A Psychometric Evaluation of the RAND 36 Scale Among Japanese Adults With Insomnia in 2023

OBJECTIVES: The RAND 36-Item Health Survey 1.0 Questionnaire is a widely used health-related quality of life (HRQoL) measure. Few studies have evaluated the psychometric properties of the instrument in a disease-specific population. This study aimed to assess the validity of the RAND-36 as an HRQoL measure among Japanese participants of the National Health and Wellness Survey (NHWS) with insomnia, using classical test (CTT) and item-response theory (IRT) approaches.

METHODS: Adults (age 18 and up) with a self-reported physician diagnosis of insomnia from the 2023 Japan NHWS (n=2,084) were included in the analysis. Response frequencies of all items in the RAND-36 were used to check for a balanced response, and associations between items were assessed with Pearson correlation tests. To determine the internal reliability of the RAND-36, Cronbach’s alphas were computed globally and for the 8 subscales. Subscale homogeneity was tested with Mokken scaling analysis. Lastly, the unidimensionality of subscales was identified with automatic item selection procedure (AISP) at increasing homogeneity thresholds.

RESULTS: Among the sample, all responses were well represented in the data, and no negative associations between items were found after reverse-coding (all Pearson’s rho>0.0). The internal consistency reliability was acceptable, with Cronbach’s alphas greater than 0.7 for the global composite measure and all 8 subscales (α=0.724 - 0.937). The Mokken scaling homogeneity coefficients indicated strong homogeneity within the subscales (H = 0.502- 0.753). Tests for local independence did not indicate any violations of learning effects within the intra-subscale items. Analysis of monotonicity detected violations for the Energy/Fatigue and Emotional Well-being subscales.

CONCLUSIONS: A stepped psychometric analysis of the RAND-36 has provided evidence of the scale’s validity to measure HRQoL in an insomnia-burdened patient population. Additional IRT analyses will be conducted to further assess the validity of RAND-36 in this cohort.