on health system responsiveness, were investigated.Statistical description of general information and factor analysis of health system responsiveness score were done under SPSS16.0. Two level models for influencing factors were fitted under MLwiN2.02. Parameters were estimated by restricted iterative generalized least squares, hypothesis test for residual variance and regression coefficients were done by Wald test, and -21oglikelihood function was used to compare the effect of different models. The significant levelαis 0.05. Results1. In 7 aspects of health system responsiveness in Henan province, the best evaluation was given to dignity, "very good" and "good " was added up to 70.8%; the lowest evaluation was given to quality of basic amenities, the proportion of "very good" and "good " was only 32.9%.2. The distribution of responsiveness had regional hierarchical structures. Intra-class correlation coefficient (ICC) was 14.2%. The hypothesis test of variance of level 2(cities)’s residual got a significant result:x2=5.76, P=0.016,δuo2≠0.3. ICC in two level variance component model was 13.97%, lower than that in zero model. The hypothesis test of variance of level 2’s residual got a significant result: x2=5.661,P=0.017,δuo2≠0.4. Four kinds of population (with age of 60-69 years, frequent alcohol drinking, having slight difficulties in social activities, having outpatient insurance) evaluated the responsiveness lower, with regression coefficients 0.231,0.086,0.119,0.488 (p <0.05 for Wald test).5. The standard errors of parameter estimation in two level variance component model were smaller than those in one level linear regression model, and the former model got more precise parameter estimation. The residual variance reduced in the former model, and-21oglikelihood function value decreased by 137.436, according to x2 distribution, p<0.001. The two level variance component model fitted better.Conclusions1. The distribution of health system responsiveness in Henan province has region cluster, and the data has hierarchical structure.2. Multilevel model is better than traditional model in analysis of influencing factors of health system responsiveness.3. Age, medical insurance form, degree of difficulties in social activities, frequency of drinking are influencing factors of health system responsiveness in Henan province, which should be the concern of health workers.更多还原