Dy Christopher J, DeMartini Stephen J, Sane Eshan, Brogan David M
Department of Orthopaedic Surgery, Washington University School of Medicine, St. Louis, Missouri.
JB JS Open Access. 2025 Jun 24;10(2). doi: 10.2106/JBJS.OA.25.00012. eCollection 2025 Apr-Jun.
Following surgical reconstructions considered successful, many patients with brachial plexus injuries (BPI) have limited limb motion. In addition to manual muscle testing, clinicians typically measure active range of motion (AROM). AROM measurement relies on in-person examination and manual goniometers or visual estimation, both of which are subject to interobserver variability. The purpose of this proof-of-concept study was to evaluate reliability of video-based AROM assessments. We hypothesized that video-based assessment of AROM would have high inter-rater reliability (IRR) among surgeons who perform BPI reconstruction.
We video recorded a standardized examination in a convenience sample of 8 postreconstruction BPI patients performing 3 motions: elbow flexion (EF), shoulder flexion (FF), and shoulder abduction (ABD). Eight BPI surgeons were given access to the videos and instructed on how to measure AROM first visually and then digitally using ImageJ. We examined the correlation between video-based and in-person goniometry measurements and assessed IRR of visual estimates and digital goniometry using Shrout-Fleiss Intraclass 3 fixed set correlations.
For EF, digital goniometry had a higher correlation (r = 0.92; p < 0.01) than visual assessment (r = 0.73; p < 0.01) relative to in-person measurements. IRR for EF was 0.80 for visual assessments and 0.96 for digital assessments. For FF, digital goniometry (r = 0.80; p < 0.01) and visual assessment (r = 0.80; p < 0.01) had similar correlations relative to in-person measurements. IRR for FF was 0.95 for visual assessments and 0.99 for digital assessments. For ABD, digital goniometry had a higher correlation (r = 0.85; p < 0.01) than visual assessment (r = 0.80; p < 0.01) relative to in-person measurements. IRR for ABD was 0.91 for visual assessments and 0.96 for digital assessments.
Using standardized footage, visual estimates and digital goniometry of patient with EF, FF, and ABD were highly reliable among BPI surgeons. Digital goniometry of AROM was slightly more reliable than visual estimates for all 3 motions.
In addition to facilitating remote assessments to minimize patient travel, video-based assessments may allow opportunity to minimize reporting bias in clinical research through evaluation of results by multiple raters.
在被认为成功的手术重建后，许多臂丛神经损伤（BPI）患者的肢体活动受限。除了徒手肌力测试外，临床医生通常还会测量主动活动范围（AROM）。AROM测量依赖于现场检查和手动量角器或视觉估计，这两种方法都存在观察者间的差异。本概念验证研究的目的是评估基于视频的AROM评估的可靠性。我们假设，对于进行BPI重建的外科医生来说，基于视频的AROM评估将具有较高的评分者间信度（IRR）。
我们对8例BPI重建术后患者进行了标准化检查的视频记录，这些患者进行了3种动作：肘部屈曲（EF）、肩部屈曲（FF）和肩部外展（ABD）。8位BPI外科医生可以访问这些视频，并接受了如何先通过视觉然后使用ImageJ进行数字测量AROM的指导。我们检查了基于视频的测量与现场量角器测量之间的相关性，并使用Shrout-Fleiss组内相关系数3固定集相关性评估了视觉估计和数字量角器测量的IRR。
对于EF，相对于现场测量，数字量角器测量的相关性更高（r = 0.92；p < 0.01），而视觉评估的相关性为（r = 0.73；p < 0.01）。EF的视觉评估IRR为0.80，数字评估IRR为0.96。对于FF，相对于现场测量，数字量角器测量（r = 0.80；p < 0.01）和视觉评估（r = 0.80；p < 0.01）的相关性相似。FF的视觉评估IRR为0.95，数字评估IRR为0.99。对于ABD，相对于现场测量，数字量角器测量的相关性更高（r = 0.85；p < 0.01），而视觉评估的相关性为（r = 0.80；p < 0.01）。ABD的视觉评估IRR为0.91，数字评估IRR为0.96。
使用标准化的视频片段，BPI外科医生对EF、FF和ABD患者的视觉估计和数字量角器测量具有高度可靠性。对于所有3种动作，AROM的数字量角器测量比视觉估计稍可靠。
除了便于进行远程评估以尽量减少患者的出行外，基于视频的评估还可能提供机会，通过多名评估者对结果的评估，尽量减少临床研究中的报告偏倚。
