Reliability and validity of 2-dimensional video analysis for a running task: A systematic review

Objective: Evaluate the reliability and validity of 2-dimensional (2D) video-based motion analysis during running. Methods: A systematic search of MEDLINE, Cochrane Library, EMBASE, CINAHL, PEDro, SPORTDiscus, and IEEE Xplore was conducted in March 2020 and updated in May 2021. We included studies assessing reliability and/or validity of 2D video-based motion analysis (gold standard: 3D motion analysis) during running. Results: 11 studies (251 runners; mean age range: 18.7–37.0 years; 57.4% female; 63.7% injury-free) met inclusion criteria. Eight studies examined kinematics of the pelvis/hip, eight of the knee, and six of the ankle/foot. Low-to-moderate risk of bias was present in all studies. Heterogeneous study designs, measurement methods, and statistical approaches across studies precluded statistical synthesis. Intrarater reliability [Interclass correlation coefficient (ICC) range: 0.56–1.00; kappa range: 0.49–0.81] was better than interrater reliability (ICC range: 0.31–1.00; kappa range 0.00–0.85). ICC values for validity were poor to good (0.06–0.89). One study examining foot strike pattern found good to excellent validity (using Gwet AC statistics) when movement kinematics were categorized. Conclusions: A wide range of methods were reported in 2D video-based motion analysis of joint angular kinematics during a running task. Further research to develop standardized 2D video-based motion analysis for running is needed. Categorizing movement patterns may be more useful than angularly quantifying joint kinematics.