Item Level Psychometrics of the Dizziness Handicap Inventory in Vestibular Migraine and Meniere’s Disease

Abstract

Objectives: Evaluate the measurement properties of the Dizziness Handicap Inventory (DHI) using item response theory in patients diagnosed with vestibular migraine (VM) and Meniere’s disease (MD). Design: One hundred twenty-five patients diagnosed with VM and 169 patients diagnosed with MD by a vestibular neurotologist according to the Bárány Society criteria in two tertiary multidisciplinary vestibular clinics and who completed the DHI at their initial visit, were included in the study. The DHI (total score and individual items) was analyzed using the Rasch Rating Scale model for patients in each subgroup, VM and MD, and as a whole group. The following categories were assessed: rating-scale structure, unidimensionality, item and person fit, item difficulty hierarchy, person-item match, and separation index, standard error of measurement, and minimal detectable change (MDC). Results: Patients were predominantly female (80% of the VM subgroup and 68% of the MD subgroup) with a mean age of 49.9 ± 16.5 years and 54.1 ± 14.2 years, respectively. The mean total DHI score for the VM group was 51.9 ± 22.3 and for the MD group was 48.5 ± 26.6 (p > 0.05). While neither all items nor the separate constructs met all criteria for unidimensionality (i.e., items measuring a single construct), post hoc analysis showed that the all-item analysis supported a single construct. All analyses met the criterion for showing a sound rating scale and acceptable Cronbach’s alpha (≥0.69). The all-item analysis showed the most precision, separating the samples into three to four significant strata. The separate-construct analyses (physical, emotional, and functional) showed the least precision, separated the samples into less than three significant strata. Regarding MDC, the MDC remained consistent across the analyses of the different samples; approximately 18 points for the full analyses and approximately 10 points for the separate construct (physical, emotional, and functional). Conclusions: Our evaluation of the DHI using item response theory shows that the instrument is psychometrically sound and reliable. The all-item instrument fulfills criteria for essential unidimensionality but does seem to measure multiple latent constructs in patients with VM and MD, which has been reported in other balance and mobility instruments. The current subscales did not show acceptable psychometrics, which is in line with multiple recent studies favoring the use of the total score. The study also shows that the DHI is adaptable to episodic recurrent vestibulopathies. The total score shows better precision and separation of subjects in up to four strata compared to the separate construct that separate subjects into less than three strata. The measurement error smallest detectable change was found in our analysis to be 18 points, which means any change in the DHI of less than 18 points is not likely to be clinically significant. The minimal clinically important difference remains indeterminate.