The required minimum length of video sequences for obtaining a reliable interobserver diagnosis in wrist arthroscopies

Purpose To examine the relationship between video length for wrist arthroscopy and interobserver reliability. 100 consecutive wrist arthroscopies were documented by long and short videos of the radiocarpal and the midcarpal joints. The long videos were about twice as long as the short videos. They were presented randomly to two independent and blinded examiners. Their diagnoses were compared to the diagnoses made by the surgeon who performed the arthroscopies. Kappa coefficients were calculated. Kappa statistics were inconsistent and did not show that the long video provided an obvious advantage over the short video. The Kappa coefficients of the two examiners for the assessment of the cartilage status were 0.524 and 0.700 for the long videos and 0.465 and 0.639 for the short videos, respectively. The examiners diagnosed twice as many false-positive cartilage lesions on short videos than on long videos. The assessment of ligament lesions was more accurate on long than on short videos. The results confirmed the hypothesis that the reproducibility of diagnoses based on video documents was influenced by the length of the video sequences. Therefore, it may be advisable for video documentation to be done diligently. The video sequence of the radiocarpal joint should last about 60 s, and that of a midcarpal joint should last about 45 s. Videos of difficult joints should last appropriately longer. Diagnostic II.