Jelles F, Van Bennekom C A, Lankhorst G J, Sibbel C J, Bouter L M
Free University Hospital, Department of Rehabilitation Medicine, Amsterdam, The Netherlands.
J Clin Epidemiol. 1995 Mar;48(3):407-16. doi: 10.1016/0895-4356(94)00152-g.
The objective of the study was to determine the inter- and intra-rater agreement of the Rehabilitation Activities Profile (RAP). The RAP is an assessment method that covers the domains of communication, mobility, personal care, occupation and relationships. Each domain consists of items which are further divided in sub-items for in-depth analysis. The RAP allows quantification of the severity of disabilities, handicaps and perceived problems of a patient with regard to the items and sub-items. For this purpose ordinal 4-point Likert scales were constructed. The RAP can be used for goal setting and evaluation of rehabilitation. Because of the broad intended use of the RAP and its construction, a special design for the reliability study was needed. The study was carried out in 5 rehabilitation facilities with the participation of various professions. The items and sub-items of the RAP were divided over these professions according to their expertise. Pairs of interviewers were formed that questioned a patient. For the determination of inter- and intra-rater agreement each pair of interviewers was allowed to question a patient only once. To establish the intra-rater agreement, video recordings were made during the interviews. The median (weighted) kappa value and percentage of agreement about the severity grading of a disability or handicap for all items and sub-items exceeded 0.84 and 81%, respectively, with regard to the inter- and intra-rater agreement. For the severity grading of perceived problems these values were 0.91 and 86%. The interpretation of kappa was hindered by two paradoxes recently described in the literature. The paradox "high agreement but low kappa" manifested itself in particular. It is concluded that inter- and intra-rater agreement of the RAP can be considered to be good to very good.
本研究的目的是确定康复活动概况（RAP）在评估者间和评估者内的一致性。RAP是一种评估方法，涵盖沟通、移动性、个人护理、职业和人际关系等领域。每个领域由项目组成，这些项目又进一步细分为子项目以进行深入分析。RAP允许对患者在项目和子项目方面的残疾、障碍及感知问题的严重程度进行量化。为此构建了4级有序李克特量表。RAP可用于康复目标设定和评估。由于RAP的广泛预期用途及其构建方式，需要为可靠性研究设计一种特殊方案。该研究在5个康复机构开展，有不同专业人员参与。RAP的项目和子项目根据专业人员的专业知识分配给他们。组成访谈者对来询问一名患者。为了确定评估者间和评估者内的一致性，每对访谈者只能询问患者一次。为了确定评估者内的一致性，在访谈期间进行了录像。就评估者间和评估者内的一致性而言，所有项目和子项目关于残疾或障碍严重程度分级的中位数（加权）kappa值和一致性百分比分别超过0.84和81%。对于感知问题的严重程度分级，这些值分别为0.91和86%。kappa的解释受到文献中最近描述的两个悖论的阻碍。特别是出现了“高一致性但低kappa”的悖论。结论是，RAP在评估者间和评估者内的一致性可被认为是良好到非常好。
