Damsted C, Larsen L H, Nielsen R O
Institute of Sports Science and Clinical Biomechanics, University of Southern Denmark, Campusvej 55, 5230 Odense M, Denmark.
University Collage of Northern Jutland, Selma Lagerloefs Vej 2, 9220 Aalborg East, Denmark.
Gait Posture. 2015 Jun;42(1):32-5. doi: 10.1016/j.gaitpost.2015.01.029. Epub 2015 Apr 14.
Two-dimensional video recordings are used in clinical practice to identify footstrike pattern. However, knowledge about the reliability of this method of identification is limited.
To evaluate intra- and inter-rater reliability of visual identification of footstrike pattern and video time frame at initial contact during treadmill running using two-dimensional (2D) video recordings.
Thirty-one recreational runners were recorded twice, 1 week apart, with a high-speed video camera. Two blinded raters evaluated each video twice with an interval of at least 14 days.
Kappa values for within-day identification of footstrike pattern revealed intra-rater agreement of 0.83-0.88 and inter-rater agreement of 0.50-0.63. Corresponding figures for between-day identification of footstrike pattern were 0.63-0.69 and 0.41-0.53, respectively. Identification of video time frame at initial contact ranged from five frames to 12 frames (95% limits of agreement).
For clinical use, the intra-rater within-day identification of footstrike pattern is highly reliable (kappa>0.80). For the inter-rater between-day identification inconsistencies may, in worst case, occur in 36% of the identifications (kappa=0.41). The 95% limits of agreement for identification of video time frame at initial contact may, at times, allow for different identification of footstrike pattern. Clinicians should, therefore, be encouraged to continue using clinical 2D video setups for intra-rater identification of footstrike pattern, but bear in mind the restrictions related to the between day identifications.
临床实践中使用二维视频记录来识别着地方式。然而，关于这种识别方法可靠性的知识有限。
使用二维（2D）视频记录评估在跑步机跑步初始接触时着地方式和视频时间帧视觉识别的评分者内和评分者间可靠性。
31名休闲跑步者用高速摄像机进行两次记录，间隔1周。两名不知情的评分者对每个视频评估两次，间隔至少14天。
着地方式日内识别的Kappa值显示评分者内一致性为0.83 - 0.88，评分者间一致性为0.50 - 0.63。着地方式日间识别的相应数字分别为0.63 - 0.69和0.41 - 0.53。初始接触时视频时间帧的识别范围为5帧至12帧（一致性界限的95%）。
对于临床应用，着地方式的评分者日内识别高度可靠（kappa>0.80）。对于评分者间日间识别，在最坏情况下，36%的识别可能会出现不一致（kappa = 0.41）。初始接触时视频时间帧识别的95%一致性界限有时可能导致着地方式的不同识别。因此，应鼓励临床医生继续使用临床2D视频设置进行评分者内着地方式识别，但要记住与日间识别相关的限制。
