Psychometric validation of the modified low luminance questionnaire in the X-linked retinitis pigmentosa patient population

Background X-linked retinitis pigmentosa (XLRP) is a rare inherited eye disease that has negative impacts on patients’ health-related quality of life and increased lifetime burden. The Low Luminance Questionnaire (LLQ) was modified to enhance clarity and relevance for use in XLRP. This study evaluated the psychometric measurement properties of the 30-item modified Low Luminance Questionnaire (mLLQ) in adults with XLRP. Methodology Data from 2 clinical studies of participants with XLRP underwent psychometric and statistical analyses. Validation analyses included: item-level (missingness, distribution, correlations, discrimination), Rasch, confirmatory factor (CFA), and tests of reliability (internal consistency, test-retest), validity (convergent/divergent, known groups), and responsiveness. Results Data from 141 adult participants in the EXPLORE XLRP-2 study were included, of which 115 (81.6%) were male, mean age was 42.6 years (range 18–88 years), and 49 (35.0%), 36 (25.7%) and 55 (39.3%) had mild, moderate or severe disease, respectively. mLLQ domains demonstrated excellent internal reliability (Cronbach’s alpha: 0.843 to 0.957). Convergent validity was supported by moderate-to-strong correlations with reference measures ( r > 0.60 in expected domains). The mLLQ domains demonstrated acceptable to good capacity to discriminate between groups of participants known to differ on XLRP disease stage. Item-level rates of missing responses were very low across all items (range 0.0% to 2.1%) and each mLLQ response option was used for all items. Floor effects were identified for 22/30 items (73.%). Items within the same domain correlated well with one another and with their respective domain. 63% (19/30) of items had a discrimination index > 0.40. The hypothesized 6-factor CFA model provided relatively good fit to the data. Assigning 2 Extreme Lighting items to the Driving domain also demonstrated an improved model-data agreement. Rasch analysis suggested a lack of items to assess participants that are more severe, consistent with the observed floor effects for most items. Strong evidence of responsiveness was observed for all mLLQ domains except Driving and Peripheral vision domains, where observed evidence was weaker. Conclusions Psychometric analyses indicate the mLLQ has acceptable measurement properties to assess self-reported task difficulty under low luminance and at night experienced by adult patients with XLRP.