PP34Rasch analysis reveals combining ADL/IADL summary scores from different residential settings is inappropriate

• Published in: Journal of epidemiology and community health (1979) Vol. 69; no. Suppl 1; p. A67
• Main Authors: Lutomski, JE, Krabbe, PFM, den Elzen, WPJ, Olde-Rikkert, MGM, Steyerberg, E W, Muntinga, ME, Bleijenberg, N, Kempen, Gijm, Melis, RJF
• Format: Journal Article
• Language: English
• Published: 01.09.2015
• Subjects: Netherlands
• ISSN: 0143-005X
• DOI: 10.1136/jech-2015-206256.131

BackgroundNumerous Activities of Daily Living (ADL) and Instrumental Activities of Daily Living (IADL) scales have been developed to measure functional limitations; however, not all of these scales have not undergone rigorous validity assessments. The primary aim of our study was to internally validate a 15-item dichotomous ADL/IADL index scale.MethodsData were extracted from The Older Persons and Informal Caregivers Survey Minimal DataSet (TOPICS-MDS, www.topics-mds.eu), a large, public access database containing prospectively collected data in the Netherlands. To assess validity, we applied a Rasch measurement model, which assumes that the probability of item endorsement is a logistic function of the relative distance between the item's difficulty and the participant's ability. Based on this model, six aspects of the ADL/IADL scale were assessed: (1) overall fit; (2) internal consistency; (3) individual item and person fit; (4) local dependency; (5) targeting; and (6) differential item functioning (RUMM 2030). All analyses were stratified by residential setting (community-dwelling or residential care facility).ResultsData were available on 21,926 community-dwelling older persons and 2,458 older persons residing in residential care facilities. 'Eating' was regarded as the easiest activity on the scale and 'performing household tasks' was the most difficult activity in both living situations. However, based on the location on the logit scale, the level of difficulty for certain items varied between residential settings, suggesting summary scores are not equivalent between these settings. Notably, the 15-item ADL/IADL scale exhibited poor fit to a Rasch model in community-dwelling older persons, suggesting sub-optimal construct validity. Moreover, the scale was poorly targeted for this population. In both residential settings, differential item functioning by gender and age group was observed for several items, underscoring potential measurement bias in the scale.ConclusionADL/IADL scales are often summated to indicate the number of functional limitations experienced by an individual. However, when performing validity studies, reviewing only the summation of functional limitations obscures critical information regarding varying difficulty levels between ADL/IADL items. Through applying a Rasch analysis, we found that the 15-item dichotomous ADL/IADL exhibited different patterns of difficulty in community-dwelling and residential care facility settings. Unless adjustments are undertaken, ADL/IADL summary scores retrieved from older persons residing in the community or residential care facilities should not be directly combined. This 15-item scale is poorly targeted for a community-dwelling older population, and further instrumental items are warranted to better discriminate between higher functioning adults.