Abstract: Background　Coronary heart disease is difficult to cure，and the course of which is long.Traditional evaluation indexes （such as mortality and cure rate） cannot evaluate the health status and treatment effect on patients.Patient reported outcome is the direct evaluation of patients' own health status without interpretation of medical professionals.The key to the assessment of patient reported outcome is to develop an appropriate measurement scale and evaluate it accordingly．Objective　To evaluate the reliability of the Patient Reported Outcome Instruments for Chronic Disease （PROISCD-CHD）．Methods　To evaluate the reliability of PROISCD-CHD by classical test theory （CTT） and modern test theory （Generalizability Theory，GT），a total of 189 patients with coronary heart disease who were literate and volunteered were investigated at the time of admission and 75 of them were randomly selected for the second survey after one day interval.The correlation coefficients of the first and second measurements were used to evaluate the test-retest reliability while Cronbach's α coefficients were used to evaluate internal consistency.The source of the patient's reported outcome score was explored in G-study and the suggested number of items was given based on their G-coefficient （Ep2） and indexes of dependability （φ） in D-study．Results　Coefficients of physiological health，mental health，social health，spirit/belief health，common modules，specific modules and total score of PROISCD-CHD scale were 0.89，0.84，0.75，0.86，0.88，0.89，0.71，respectively （P<0.001），and Cronbach's α coefficients of each item were 0.71，0.75，0.75，0.69，0.88，0.78，0.89，respectively.G-study showed that the biggest source of variation was the interaction between the patients and items，and the smallest was the item.D-study revealed that G-coefficients of all domains were greater than 0.70，except for the domain of spirit/belief health，which had the same result with the reliability analysis of classical test theory.And according to Ep2 andφ，the optimal number of items was also given．Conclusion　PROISCD-CHD can be used to assess patient-reported outcome for patients with coronary heart disease with good reliability.

Key words: Coronary disease, Patient reported outcome, Reliability

摘要： 背景　冠心病病程慢性迁延，较难治愈，传统的评价指标（如病死率、治愈率）等不能很好地评价其健康状况和治疗效果。患者报告结局是未经医疗专业人士解释的患者对自身健康状况、直接的评价，患者报告结局测定的关键是制定合适的患者报告结局测定量表并对其评价。目的　分析冠心病患者报告结局量表（PROISCD-CHD）的信度。方法　选取2015年3月—2017年1月广东医科大学附属医院心血管内科住院的有一定文化程度且自愿填写PROISCD-CHD的冠心病患者189例为调查对象，间隔1 d后随机选取75例冠心病患者进行第2次调查。采用经典测量理论与现代测量理论（概化理论）相结合的方法对其进行信度考评。重测信度采用第1、2次问卷调查的相关系数；内部一致性采用Cronbach'sα系数。G研究计算变异的来源；D研究根据概化系数（Ep2）和可靠性指数（φ）给出适宜的条目建议个数。结果　PROISCD-CHD生理健康、心理健康、社会健康、精神/信仰健康、共性模块、特异模块和量表总分的相关系数分别为0.89、0.84、0.75、0.86、0.88、0.89、0.71（P<0.001），Cronbach'sα系数分别为0.71、0.75、0.75、0.69、0.88、0.78、0.89。G研究结果显示，变异的来源中患者与条目的交互作用所占的比重最大，条目所占的比重最小。D研究结果显示：除精神/信仰健康领域外，其余领域的Ep2均>0.70，与经典测量理论对信度的分析一致。根据Ep2和φ，给出了适宜的条目建议个数。结论　PROISCD-CHD具有良好的信度，可以作为我国冠心病患者报告结局的测评工具。

关键词: 冠心病, 患者报告结局, 信度