Abstract
Introduction: Reliable measures of muscle strength and function in older adults are essential for meaningful clinical assessments and to enable accurate conclusions to be made about interventions. However, multiple familiarisation sessions prior to assessments to ensure reliability are not feasible in many settings. Coefficients of variation (CVs) could be an efficient means of determining if an individual can produce repeatable subsequent performances in tests of muscle strength and function. The aims of this study were to determine: (i) whether CVs for each participant at the first test session can predict repeatability of performance in a subsequent session; and (ii) the level of CV that produces acceptable repeatability (<6% variation between the first and second test).
Methods: Forty-eight healthy older adults (average age 68.6 ± 6.1 years) completed two assessment sessions (S1 and S2), and on each occasion undertook: Isokinetic dynamometry for isometric and isokinetic quadriceps strength (60, 120, 240 and 360°/s), 6 metre fast walk (6mFWT), timed up and go (TUG), stair climb and descent, and vertical jump.
Results: Significant linear relationships were observed between CVs in S1 and the percentage difference between S1 and S2 (%diff) for torque at 60°/s (r = 0.484; P < 0.01), 120°/s (r = 0.730; P < 0.01), 240°/s (r = 0.702; P < 0.01) and 360°/s (r = 0.329; P < 0.05), 6MFWT (r = 0.336; P < 0.05), TUG (r = 0.372; P < 0.01), stair climb (r = 0.529; P < 0.01) and stair descent (r = 0.340; P = 0.01). Participants who in S1 achieved CVs of less than 2.84% in the 60°/s, 120°/s, 240°/s and 360°/s dynamometer tests were likely (P < 0.05) to have a %diff of less than 6%. Participants who achieved CVs of less than 4.50% in the 6MFWT and stair climb test in S1 were likely (P < 0.05) to have a %diff of less than 5%. Participants who achieved CVs of less than 1.37% in the TUG were likely (P < 0.05) to have a %diff of less than 6%. Males who achieved CVs of less than 0.18% in the stair descent test in S1 were likely (P < 0.05) to have a %diff of less than 6% while females who achieved CVs of less than 2.79% in the stair descent test in S1 were likely (P < 0.05) to have a %diff of less than 5%.
Discussion: Coefficients of variation can predict repeatability of performance in individuals in some tests, and may therefore be used to determine whether further familiarisation is required prior to the attainment of valid and reliable assessment data.