Free
Poster Session
Issue Date: July 01, 2015
Published Online: February 09, 2016
Updated: January 01, 2020
Cracking the Code: Determining Whether a Single Measurement Model Underlies Different Activities of Daily Living (ADL) Scales
Author Affiliations
  • Medical University of South Carolina
  • Medical University of South Carolina
  • Medical University of South Carolina
  • Medical University of South Carolina
  • Medical University of South Carolina
  • Medical University of South Carolina
  • Medical University of South Carolina
  • Medical University of South Carolina
  • Medical University of South Carolina
Article Information
Evidence-Based Practice / Neurologic Conditions / Rehabilitation, Participation, and Disability / Stroke / Assessment/Measurement
Poster Session   |   July 01, 2015
Cracking the Code: Determining Whether a Single Measurement Model Underlies Different Activities of Daily Living (ADL) Scales
American Journal of Occupational Therapy, July 2015, Vol. 69, 6911500105. https://doi.org/10.5014/ajot.2015.69S1-PO5091
American Journal of Occupational Therapy, July 2015, Vol. 69, 6911500105. https://doi.org/10.5014/ajot.2015.69S1-PO5091
Abstract

Date Presented 4/17/2015

This study explores the feasibility of developing an underlying measurement theory of activities of daily living (ADL) item hierarchy across ADL instruments used in rehabilitation. This is a preliminary study to provide applicable information and to further facilitate evidence-based professional development and practice.

SIGNIFICANCE: Activities of daily living (ADLs) are an essential focus in occupational therapy practice. However, the current measurement scales used to assess ADLs lack a general measurement model that would jointly represent ADL tasks measured by different ADL scales. Discovery of a single underlying measurement model could (1) curtail the necessity to develop new ADL instruments and (2) support measuring patients across the continuum of care by providing a basis for translating scores across different ADL scales.
INNOVATION: This study is innovative conceptually and methodologically. Conceptually, this study examines the hypothesis that there is an underlying measurement theory across different ADL instruments. Methodologically, this study uniquely combines the occupational therapy skill of task analysis with advanced statistical approaches of item response theory (IRT) and multiple regressions.
APPROACH: Hypotheses were as follows: (1) Different ADL instruments show a similar item-difficulty hierarchy, and (2) a universal measurement theory underlies ADL instruments. Hypotheses will be supported if the two instruments show similar item difficulty hierarchies and regression models.
Presently, there are more than 85 published ADL scales, with the vast majority showing acceptable reliability and validity. Unfortunately, there is a lack of a general underlying measurement theory across ADL instruments that could support translating scores between instruments. Having an underlying measurement theory could support the hypothesis that there is the same latent trait measured across different ADL instruments.
METHOD: This is a secondary and cross-sectional study design using the published item difficulty hierarchy of the Stroke Impact Scale (SIS) and the International Classification of Functioning, Disability and Health (ICF)—Activity Measure (AM), Self-Care domain, as dependent variables. Eight task analysis–derived predictors served as independent variables.
Individuals with stroke completed the SIS on the basis of a 2003 study by Duncan and colleagues. Individuals with back pain, upper- or lower-extremity injuries, or spinal cord injuries completed the ICF–AM. This study used item calibrations from the 8 ADL items of the SIS and the 78 ADL items of the ICF–AM, Self-Care domain. Task analysis ratings were dichotomous. Regression stepwise analysis was completed using SPSS Version 21. The model fit of ICF–AM, Self-Care domain, was examined using Winsteps Rasch analysis.
RESULTS: Hypothesis 1 was supported by similar ADL item hierarchies in the two instruments (e.g., ambulation was more difficult than eating). Hypothesis 2 was partially supported when both ADL instruments fit with a significant model (p < .01) that explained more than 50% of the variance for each item hierarchy. However, amount of variance accounted for by the dependent variables differed across the instruments—that is, antigravity is the main predictor for the SIS, whereas number of joints, precision, and range of motion are the main predictors in the ICF–AM, Self-Care domain, measure.
CONCLUSION: The findings only partially support a common measurement theory underlying the SIS and the ICF–AM, Self-Care domain, measure. Several factors may have contributed to the discrepant findings—for example, differences across instruments in the level of detail of item descriptions, item calibrations based on different populations, and lack of standardization in task analysis. Future studies addressing these issues may clarify the potential for identifying a common measurement model underlying ADL scales.