Reliability, measurement error and minimum detectable change in mobility measures: a cohort study of community-dwelling adults aged 50 years and over in Ireland.

Abstract

To estimate the effects of repeat assessments, rater and time of day on mobility measures and to estimate their variation between and within participants in a population-based sample of Irish adults aged ≥50 years. Test-retest study in a population representative sample. Academic health assessment centre of The Irish Longitudinal Study on Ageing (TILDA). 128 community-dwelling adults from the Survey for Health, Ageing and Retirement in Europe (SHARE) Ireland study who agreed to take part in the SHARE-Ireland/TILDA collaboration. Not applicable. Participants performed timed up-and-go (TUG), repeated chair stands (RCS) and walking speed tests administered by one of two raters. Repeat assessments were conducted 1-4 months later. Participants were randomised with respect to a change in time (morning, afternoon) and whether the rater was changed between assessments. Within and between-participant variance for each measure was estimated using mixed-effects models. Intraclass correlation (ICC), SE of measurement and minimum detectable change (MDC) were reported. Average performance did not vary between baseline and repeat assessments in any test, except RCS. The rater significantly affected performance on all tests except one, but time of day did not. Reliability varied from ICC=0.66 (RCS) to ICC=0.88 (usual gait speed). MDC was 2.08 s for TUG, 4.52 s for RCS and ranged from 19.49 to 34.73 cm/s for walking speed tests. There was no evidence for lower reliability of gait parameters with increasing time between assessments. Reliability varied for each test when measurements are obtained over 1-4 months with most variation due to rater effects. Usual and motor dual task gait speed demonstrated highest reliability.