Reliability of upper limb movement quality metrics during everyday tasks

Quantitative assessments of an individual's functional status commonly involve the use of movement quality metrics.

The purpose of this work was to quantify the reliability of movement quality metrics in healthy adults during a variety of unconstrained activities of daily living (ADLs).

Nineteen participants performed six ADLs (lifting a laundry basket, applying deodorant, turning a doorknob, placing a pill in a pillbox, placing a pushpin in a bulletin board, and drinking water from a glass) during two separate sessions. The ADLs were divided into reaching and object manipulation phases. Movement quality for each phase was assessed using three measures of smoothness (log dimensionless jerk, spectral arc length, and number of submovements) and one measure of straightness (index of curvature). Within- and between-session reliability was quantified using intraclass correlation coefficients (ICCs) and minimum detectable changes in measured units and as a percentage of their mean value (MDC%).

Reliability was generally lower within-session than between-session and for object manipulation tasks compared to reaching tasks. The ICCs exceeded 0.75 for 5% of the within-session metrics and 73% of the between-session metrics. The average MDC% was 35% for the within-session metrics and 20% for the between-session metrics. Reliability was similar for most metrics when averaged across the tasks, but the number of submovements consistently indicated much lower reliability.

Unconstrained ADLs can reliably be used to assess movement quality in functional settings that mimic real-world challenges. However, the specific movement quality metrics used in the assessment should be chosen carefully since some metrics perform dissimilarly when applied to the same data. In particular, it may be advisable to use the number of submovements in combination with other metrics, if it is to be used at all.