Anatomical coordinate systems are fundamental tools in biomechanics that allow joint kinematics to be described in a clinically-relevant manner. However, kinematic descriptions depend on how coordinate systems were defined precluding direct comparison across studies that use different conventions. This challenge is especially prevalent when investigating shoulder kinematics where no less than nine different coordinate system conventions have been used for the scapula (Amadi et al., 2008, Calderone et al., 2014, Hebert et al., 2000, Kedgley and Dunning, 2010, Kolz et al., 2020, Ohl et al., 2015, Pearl et al., 1992, van der Helm, 1997, Wu et al., 2005).

Historically, the most commonly used scapular coordinate systems are those proposed by van der Helm (van der Helm, 1997) and the International Society of Biomechanics (ISB) (Wu et al., 2005). Originally, van der Helm proposed that the scapular axes should be constructed using the root of the scapular spine, posterior acromioclavicular joint, and inferior angle (van der Helm, 1997). Less than a decade later, however, the ISB replaced the posterior acromioclavicular landmark with the posterolateral acromion to help prevent gimbal lock (Wu et al., 2005). However, the resulting axes no longer represent the scapula’s anatomical plane, which is important for clinical interpretation (Ludewig et al., 2010) and possibly the rationale for its continued use by many researchers (e.g., Lawrence et al., 2014, Ludewig et al., 2009, McClure et al., 2006).

More recently, a glenoid-based coordinate system has been used to describe glenohumeral translations and arthrokinematics (Peltz et al., 2015) since the glenoid provides a more meaningful reference than the full scapula for these measures. Widespread use of a glenoid-based coordinate system is hindered, however, due to the glenoid’s inaccessibility to palpation, which is necessary for surface-based motion capture techniques using sensors or markers. Finally, a glenoid-oriented coordinate system has also been proposed in which the medial–lateral axis is oriented to the glenoid center instead of the acromion, offering a potential compromise between the original van der Helm and ISB conventions.

Although several researchers have compared kinematics described using different scapular coordinate systems (Calderone et al., 2014, Kolz et al., 2020, Ludewig et al., 2010), it remains unclear which convention corresponds mostly closely with the true 3D axis of scapular motion, which can be described using an instantaneous helical axis (IHA). This gap in our knowledge is especially problematic given the frequent use of Euler angles, which describe a joint’s kinematics as an ordered sequence of rotations typically about the distal segment’s axes, and inconsistencies between these axes and the 3D motion axis will result in mathematical artifact confounding clinical descriptions. Helical angles have been suggested as an alternative; however, they lack physical interpretation (Woltring, 1991). Consequently, Euler angles remain the primary method for describing kinematics despite their many limitations and uncertain correspondence with the true 3D motion axis. Therefore, the objectives of this study were to: 1) determine the extent to which the axes of four common scapular coordinate system conventions correspond to the true 3D axis of scapular motion (IHA), and 2) report the prevalence of scapulothoracic gimbal lock for each convention.