Validity and responsiveness of the Minimum Data Set

Abstract

Objective: To determine the criterion validity and responsiveness to change over time of the Minimum Data Set (MDS).

Design: Cross sectional analysis comparing prospectively collected research data with MDS data reported from nursing homes on identical subjects. Longitudinal follow-up of subjects allowed for calculation of responsiveness to change over time.

Participants: Subjects (n = 140) participating in epidemiological research on the natural history of Alzheimer's disease in the University of Washington Group Health Cooperative Alzheimer's Disease Patient Registry (ADPR) who entered nursing homes in Washington State from January 1990 through March 1996.

Main outcome measurements: Spearman's correlation coefficients to compare MDS cognitive performance scale scores with Mini-Mental State Examinations (MMSE), MDS behavior domain scores with ADPR Physician Behavior checklist scores, and MDS activities of daily living (ADL) functioning scores with Dementia Rating Scale scores for ADLs. Partial correlations were used to control for the number of days between MDS and ADPR assessments, age, and function in the other two MDS ratings of ADL, cognition, or behavior. We determined responsiveness by using the mean change in an effect size (ES) calculation for each domain of interest.

Results: MDS measures in all three domains were correlated modestly to the analogous ADPR research instruments, with correlation coefficients 0.45 for cognition, 0.50 for behavior, and 0.59 for ADL functioning. Controlling for age, time between ADPR and MDS assessments, and the MDS ratings of the other two MDS domains did not improve the correlations significantly. In a subgroup of 60 subjects followed for an average of 636 days, the Cognitive performance scale of the MDS had a slightly larger effect size (ES = 0.60) than the Folstein Mini-Mental Status Exam (ES = 0.39) used by the ADPR. However, the Dementia Rating Scale (ES = 0.77) used by the ADPR researchers reflected more change over time than the ADL domain of the MDS (ES = 0.02). The MDS behavior domain (ES = 0.058) was comparable to the research behavior checklist (ES = 0.065).

Conclusions: These data demonstrate reasonable criterion validity of the MDS and the MMSE, ADL, and behavior rating scales and suggest that the MDS data is of adequate validity for research purposes in these areas. The MDS was less able to capture dementia-related change in ADLs in this population of patients with dementia even though the same nursing home staff completing the MDS often supplied the information for completing the research scales. As a result, the MDS may be limited as an outcome assessment instrument.