Reliability and minimal detectable change of gait variables in community-dwelling and hospitalized older fallers

Purpose: Gait variables may constitute surrogate outcomes for fall risk. Their reliability in a specific population of older fallers has not been fully established, which limits their research and clinical applications. This study aimed to determine test-retest reliability and minimal detectable change (MDC) values for selected fall-related gait variables in older adults with a recent fall history. Methods: Community-dwelling (n = 30) and hospitalized (n = 30) fallers aged >= 65 years were assessed twice using an instrumented pressure-sensitive walkway, under single-and dual-task gait conditions. Intraclass correlation coefficient (ICC(2,1)), standard error of measurement (SEM; SEM%) and MDC at 95% confidence level (MDC95; MDC95%), were used as reliability estimates. Results: The ICC(2,1) for gait velocity was greater than 0.84 across all gait conditions and groups; SEM% and MDC95% did not exceed 6.5% and 18.1%, respectively. Gait variability measures returned lower ICC(2,1) (range 0.18-0.79), and markedly higher SEM% (16.3-31.9%) and MDC95% (45.3-88.3%). Overall, hospitalized fallers exhibited larger SEM and MDC95 values for variability measures compared to community-dwellers in all gait conditions, while larger values were found for all variables while dual-tasking compared to single-tasking in both groups. Conclusions: Gait velocity was found to be highly reliable and likely to be sensitive to change over repeated sessions in community-dwelling and hospitalized older fallers, both under single-and dual-task conditions. Gait variability measures showed lower reliability, irrespective of gait condition or group, displaying consistently larger measurement error, particularly under dual-task conditions. Clinicians should consider MDC95 values before using gait variability variables as evaluative outcome measures at patient level. (C) 2013 Elsevier B.V. All rights reserved.