Inter-rater Reliability of the Chinese Version of the Schedules for Clinical Assessment in Neuropsychiatry (SCAN) in Taiwan

　　目的：探討臺灣SCAN 2.1中文版評估者間的信度。方法：根據世界衛生組織(WHO)制訂的訓練課程，我們在臺灣舉辦為期5天的工作坊，內容包括授課與討論、錄影帶面談示範、實際病人的面談訓練與督導。課程講師由接受過SCAN訓練的四名精神科專科醫師擔任，他們皆有使用SCAN進行流行病學調查的經驗。學員包括7名精神科醫師、1名臨床心理師和6名精神科護理師。訓練課程結束後，所有學員觀看了含各類診斷之23名個案訪談錄影帶，然後依據錄影帶書面記錄作SCAN評分。這23名個案包含9名門診病人、8名住院病人和6名社區個案。評分結果以症狀嚴重度之量性評估與症狀呈現與否的質性評估兩種方法計算信度，並分成講師與學員兩組做比較。針對各個題項以及每一章節的全部題項評分總和，分別計算各種信度值（包含平均值、呈現百分率、一致率、組間一致率及Kappa值)。每一章節的全部題項評分總和之各種信度值，則以計算該章節所有題項的上述值之總平均值（grand mean)為之。結果：依題項症狀有無作檢定，所有評估者的一致率達61-100%，kappa值為0.47-0.88，在六（憂鬱情緒與意念)、七（思考、注意力、精力和興趣)。八（身體功能)三個章節的總分，學員組的評分比講師組高，但若以題項症狀有無作比較，則兩組無差異。結論：在接受過完整訓練後，精神科醫療人員使用SCAN作臨床會談評估時，信度相當良好。
　　Objective: To assess the item-level inter-rater reliability of the Chinese version of the Schedules for Clinical Assessment in Neuropsychiatry (SCAN) comparing a group of trainers and trainees in Taiwan. Methods: A WHO- format SCAN-training course was held by the Taiwan SCAN Club in Taipei in 2000. The trainers included four psychiatrists qualified in SCAN training and experienced in large-scale epidemiological survey using the instrument. The trainees included seven psychiatrists, one clinical psychologist and six psychiatric nurses. A reliability study was when conducted by all the trainers and trainees who assessed 23 videotaped interviews with the Chinese version of SCAN-2.1. The 23 study subjects included eight psychiatric inpatients, nine outpatients and six community respondents. The between-group mean scores for the scaled ratings were compared using the paired-t-test. The binary transformed item ratings representing the presence or absence of clinically significant symptoms (0-1) were treated with percentages of presence and agreement, and the kappa coefficient; then the percentage for between-group percentage agreement was derived for comparison. Additionally, the grand means for the scaled-rating scores and the agreement percentages for all items were calculated to generate section summary values, with the two groups also compared via paired-t-test and between group-agreement percentages. Results: For all raters, 61-100％ agreement was achieved for most of the item-level ratings after binary transformation, with kappa values of 0.47-0.88. Interestingly, based on these scaled ratings, the trainees tended to rate higher than the trainers for some mood disorder items and some section-summary scores before further analysis, however, this difference disappears when comparison is made using binary transformation of the ratings. Conclusion: Based on the finding of this study, it seems reasonable to suggest that the SCAN can be reliably applied by well-trained mental-health professionals from different disciplines.(Full text in Chinese)