Computer-adaptive tests use respondents' answers to previous questions to select the subsequent questions. They are gaining popularity for their increased measurement precision and decreased administration time compared to static questionnaires. The purpose of this study was to estimate the test-retest reliability and construct validity of the computer-adaptive test version of a participation measure, the Late Life Disability Instrument (LLDI-CAT) for people with COPD and to compare scores and administration time with those of the static LLDI. Among 76 older adults with COPD, scores on the LLDI-CAT were compared to scores on measures of related constructs, between groups based on symptom severity, prognosis and frailty phenotype, and to scores on the static LLDI. A subsample of 28 people completed the LLDI-CAT a second time within one week of the initial administration for test-retest reliability. The LLDI-CAT had very good test-retest reliability (ICC2,1 0.88; SEM 2.74 points), fair correlations with physical function (r = 0.37-0.50), anxiety (r=-0.42), and depression (r=-0.50), fair to moderately-strong correlations with quality of life (r = 0.48-0.63), and strong correlation with the static LLDI limitation domain (r = 0.80). The LLDI-CAT scores differed between people with different symptom severity, prognosis and frailty phenotype (p ≤ 0.004). The mean administration time for the LLDI-CAT was 3.3 (1.5) minutes, less than that of the static LLDI at 6.3 (2.8) minutes (p < 0.001). The LLDI-CAT demonstrates evidence of test-retest reliability and construct validity, and correlates well with the limitation domain of the static LLDI for people with COPD. The LLDI-CAT can be used to assess participation for this population.
Keywords: Community participation; computer adaptive testing; disability; psychometrics; social participation.