Validity of remote systems
Cervical rotation measurement by the CV-aided system was moderately
correlated to a clinician assessment in the axSpA group and weakly
correlated in the non-back pain groups (Table 4); in the axSpA group,
the CV-aided system demonstrated a –2.6cm bias compared to the
reference physiotherapist measurement with a positive regression slope
(Figure 2). Lumbar forward flexion and hip internal rotation were
strongly correlated in both the axSpA and non-back pain groups; both
demonstrated a postive bias (+0.4cm and +3.7cm, respectively) with one
outlier beyond the limits of agreement. Shoulder flexion and lumbar side
flexion showed a strong to very strong correlation in axSpA group and
moderate to weak correlation in non-back pain group. Shoulder flexion
demonstrated a negative bias (right –3.0 degrees, left –1.4 degrees)
with a slightly negative slope, and lumbar side flexion demonstrated
minimal bias (right –0.6cm, left 0cm). Hip abduction was moderately
correlated in axSpA group and demonstrated moderate to strong
correlation in the non-back pain group. Metrics for posture showed
strong correlation for tragus-to-wall distance (TWD) and thoracic
kyphosis measurement in the axSpA group, yet very weak (TWD) to moderate
(kyphosis) correlation in the non-back pain group; lumbar lordosis was
not significantly correlated in either group (Table 4). All measurements
showed agreement in the axSpA group with minimal bias (TWD –0.9,
kyphosis +0.4, lordosis +0.2); TWD has a positive slope and kyphosis and
lordosis have negative slopes, all with few outliers (Figure 3).