PURPOSE: The purpose of this study was to validate DEQ-5 and mOSDI in children using Rasch analysis and determine the optimal cutoff value for DEQ-5 to differentiate dry eye symptoms defined by mOSDI in children aged 6-17 years. In addition, the repeatability of mOSDI and DEQ-5 were investigated in children. METHODS: A total of 225 children were recruited. Both DEQ-5 and mOSDI were administered to participants in a randomized sequence. Retests of the two questionnaires were performed after 30 min in 72 participants. A receiver operating characteristic (ROC) curve was generated to determine the sensitivity and specificity of the DEQ-5 questionnaire for diagnosis of dry eye symptoms. Repeatability of both questionnaires was evaluated using coefficient of repeatability (CoR) and 95% limits of agreement. RESULTS: The median of mOSDI score was 11.4 (Inter-quartile range: 4.5-20.5). The median DEQ-5 score was 4.0 (Inter-quartile range: 2.0-7.0). The area under the curve of the ROC curve for DEQ-5 was 0.78 (p < 0.0001). A DEQ-5 threshold of 4.5 yielded maximum sensitivity (72.2%) and specificity (66.1%) to differentiate dry eye symptoms in children. Females had significantly higher OSDI scores than males. CoR was ±4.8 for DEQ-5 and ±16.6 for mOSDI. The 95% limits of agreement were 5.5 to -4.2 for DEQ-5 and 16.5 to -16.7 for modified OSDI. Rasch analysis demonstrated that the five-level response scales in both the mOSDI and DEQ-5 produced disordered item thresholds. Collapsing the response scales from five to three categories improved overall model fit. CONCLUSIONS: Although we identified a DEQ-5 cutoff value of ≥5 for differentiating dry eye symptoms in children, which is different from that used in adults (≥6), the five-level response scales of both the mOSDI and DEQ-5 may not be optimal for the pediatric population and certain items might not be adequate to measure dry eye symptoms, thus we do not recommend using DEQ-5 in children to measure dry eye symptoms. This underscores the need to develop a dry eye questionnaire specifically designed for children to ensure more accurate assessment.
Pang et al. (Wed,) studied this question.