Measuring physical function in people with dementia: a psychometric investigation

Abstract

Dementia is a global issue, and it is a growing problem. Exercise is an effective intervention strategy for improving cognition and physical function for people of all ages, and measurement is fundamental to scientific endeavour. This thesis sits at the crossroads to these three key areas. Without appropriate psychometric justification behind the choice of measurement tools for exercise intervention studies, the validity of outcomes to suggest that exercise is beneficial for people with dementia should be called into question. This thesis aims to identify common measures of physical function, a key construct to remaining independent for people with dementia, assess their psychometric properties and determine whether such measures are suitable for use for people with dementia. This thesis uses a multi-stage, mixed methods approach, with each stage founded upon its own question to answer. Stage One answers the question; what measures of physical function are currently used in exercise intervention studies? Stage Two answers the question; have common measures been rigorously assessed for their psychometric properties in a dementia specific population? Stage Three answers the question; are common measures, previously unassessed, valid and reliable in a sample of people with dementia? Stage Four answers the question; what does it mean for a person with dementia to be lphysically abler? In Stage One, the systematic review of the benefits of exercise for people with dementia demonstrated significant heterogeneity, making a meta-analysis and final recommendation on the frequency, intensity and time of exercise for this population difficult. A significant factor in this was the large number of outcome measures chosen to demonstrate the benefits of exercise. Sub-group analysis with identical measures was able to demonstrate homogeneity. Part of the findings from this review was that physical function as a construct is poorly defined and measured. In Stage Two, a second systematic review was undertaken. This examined the psychometric properties of all measures from Stage One in dementia specific populations. While some measures had garnered the attention of researchers, many measures had no, or very little psychometric investigation. This potentially undermines the validity of the intervention studies reviewed in Stage One. In Stage Three, a subset of commonly used measures of physical function, which had not previously been assessed, were subject to a seven day test-retest reliability trial and video analysis of the standardised protocol. The majority of measures had high levels of relative reliability (ICC g 0.71), but low levels of absolute reliability, as demonstrated by large Standard Error of Measures and Limits of Agreement. The prevalence of error in task execution and non-standardised instruction was significantly different in the first trial (pl0.05), generally these differences were no longer present by the third trial (pg0.05). It was evident that the majority of our sample were able to complete the measures to standardised the protocol. In Stage Four, the previous definitions of physical function, which centre on capacity to complete complex and basic daily tasks, is appropriate for people with dementia. However, further attention should be given to the tasks most basic and central to remaining independent. The discussion focussed on three main findings from this thesis. These are; there is reasonable evidence to support the use of assessed measures, intended use should guide clinicians and researchers, and simple measures for people that have low physical and low cognitive function should be used. Implications will translate these main findings into expected consequences of this work. A Rasch Analysis of the data from Stage Three was briefly presented to demonstrate the applicability of this approach to future psychometric investigation into this area.