The dimensionality and validity of the Older Americans Resources and Services (OARS) Activities of Daily Living (ADL) Scale. (Doble & Fisher, 1998)

Doble, S. E., & Fisher, A. G. (1998). The dimensionality and validity of the Older Americans Resources and Services (OARS) Activities of Daily Living (ADL) Scale. Journal of Outcome Measurement, 2(1), 4-24.

School of Occupational Therapy, Dalhousie University, Halifax, Nova Scotia, Canada.

The psychometric properties of the OARS ADL scale, comprised of seven physical activities of daily living (PADL) and seven instrumental activities of daily living (IADL) items, were examined using a Rasch measurement approach. Two of the PADL items failed to demonstrate acceptable goodness-of-fit with the measurement model but the remaining 12 items could be combined into a single measure of ADL ability. Although the OARS ADL scale was designed to identify those community-dwelling elderly who need supports and services to continue to live in the community, the scale items were found to be poorly targeted to community-dwelling elderly since almost half of our sample received maximal scores. Rasch analysis identified how we might improve the sensitivity of the OARS ADL scale but its utility in outcome and longitudinal studies remains questionable.

Notes:

Literature review

1. The Older Americans Resources and Services (OARS) Activities of Daily Living (ADL) Scale
   
   
   1. 7 Physical Activities of Daily Living (PADL)
      
   2. 7 Instrumental Activities of Daily Living (IADL)
      
2. The importance of PADL and ADL; the dependency in PADL and IADL are associated with
   
   
   1. Poor quality of life
      
   2. Increased risk of nursing home placement
      
   3. Increased risk of death
      
3. 2 major problems associated with self- and proxy-based assessment of PADL and IADL
   
   
   1. Evaluating PADL and ADL with few items
      
      
      1. Ceiling and floor effect
         
      2. Compromised reliability
         
      3. In a study conducted by Suurmeijer et al (1994), it is found that when PADL and IADL items are included in a single scale, the unidimensionality of the scale items is compromised. However, results of other researchers supported the unidimensionality of the combined scale (Finch, Kane, & Philp, 1995; Kempen & Suurmeijer, 1990, Siu, Reuben & Hays, 1990; Silverstein, Fisher, Kilgore, Harley, & Harvey, 1992; Spector et al., 1987)..
         
   2. Summing ordinal rating
      
      
      1. Compromise the true measure
         
      2. Guttman scaling is a deterministic model and it requires the data to maintain an absolutely rigid structure, which is rarely the case in social science. In addition, this model holds that people will pass all items lower than their ability level and fail all items higher than their ability level. As a result, the differences have to be big between item difficulties in order for a scale to confirm to this expectation. The resulting trade-off is that it might reduce the sensitivity to identifying changes within an individual over time or small changes across individuals.
         
4. Rasch model
   
   
   1. Difficulty level of items are based on the Likelihood of the item being passed
      
   2. Rasch analysis could generates item-parameter related statistics
      
      
      1. a calibrated difficulty level for each test item
         
      2. Mean square residual (MnSq) with an expected value of 1. MnSq greater than 1.4 is no good
         
      3. A standardized goodness-of-fit statistics (z) with an expected value of 0. z greater or equal to 2 is no good.
         
   3. Rasch model also provided us with estimation for person-related parameter, theta, and the associated statistics. Greater separation suggests higher sensitivity.
      

Subjects

Elderly, at least 60 years old and at least one health condition affecting daily living

Analyses

1. Conducted using BIGSTEPS:
   
   
   1. Item separation statistics: whether subjects separate items into levels of difficulty
      
   2. Person separation statistics: whether items separate subjects into levels of ability
      
   3. Item difficulty measure
      
   4. Person ability measure
      
   5. Goodness-of-fit statistics
      
   6. Unidimensionality of the ADL scale was determined by examining MnSq and associated z fit statistic for each item
      
2. Greater separation suggests increased sensitivity of the scale
   
3. If more than 5% of the subjects' failed to demonstrate goodness-of-fit with the measurement model when z is set at 2, then there might be problem with the validity of the scale
   

Results

1. Fit of items to the measurement model p. 12-17
   
   
   1. Item separation statistic
      
   2. Items located within .25 logitsè actual order might be different because the distance is too small
      
   3. Item order (Finch et al., 1995, Fillenbaum, 1988; Kempen & Suurmeijer, 1990; Siu et al., 1990; Silverstein et al., 1992; Spector et al., 1987; Suurmeijer et al., 1994; Lawton and Brody, 1969)
      
   4. Check for large gaps to see whether we might be missing items to measure certain ability level
      
   5. Check for items with large error estimates
      
   6. Check goodness-of-fit statistics
      
   7. Get rid of the bad items and redo the analyses
      

1. Fit of the persons to the measurement model P. 17-20
   
   


1. Person separation statistics
   


2. Check the number of groups
   

Doble and Fisher (1998) conducted a study to evaluate the psychometric properties of the Older Americans Resources and Services (OARS) Activities of Daily Living (ADL) Scale, which contains 7 ADL and 7 IADL items. Participants were community-dwelling elderly adults in either Canada and United States age 60 and above. Most subjects were interviewed directly through self-report while, for certain participants', such as those having dementia or memory impairments, information was obtained through proxy report. All except for the continence item were rated based on a 3 point scale with 0 indicating the subject is completely incapable of performing the task and 2 indicating no problem at all. The authors used the program BIGSTEPS to run the Rasch model analysis. Their results found that, after dropping two misfitting items (i.e., bathing and continence), the remaining items were found to be hierarchically ordered and measure the same underlying construct (e.g., perceived ADL ability). However, based on the parameter estimates of the ability score, the authors concluded that the OARS ADL items have limited applications for the community-dwelling individuals they were designed for.