Assessing gait efficacy in older adults: An analysis using item response theory

Abstract

BackgroundAssessing confidence in walking in older adults is important, as mobility is a critical aspect of independence and function, and self-report provides complementary information to performance-based measures. The modified Gait Efficacy Scale (mGES) is a self-report measure used to examine confidence in walking. Research Question: What are the psychometric properties of the mGES at the item level? Are there opportunities for improvement? MethodsWe performed a secondary analysis of baseline data from a cluster randomized trial of 424 community-dwelling older adults and reliability data from 123 participants. We fitted a graded response model to dissect the mGES to the item and individual response level and examined opportunities to improve and possible shorten the mGES. We examined psychometric characteristics such as internal consistency, test-retest reliability and construct validity with respect to other relevant measures. ResultsMobility tasks such as navigating stairs and curbs with separate items for going up and down largely provide the same information on confidence, with downward direction providing slightly more. It may be reasonable to consider removal of walking 1/2 mile, stepping down and/or stair tasks with railings items due to little or duplicate information contributed compared to other items. The shortened scales proposed by removing the above items had similar psychometric properties to mGES. SignificanceThe mGES has good psychometric properties, but can be potentially shortened to substantially reduce responder burden. The upward direction curb and stairs items can be removed to result in a 7-item scale with virtually no loss of desirable psychometrics. An alternative 3-item version, level surface walking, stepping down curb and climbing up stairs without a railing items, entails only a minimal loss in psychometric properties.