應用影像軟體於腦性麻痺兒童攝影步態分析之心理計量特性

背景與目的：異常步態為腦性麻痺兒童常見問題，容易造成疼痛、肌肉疲勞、增加跌倒風險等問題，因此改善異常步態為臨床的治療重點。臨床常用步態分析量表評估腦性麻痺兒童步態，攝影步態分析為拍攝步態影片後再以步態分析量表評估，為臨床常用於步態分析之工具。由於攝影步態分析為評估者以肉眼直接觀看影片評分，容易出現主觀上的偏差。影像軟體可直接在影像上標記關節角度和距離，過去研究曾提出使用影像軟體協助攝影步態分析可以增加評估者間信度，但未曾討論效度與反應性，因此本研究欲探討應用影像軟體協助腦性麻痺兒童攝影步態分析之心理計量特性，包含信度、效度與反應性。方法：本研究使用愛丁堡視覺步態量表 (Edinburgh Visual Gait Scale)，將11個項目以影像軟體協助量測角度後評分 (Kinovea assist visual scale-11, KAVS-11) 對照直接觀看影像評分 (EVGS-11) 做為信度分析。將使用影像軟體協助量測之總分 (KAVS-17) 及直接觀看影像之總分 (EVGS-17)，相較於粗大動作功能量表-88向度E (Gross Motor Function Measure-88 item E, GMFM-88, E) 做為收斂效度分析，相較於粗大動作功能分級系統 (Gross Motor Function Classification System, GMFCS) 做為區辨效度分析。追蹤6個月後EVGS-17及KAVS-17步態分數變化，與GMFM-88, E比較做為反應性分析。結果：本研究共納入38位腦性麻痺兒童 (男生15人/女生23人)。評估者內信度：KAVS-11 kappa (Ka) 平均由0.51增加為0.60，PA (percentage of agreement) 由77%增加為80%。評估者間信度：KAVS-11 Ka平均由0.41增加為0.46，PA由72.6%改變為72.9%。證實使用影像軟體可增加信度之準確性。EVGS-17與KAVS-17總分與GMFM-88, E均呈現高度相關 (EVGS-17:rs=-0.77; KAVS-17:rs=-0.73)，證實使用EVGS-17與KAVS-17評估腦性麻痺兒童異常步態具有良好的收斂效度。GMFCS三組兒童在EVGS-17與KAVS-17分數均呈現統計上顯著差異 (P<0.001)。證實EVGS-17和KAVS-17具有良好之區辨效度，能區辨不同GMFCS等級之腦性麻痺兒童。追蹤6個月後步態分數變化，KAVS-17具有較佳效果量 (0.19)，且與GMFM-88, E (0.18) 具有一致之反應性，證實使用KAVS-17評估腦性麻痺兒童異常步態可反應出行走動作之變化。結論：本篇研究證實使用影像軟體協助攝影步態分析具有良好之心理計量特性，但相較於使用肉眼直接評估步態，使用影像軟體於評估者的一致性並無明顯差異。

Background and Purpose:Abnormal gait pattern is one of the common problems in children with cerebral palsy, which cause pain, muscle fatique, and increase the fall risk. Therefore, improving the abnormal gait is important in clinical treatment. The gait analysis scales are commonly used to assess the gait in children with cerebral palsy. Video gait analysis is an important clinical evaluation tool. However, using visual to assess gait pattern from video is prone to observer bias. Pervious studies had pointed out the video software which enabled the users to perform angle measurements on video offered the potential to decrease the bias on video gait analysis. Therefore, the purpose of this study was to examine the psychometric properties with using video software in video gait analysis for children with cerebral palsy, including reliability, validity and responsiveness.Methods:Compare the 11 numeric items of the Edinburgh visual gait scale with visual (EVGS-11) and with video software (KAVS-11) as intra-rater reliability and inter-rater reliability. Convergent validity: The correlation between total scores of EVGS (EVGS-17) and KAVS (KAVS-17) and dimension E of GMFM-88. Discriminant validity: The difference between EVGS-17, KAVS-17 and at each level of GMFCS. Responsiveness: the change between EVGS-17, KAVS-17 and the dimension E of GMFM-88 under half-year following.Results:There are 38 children of cerebral palsy (boy:15/girl:23). (1) Intra-rater reliability: KAVS-11 kappa (Ka) increased from 0.51 to 0.60 on average, percentage of agreement (PA) increased from 77% to 80% on average. Inter-rater reliability: KAVS-11 Ka increased from 0.41 to 0.46 on average, PA increased from 72.6% to 72.9% on average. It is proved that using video gait analysis with video software can increase the accuracy of reliability. (2) Convergent validity: EVGS-17 rs = -0.77; KAVS-11 rs: = -0.73. It is good convergent validity that using EVGS-17 and KAVS-17 to assess abnormal gait in children with cerebral palsy. Discriminat validity: It is significant differences in the scores of EVGS-17 and KAVS-17 with the three groups of children in GMFCS (P<0.001). It is confirmed that EVGS-17 and KAVS-17 have good discriminat validity and can discriminate children with cerebral palsy in different GMFCS grades. (3) responsiveness: effect size: GMFM-88, E=0.18; EVGS-17=0.05; KAVS-17=0.19. The responsiveness of KAVS-17 is cross to GMFM-88, E, and better than EVGS-17. It is proved that KAVS-17 can reflect changes in the gross motor function in children with cerebral palsy.Conclusion: It is good psychometric properties in using video gait analysis with video software for children with cerebral palsy. However, compared with using visual gait analysis, there is no significant difference in the reliability of using video gait analysis with video software.