American Psychiatric Association, Division of Research and American Psychiatric Institute for Research and Education, Arlington, Va., USA.
Am J Psychiatry. 2013 Jan;170(1):71-82. doi: 10.1176/appi.ajp.2012.12071000.
The authors sought to document, in adult and pediatric patient populations, the development, descriptive statistics,and test-retest reliability of cross-cutting symptom measures proposed for inclusion in DSM-5.
Data were collected as part of the multisite DSM-5 Field Trials in large academic settings. There were seven sites focusing on adult patients and four sites focusing on child and adolescent patients.Cross-cutting symptom measures were self-completed by the patient or an informant before the test and the retest interviews, which were conducted from 4 hours to 2 weeks apart. Clinician-report measures were completed during or after the clinical diagnostic interviews. Informants included adult patients, child patients age 11 and older, parents of all child patients age 6 and older, and legal guardians for adult patients unable to self-complete the measures. Study patients were sampled in a stratified design,and sampling weights were used in data analyses. The mean scores and standard deviations were computed and pooled across adult and child sites. Reliabilities were reported as pooled intraclass correlation coefficients (ICCs) with 95% confidence intervals.
In adults, test-retest reliabilities of the cross-cutting symptom items generally were good to excellent. At the child and adolescent sites, parents were also reliablereporters of their children’s symptoms,with few exceptions. Reliabilities were not as uniformly good for child respondents, and ICCs for several items fell into the questionable range in this age group. Clinicians rated psychosis with good reliability in adult patients but were less reliable in assessing clinical domains related to psychosis in children and to suicide in all age groups.
These results show promising test-retest reliability results for this group of assessments, many of which are newly developed or have not been previously tested in psychiatric populations
作者旨在记录跨症状测量的发展、描述性统计和重测信度，这些测量被提议纳入 DSM-5，对象为成年和儿科患者群体。
数据是作为多地点 DSM-5 现场试验的一部分在大型学术环境中收集的。有七个成人患者的地点和四个儿童和青少年患者的地点。跨症状测量是在测试前和测试后的访谈中由患者或知情者自我完成的，两次访谈之间的间隔时间为 4 小时至 2 周。临床报告测量是在临床诊断访谈期间或之后完成的。知情者包括成年患者、年龄在 11 岁及以上的儿童患者、所有 6 岁及以上儿童患者的父母，以及不能自我完成测量的成年患者的法定监护人。研究患者采用分层设计抽样，数据分析中使用了抽样权重。平均分数和标准差是在成人和儿童地点之间计算和汇总的。可靠性报告为汇总的组内相关系数（ICC），置信区间为 95%。
在成年人中，跨症状项目的重测信度通常较好至极好。在儿童和青少年地点，父母也是其子女症状的可靠报告者，除了少数例外。儿童的可靠性并非如此一致良好，几个项目的 ICC 在该年龄组中落入可疑范围。临床医生在成年患者中评估精神病的可靠性较好，但在评估儿童的精神病相关临床领域和所有年龄段的自杀方面的可靠性较差。
这些结果表明，这组评估具有有希望的重测信度结果，其中许多是新开发的，或以前没有在精神科人群中进行过测试。
