Assessing the effectiveness of Compare Assistant for improving intra-rater reliability of ultrasound-measured muscle thickness

Muscle thickness measured via ultrasound is commonly used to assess muscle size. The purpose of this study was to determine if the reliability of this measurement will improve if using the Compare Assistant tool, and whether this depends on technician experience and the muscle being assessed.

Individuals came to the laboratory for two visits each separated by 24 h. On day 1, two ultrasound images were taken on the individual's anterior upper arm (elbow flexors) and anterior lower leg (tibialis anterior) by two inexperienced and one experienced ultrasound technician. On day 2, three images were taken: (1) without looking at the previous images taken on day 1; (2) after re-examining the images taken on day 1, and (3) side-by-side with the images taken on day 1 via Compare Assistant. Bayes Factors (BF₁₀) were used to provide evidence for the null (< 0.33) or alternative (> 3) hypotheses.

There was no rater by measurement technique interaction (upper body: BF₁₀ = 0.04, lower body: BF₁₀ = 0.138), nor was there a main effect of measurement technique (upper body: BF₁₀ = 0.052, lower body: BF₁₀ = 0.331), indicating that reliability measures were not improved for either the upper body (CV%, no look: 2.92 vs. Compare Assistant: 2.87) or lower body (CV%, no look: 1.81 vs. Compare Assistant: 1.34) as a result of using Compare Assistant.

The results of this study suggest that day-to-day reliability of muscle thickness measurement may be limited by random biological variability as opposed to technician error.