Sarah McGrory1, Susan D Shenkin2, Elizabeth J Austin3, John M Starr4. 1. Alzheimer Scotland Dementia Research Centre, University of Edinburgh, 7 George Square, Edinburgh EH8 9JZ, UK. 2. Geriatric Medicine, University of Edinburgh, Edinburgh, UK Centre for Cognitive Ageing and Cognitive Epidemiology, University of Edinburgh, Edinburgh, UK. 3. Psychology, University of Edinburgh, Edinburgh, UK. 4. Alzheimer Scotland Dementia Research Centre, University of Edinburgh, 7 George Square, Edinburgh EH8 9JZ, UK Geriatric Medicine, University of Edinburgh, Edinburgh, UK Centre for Cognitive Ageing and Cognitive Epidemiology, University of Edinburgh, Edinburgh, UK.
Abstract
BACKGROUND: impairment of functional abilities represents a crucial component of dementia diagnosis. Current functional measures rely on the traditional aggregate method of summing raw scores. While this summary score provides a quick representation of a person's ability, it disregards useful information on the item level. OBJECTIVE: to use item response theory (IRT) methods to increase the interpretive power of the Lawton Instrumental Activities of Daily Living (IADL) scale by establishing a hierarchy of item 'difficulty' and 'discrimination'. METHODS: this cross-sectional study applied IRT methods to the analysis of IADL outcomes. Participants were 202 members of the Scottish Dementia Research Interest Register (mean age = 76.39, range = 56-93, SD = 7.89 years) with complete itemised data available. RESULTS: a Mokken scale with good reliability (Molenaar Sijtsama statistic 0.79) was obtained, satisfying the IRT assumption that the items comprise a single unidimensional scale. The eight items in the scale could be placed on a hierarchy of 'difficulty' (H coefficient = 0.55), with 'Shopping' being the most 'difficult' item and 'Telephone use' being the least 'difficult' item. 'Shopping' was the most discriminatory item differentiating well between patients of different levels of ability. CONCLUSIONS: IRT methods are capable of providing more information about functional impairment than a summed score. 'Shopping' and 'Telephone use' were identified as items that reveal key information about a patient's level of ability, and could be useful screening questions for clinicians.
BACKGROUND: impairment of functional abilities represents a crucial component of dementia diagnosis. Current functional measures rely on the traditional aggregate method of summing raw scores. While this summary score provides a quick representation of a person's ability, it disregards useful information on the item level. OBJECTIVE: to use item response theory (IRT) methods to increase the interpretive power of the Lawton Instrumental Activities of Daily Living (IADL) scale by establishing a hierarchy of item 'difficulty' and 'discrimination'. METHODS: this cross-sectional study applied IRT methods to the analysis of IADL outcomes. Participants were 202 members of the Scottish Dementia Research Interest Register (mean age = 76.39, range = 56-93, SD = 7.89 years) with complete itemised data available. RESULTS: a Mokken scale with good reliability (Molenaar Sijtsama statistic 0.79) was obtained, satisfying the IRT assumption that the items comprise a single unidimensional scale. The eight items in the scale could be placed on a hierarchy of 'difficulty' (H coefficient = 0.55), with 'Shopping' being the most 'difficult' item and 'Telephone use' being the least 'difficult' item. 'Shopping' was the most discriminatory item differentiating well between patients of different levels of ability. CONCLUSIONS: IRT methods are capable of providing more information about functional impairment than a summed score. 'Shopping' and 'Telephone use' were identified as items that reveal key information about a patient's level of ability, and could be useful screening questions for clinicians.
Authors: John D Peipert; Lee A Jennings; Tahmineh Romero; Ron D Hays; Neil S Wenger; Emmett Keeler; David B Reuben Journal: J Am Geriatr Soc Date: 2020-11-30 Impact factor: 5.562
Authors: Astrid Etman; Carlijn B M Kamphuis; Frank H Pierik; Alex Burdorf; Frank J Van Lenthe Journal: Int J Health Geogr Date: 2016-11-15 Impact factor: 3.918