Test-retest reliability of assessing psychiatrically ill patients in a multi-center design.
Test-retest reliability of assessing psychiatrically ill patients in a multi-center design.
复制标题
在多中心设计中评估精神疾病患者的重测可靠性。
DOI：
10.1016/0022-3956(81)90016-9
复制
发表时间：
1981
影响因子：
4.8
通讯作者：
McDonald-Scott,P
中科院分区：
文献类型：
--
作者：
Keller,MB;Lavori,PW;Andreasen,NC;Grove,WM;Shapiro,RW;Scheftner,W;McDonald-Scott,P
方法SADS是一种记录精神疾病患者描述性和诊断性数据的方法。它的格式是一份面试指南，由一系列问题、项目和标准组成。在实际操作中，SADS评分和RDC诊断是根据在患者面谈期间以及从朋友、亲戚、治疗人员和患者记录中收集的信息做出的。第一部分详细描述了当前疾病发作的特点，并描述了精神病理学的严重性。访谈引出的信息使评分者能够做出明确的标准诊断，建立精神病理学的维度标尺，并允许在不同的症状特征中进行现象学分析。4这是一个复杂的、结构化的面试，它有许多有序和无序的分类项目，有比例的项目，以及在没有某些症状的情况下跳过面试部分的项目。这些都不能用相同的统计方法来分析；例如，在无序分类项目“当前条件的分类”上，不同组之间的差异不能通过方差分析来分析。再举一个例子，SADS第234项“主观忧郁感”作为两个后续项目的跳过项，这影响了它们的可靠性。为了估计这类复杂项目的信度系数，我们采用了不同的策略，包括用多变量替换无序类别，以及将似是而非的值归因于跳过的项目。
MethodsThe SADS was developed as a method of recording descriptive and diagnostic data on psychiatric patients. Its format is an interview guide organized as a progression of questions, items and criteria. In actual practice SADS ratings and RDC diagnoses are made on the basis of information collected during the patient interview and from friends, relatives, treating staff and the patient records. The first part provides detailed descriptions of the features of current episodes of illness and descriptions of the severity of psychopathology. The interview elicits information that enables the rater to make categorical criterion diagnoses, establishes dimensional scales of psychopathology, and permits phenomenological analyses within different symptom profiles. 4 It is a complex, structured interview which has many ordered and unordered categorical items, scaled items and items which skip interview sections if certain symptoms are absent. These are not all analyzable by the same statistical methods; for example, differences between groups on the unordered categorical item “classification of current condition” cannot be analyzed by an analysis of variance. For another example, SADS item 234,“Subjective feelings of depression” acts as a skipout for two subsequent items, which affects their reliability. To estimate reliability coefficients for such complex items we employed different strategies including replacing unordered categories by multiple variables and imputing plausible values to skipped items.