Intra-protocol repeatability and inter-protocol agreement for the analysis of scapulo-humeral coordination

Abstract

Multi-center clinical trials incorporating shoulder kinematics are currently uncommon. The absence of repeatability and limits of agreement (LoA) studies between different centers employing different motion analysis protocols has led to a lack dataset compatibility. Therefore, the aim of this work was to determine the repeatability and LoA between two shoulder kinematic protocols. The first one uses a scapula tracker (ST), the International Society of Biomechanics anatomical frames and an optoelectronic measurement system, and the second uses a spine tracker, the INAIL Shoulder and Elbow Outpatient protocol (ISEO) and an inertial and magnetic measurement system. First within-protocol repeatability for each approach was assessed on a group of 23 healthy subjects and compared with the literature. Then, the between-protocol agreement was evaluated. The within-protocol repeatability was similar for the ST ($\overline{\text{RMSE}}$ = 2.35°, $\sigma_{\text{RMSE}}$ = 0.97°, SEM = 2.5°) and ISEO ($\overline{\text{RMSE}}$ = 2.24°, $\sigma_{\text{RMSE}}$ = 0.97°, SEM = 2.3°) protocols and comparable with data from published literature. The between-protocol agreement analysis showed comparable scapula medio-lateral rotation measurements for up to 120° of flexion-extension and up to 100° of scapula plane ab-adduction. Scapula protraction–retraction measurements were in agreement for a smaller range of humeral elevation. The results of this study suggest comparable repeatability for the ST and ISEO protocols and between-protocol agreement for two scapula rotations. Different thresholds for repeatability and LoA may be adapted to suit different clinical hypotheses.

Appendix

In order to compare our findings of repeatability of each method to those in the literature it was necessary to compute the mean and standard deviation of RMSE (of Eq. 1) as well as the SEM.

• i) The mean and standard deviation (SD) of the dataset are computed by polling together the scapula ${\text{RMSE}}_{\text{within - protocol}}^{h}$ (Eq. 1) of all humerus elevation angles [21]:

  $$\overline{\text{RMSE}} = \frac{{\sum\nolimits_{h = 1}^{11} {{\text{RMSE}}_{\text{within - protocol}}^{h} } }}{11}$$
  $$\sigma_{\text{RMSE}} = {\text{SD}}({\text{RMSE}}_{\text{within - protocol}}^{1} , \ldots ,{\text{RMSE}}_{\text{within - protocol}}^{11} )$$

• ii) the SEM for the scapula kinematics is [22]:

  $${\text{SEM}}\; = \;\sqrt {{\text{MS}}_{W} }$$

  where MS_W is the sum of squares due to the within-subject variability, computed from one-way repeated measures ANOVA [14, 26]

CR is related to SEM [1] by the formula:

$${\text{SEM }} = {\text{ CR}}/( 1. 9 6*\surd 2)$$

This relationship was used to convert values of SEM that was published by van Andel [22] to CR_within and compare the results of the within-protocol repeatability of this study with the literature.