Test–retest reliability of health utilities index scores: Evidence from hip fracture

Abstract

There is relatively little evidence on the test-retest reliability of utility scores derived from multiattribute measures. The objective was to estimate test-retest reliability for Health Utilities Index Mark 2 (HUI2) and Mark 3 (HUI3) utility scores in patients recovering from hip fracture. We enrolled an inception cohort of hip fracture patients within 3 to 5 days of surgery. Baseline assessments included the Functional Independence Measure (FIM), Folstein Mini-Mental State Examinations, and the HUI2 and HUI3 questionnaire. Follow-up assessments at 1, 3, and 6 months also included a global change question. Test-retest reliability was assessed as agreement between 3- and 6-month scores using the intraclass correlation coefficient (ICC). Two approaches were used to classify patients as stable; a third approach based on the generalizability theory was also used. Patients were classified as stable if their FIM overall scores changed by 10 points or fewer and if they classified themselves as having experienced no or only a little change according to their global change question. Complete data at both the 3- and 6-month assessments based on self-report were available for 196 patients; 141 patients with complete data were classified as stable. The ICCs for HUI2 and HUI3 for stable patients were 0.71 and 0.72; the ICCs derived from the generalizability theory were 0.76 and 0.77. Test-retest reliability for HUI in this cohort was similar to reliability estimates for other preference-based multiattribute and generic health-profile measures--in the acceptable range for making valid group-level comparisons.