國內中風病人職能治療 (occupational therapy, OT) 病歷紀錄之內容多元。治 療師執行評估再記錄個案各面向（如動作能力、平衡與日常生活功能等）之評估 結果。病歷資料擁有三項優點：一、資料量龐大；二、病歷內容通常多元（面向 廣）且常有追蹤評估紀錄；三、可行性高。且近年來，國內多個OT 單位已推動 病歷電子化，更有助於提升病歷分析之效能。 病歷分析之研究類型，包含描述性分析、相關及探索性分析等。中風病人OT 病歷之研究題材包含：疾病預後與危險因子探索、各種功能特質之恢復模型建立 與驗證、理論之驗證、評估工具驗證與發展、治療成效探索、發展臨床決策輔助 系統、治療師專長之確認等。 然而目前國內中風病人OT 病歷之紀錄內容具有四項缺失：一、標準化不足； 二、欠缺心理計量特性驗證；三、面向涵蓋不足；四、多數評估面向之項目簡略， 精準度有限。改良上述缺失之方法包含：一、標準化施測流程與情境，並建立完 整之施測手冊；二、驗證所有病歷面向（內容）之心理計量特性；三、依據心理 計量驗證結果，改良病歷面向，改良之後再進行心理計量特性驗證，直到病歷所 有面向之心理計量特性良好為止。另外，改良病歷時亦須兼顧快速可行，以利臨 床人員執行。

The content of records of occupational therapy (OT) is usually plentiful and diverse in Taiwan. Occupational therapists assess and record patients’ conditions in terms of several dimensions (e.g., motor status, balance, and activities of daily living). Research using medical records has at least three advantages: the data set is large, the records contain several dimensions, and the feasibility is high. In addition, several OT institutes in Taiwan use electronic records, which greatly facilitate data analysis. The types of research on medical records include descriptive studies, correlational studies, and exploratory studies. The research topics of OT records include explorations of disease prognosis and risk factors, establishment and validation of recovery models for specific functions, validation of theories, validation and development of measures, exploration of treatment effects of specific interventions, development of clinical decision-making support systems, and verification of therapists’ expertise. However, four flaws exist in most of our OT records for patients with stroke: limited standardization, limited psychometric validation, narrow dimension coverage, and limited numbers of items affecting the precision of assessment results. These flaws can be overcome by the following methods: standardization of every assessment and establishment of testing manuals, validation of every dimension of assessment in terms of reliability and validity, and revision of the assessment content on the basis of psychometric validation. Any revisions of measures have to be further validated until excellent psychometric properties have been supported.