Multidimensionality of the PROMIS sleep disturbance 8b short form in working adult populations

Abstract

ObjectivesThe Patient-Reported Outcomes Measurement Information System sleep disturbance measures were developed using item response theory assumptions of unidimensionality and local independence. Given that sleep health is multidimensional, we evaluate the factor structure of the Patient-Reported Outcomes Measurement Information System sleep disturbance 8b short form to examine whether it reflects a unidimensional or multidimensional construct. MethodsSix full-time working adult samples were collected from civilian and military populations. Exploratory and confirmatory factor analyses were conducted. Single-factor and two-factor models were performed to evaluate the dimensionality of sleep disturbance using the 8b short form. Sleep duration and subjective health were examined as correlates of the sleep disturbance dimensions. ResultsAcross six working adult samples, single-factor models consistently demonstrated poor fit, whereas the two-factor models, with insomnia symptoms (ie, trouble sleeping) and dissatisfaction with sleep (ie, subjective quality of sleep) dimensions demonstrated sufficient fit that was significantly better than the single-factor models. Across each sample, dissatisfaction with sleep was more strongly correlated with sleep duration and subjective health than insomnia symptoms, providing additional evidence for distinguishability between the two sleep disturbance factors. ConclusionsIn working adult populations, the Patient-Reported Outcomes Measurement Information System sleep disturbance 8b short form is best modeled as two distinguishable factors capturing insomnia symptoms and dissatisfaction with sleep, rather than as a unidimensional sleep disturbance construct.