Responsiveness and Construct Validity of the Health Utilities Index in Patients with Dementia
Published In: Medical Care, v. 47, no. 6, June 2009, p. 651-661
Posted on RAND.org on January 01, 2009
BACKGROUND: Assessment of health-related quality of life (HRQOL) is important for cost-effectiveness analyses, but the validity of generic HRQOL instruments has not been adequately evaluated in persons with dementia. OBJECTIVE: To evaluate the validity (including responsiveness to change) of the Health Utilities Index Mark 2 (HUI2) and Mark 3 (HUI3), 2 commonly used generic HRQOL measures, in patients with dementia. SUBJECTS: Four hundred eight patient-caregiver dyads in an 18-month dementia care management trial. METHODS: The authors assessed construct validity by evaluating correlations of proxy (caregiver)-reported HUI2 and HUI3 with the Blessed Dementia Rating Scale, the Charlson Comorbidity Index, and a behavior rating scale. Responsiveness was estimated using effect size (ES) statistics for behavior scale change (unchanged, small, medium, large change) and for residential status change (home to skilled nursing facility), as a global external change criterion. RESULTS: The HUI2 and HUI3 were responsive to behavioral worsening (multiattribute ES range: -0.48 to -0.78) and global decline (multiattribute ES range: -0.50 to -0.76), but not improvement. The HUI2 was more responsive than the HUI3. Correlations with the Blessed Dementia Rating Scale (r = -0.69 with both HUI2 and HUI3 multiattribute scores) and behavior scale (r = 0.44 and 0.41, respectively, for HUI2 and HUI3 multiattribute scores) supported the validity of the HUI in patients with dementia. CONCLUSIONS: Our results support the construct validity of the proxy-rated HUI2/3 in patients with moderate to severe dementia, but responsiveness results were mixed. Further studies are needed of the HUI2/3's validity, including responsiveness, in patients across the full range of dementia severity, using both self and proxy report, with particular attention to the impact of general population preference weights. When possible, multiple HRQOL measures need to be used to confirm the robustness of the findings. The proxy-rated HUI should be used in patients with moderate to severe dementia, but the self-rated HUI may be appropriate for subjects with milder cognitive impairment.