摘要
目的基于多维指标构建治疗质量综合评价模型,探讨模型估计治疗质量的有效性;综合评价医院、医生乳腺癌的治疗质量。方法构建两参数logistic潜变量模型作为测量模型,将测量模型中的潜变量作为因变量整合进多水平结构模型,设定虚拟水平代表多维潜变量,构建多水平结构模型。应用MCMC估计模型参数,拟合三水平二维模型分析实例数据。结果模拟研究结果提示,评价指标数及医生数均为20,模型估计治疗质量的有效性最好;应用该模型分析可以获得乳腺癌诊断、手术操作质量较优的医生及医院。结论多水平多维潜变量模型可以分析具有层次结构和多维性的治疗质量评价数据,能够做到全面、合理的进行疾病治疗质量综合评价;治疗质量综合评价使医院的等级划分、医生的绩效考核与治疗质量的衔接成为可能。
Objective The study aimed to construct a composite score method based on multidimensional quality indi-cators and conduct simulation trials to validate the method. Besides,the quality of breast cancer care for both hospitals and sur-geons was evaluated by the method. Methods The two-parameter logistic latent variable model was constructed as measure-ment model;the latent variables in the measurement model were further incorporated into multilevel structural model as depend-ent variables and one pseudo level was designed for representing multiple latent variables. MCMC method was used to estimate model parameters. Three level and two-dimensional latent variable model was used to analyze the actual data. Results The sim-ulation study showed that the number of quality indicators and surgeons should not be less than 20 to obtain efficient estimate of quality of care;the multilevel and multidimensional latent variable model was applied to analyze the data;surgeons and hospitals who provided superior quality of breast cancer diagnosis and operative procedure were obtained. Conclusion The newly con-structed multilevel and multidimensional latent variable model could effectively address the hieratical structure in quality of care data as well as the multidimensional nature of quality of care,thus,the model can be used to comprehensively and rationally as-sess the quality of care;comprehensive evaluation of quality of care provided ground for linking the ranking of hospitals and per-formance appraisal of surgeons to the quality of care.
出处
《中国卫生统计》
CSCD
北大核心
2017年第5期700-704,共5页
Chinese Journal of Health Statistics
基金
国家自然科学基金(81273183)
关键词
治疗质量
综合评价
多水平多维
模拟研究
Quality of care
Comprehensive evaluation
Multilevel and multidimensional
Simulation study
