Reliability of medial versus lateral video analysis for footstrike assessment in treadmill and overground running

Abstract

Although video analysis is widely used to assess footstrike pattern (FSP) and footstrike angle (FSA), the reliability of these measures from medial and lateral views remains poorly understood, particularly across treadmill and overground running. This study evaluated the intra-rater, inter-rater and inter-side reliability of FSP classification and FSA measurement between treadmill and overground running. This study involved seven novice runners on an indoor treadmill and nine experienced runners during an outdoor marathon-distance run. Synchronized medial- and lateral-view videos were recorded and analyzed twice by three trained raters using Kinovea software. FSP was categorized as rearfoot, midfoot or forefoot strike, and FSA was measured as the angle between the shoe sole and running surface at initial contact. A total of 1,680 treadmill and 134 overground footfalls were analyzed. Reliability was assessed using Cohen's or Fleiss’ kappa (k) for FSP classification and intraclass correlation coefficient (ICC) for FSA measurement. Excellent intra-rater (Cohen's k ≥ 0.89, ICC<inf>(1,1)</inf> ≥ 0.96) and inter-rater reliability (Fleiss’ k ≥ 0.85, ICC<inf>(2,1)</inf> ≥ 0.95) were indicated during treadmill running. For overground running, reliability remained good but comparatively lower (Cohen's k ≥ 0.74, ICC<inf>(1,1)</inf> ≥ 0.86; Fleiss’ k ≥ 0.64, ICC<inf>(2,1)</inf> ≥ 0.84). Inter-side reliability between medial and lateral views was excellent on the treadmill (k ≥ 0.88, ICC ≥ 0.94) but only fair to moderate overground (k ≥ 0.48, ICC ≥ 0.68). While video-based footstrike assessment is highly reliable in controlled laboratory settings, its application in field-based conditions requires caution due to diminished inter-side agreement.