目的： 本研究的目的在於檢驗失眠嚴重度量表―中文版（Insomnia Severity Index- Chinese Version; ISI-C）的信、效度。方法： 本研究共分兩部分，先針對147位失眠患者及182位正常睡眠者進行ISI-C及中文版匹茲堡睡眠品質量表（Pittsburgh Sleep Quality
Index-Chinese Version; CPSQI）等量表的施測，再針對71位參與失眠認知行為治療的失眠患者，進行ISI-C以及睡眠日誌的前、後測。結果： ISI-C顯示相當好的內部一致性（Chronbach's α＝0.94），且修正後的項目-全量表相關係數皆達顯著。在效度方面，與CPSQI的總分相關達.88，與睡眠日誌的相關參數則有低度到中度的顯著相關，得分在認知行為治療後也有顯著的下降。ROC分析顯示以9分為切分點有良好的敏感度（91.8%）與特異度（91.2%）。結論： 研究結果顯示ISI-C有良好的信、效度，為非常方便的失眠篩檢工具，也可以做為失眠治療的療效測量工具。

Objective: The study was conducted to examine the psychometric properties of the Insomnia Severity Index-Chinese version (ISI-C).Method: Study 1 recruited 147 insomnia patients from a general hospital and 182 normal sleepers from local communities. A package of questionnaires including the ISI-C and the Chinese version of the Pittsburgh Sleep Quality Index (CPSQI) were administered to all subjects. Study 2 included 71 patients with insomnia complaints who went through a 6-session program of cognitive behavioral therapy for insomnia (CBT-I). The ISI-C and sleep diaries were given to the participants before and after the CBT-I program.
Results: The results showed very good internal consistency of the ISI-C, with a Chronbach’s α coefficient of 0.94.Corrected item-total items correlations were all significant. In terms of validity, ISI-C showed very good concurrent validity with the CPSQI. The total scores of these two scales correlated significantly (r = .88).The correlations between most of the items of the two scales also reached moderate to high correlations.The correlations between ISI-C items and corresponding indices derived from sleep diaries were also significant, although most of them were in the low to moderate correlation range. Furthermore, patients’scores on the ISI-C decreased significantly after receiving CBT-I program, reflecting the beneficial effects of treatment. Regarding construct validity, factor analysis revealed a single factor that can explain 73.94%of the variance. ROC analysis showed a best cut-off point of 9, with very good sensitivity (91.8%) and specificity (91.2%).