Dementia Knowledge - Psychometric Evaluation in Healthcare Staff and Students

Abstract

Dementia is a progressive life-limiting condition. Global prevalence is increasing in line with current rates of population ageing. A number of disciplines have found it useful to measure dementia knowledge levels. Challenges exist in achieving effective measurement across a range of healthcare professionals, but this is important to inform health-related education and professional development interventions. Similarly, effective measurement is required in informal care givers as well as lay populations, to improve evaluation of national public health and awareness campaigns.

The field of dementia knowledge measurement has seen rapid growth over the past decade, with a proliferation in the number of available instruments. To identify and appraise existing measurement instruments, a systematic review was undertaken. 14 instruments were identified; critical appraisal showed that the overall psychometric quality was poor, with no instrument emerging as ‘gold standard’. A key finding from this review was that measurement of dementia knowledge is currently anchored by classical test theory (CTT) methods, with a notable absence of any item response theory (IRT) methods used in instrument development or evaluation. To address this gap, and to highlight the potential usefulness of IRT methods in evaluating the measurement properties of instruments, this study sought to apply IRT methods in dementia knowledge instruments, to determine what information could be generated
about the measurement properties of items and item-sets using IRT methods, as opposed to CTT methods alone.

Two datasets were used in this thesis: the first comprised 521 sets of responses from healthcare professionals to the 16-item Knowledge in Dementia Scale (KIDE). The second comprised 404 sets of responses to the KIDE and the 25-item Dementia Knowledge Assessment Scale (DKAS).

This PhD thesis advances the field and contributes to knowledge by synthesising the current measurement literature and demonstrating the value and potential of strong measurement modelling under the Rasch paradigm, highlighting areas of further development but also areas of weakness within current items and instrumentation.

Date of Award	2022
Original language	English
Awarding Institution	University of Dundee
Supervisor	Tim Croudace (Supervisor), Judith Sixsmith (Supervisor) & Wendy Moncur (Supervisor)