Støre-Valen Jakob, Ryum Truls, Pedersen Geir A F, Pripp Are H, Jose Paul E, Karterud Sigmund
Department of Psychology, Norwegian University of Science and Technology.
Department of Personality Psychiatry, Division of Mental Health and Addiction, Oslo University Hospital.
Psychol Assess. 2015 Sep;27(3):865-73. doi: 10.1037/pas0000086. Epub 2015 Feb 2.
The Global Assessment of Functioning (GAF) Scale is used in routine clinical practice and research to estimate symptom and functional severity and longitudinal change. Concerns about poor interrater reliability have been raised, and the present study evaluated the effect of a Web-based GAF training program designed to improve interrater reliability in routine clinical practice. Clinicians rated up to 20 vignettes online, and received deviation scores as immediate feedback (i.e., own scores compared with expert raters) after each rating. Growth curves of absolute SD scores across the vignettes were modeled. A linear mixed effects model, using the clinician's deviation scores from expert raters as the dependent variable, indicated an improvement in reliability during training. Moderation by content of scale (symptoms; functioning), scale range (average; extreme), previous experience with GAF rating, profession, and postgraduate training were assessed. Training reduced deviation scores for inexperienced GAF raters, for individuals in clinical professions other than nursing and medicine, and for individuals with no postgraduate specialization. In addition, training was most beneficial for cases with average severity of symptoms compared with cases with extreme severity. The results support the use of Web-based training with feedback routines as a means to improve the reliability of GAF ratings performed by clinicians in mental health practice. These results especially pertain to clinicians in mental health practice who do not have a masters or doctoral degree.
全球功能评估（GAF）量表在常规临床实践和研究中用于评估症状及功能的严重程度和纵向变化。人们对评定者间信度较低表示担忧，本研究评估了一个基于网络的GAF培训项目在常规临床实践中提高评定者间信度的效果。临床医生在线对多达20个病例 vignette 进行评分，并在每次评分后收到偏差分数作为即时反馈（即自己的分数与专家评定者的分数相比较）。对各个病例 vignette 的绝对标准差分数的增长曲线进行了建模。一个线性混合效应模型，以临床医生与专家评定者的偏差分数作为因变量，表明在培训期间信度有所提高。评估了量表内容（症状；功能）、量表范围（平均；极端）、以往GAF评分经验、职业和研究生培训的调节作用。培训降低了无经验的GAF评定者、护理和医学以外临床专业人员以及无研究生专业背景人员的偏差分数。此外，与症状极端严重的病例相比，培训对症状严重程度为平均水平的病例最为有益。结果支持使用带有反馈程序的基于网络的培训，作为提高心理健康实践中临床医生进行GAF评分信度的一种手段。这些结果尤其适用于没有硕士或博士学位的心理健康实践临床医生。
