In this study, we developed a Chinese version of the GQL-15 and evaluated its psychometric properties in Chinese glaucoma patients. Overall, we demonstrated that the GQL-15-CHI was reliable, valid, and able to discriminate the severity of glaucoma.
With respect to the reliability, the overall Cronbach’s α coefficient was 0.96 in this study, which was similar to that of the original version (0.95) . The Cronbach’s α coefficients for the subscales were all greater than 0.70, indicating satisfactory internal consistency of the questionnaire in the studied population. Considering that intraclass correlation coefficients greater than 0.70 are generally accepted as satisfactory, our results suggested that the GQL-15-CHI demonstrated good test-retest reliability. High item-scale and item-subscale correlations further confirmed the excellent homogeneity of the questionnaire. A ceiling effect was noted for the subscale of central and near vision, peripheral vision, and outdoor mobility. Good performance in the three subscales may indicate patients’ adaptation in handling the daily activities involving these aspects. Furthermore, GQL-15-CHI is probably to be sensitive to any worsening of QoL throughout all four domains as no floor effect was observed.
Because the GQL-15 was designed for subjective assessment of visual functions in glaucoma patients to guide the treatment, it was important for the questionnaire to have strong discriminatory power. Our results showed that the GQL-15-CHI had a good capacity to discriminate the severity of glaucoma. Greater difficulty with vision-related activities and poorer QoL were found to be highly correlated with increasing disease severity even after adjustment for all sociodemographic and clinical confounders. This finding indicated that the subjective visual function assessment could function as a complement to objective visual measures, and was informative and predictive in the follow-ups of glaucoma patients. This suggests value for the ophthalmologist in establishment of a baseline QoL upon diagnosis of glaucoma and periodically thereafter.
Nelson et al.  observed a marked difference in GQL-15 summary scores only between patients with mild and severe glaucoma. Onakoya et al.  detected statistically significant differences between patients with moderate and severe glaucoma and mild and severe glaucoma. However, in our study the differences among patients in all stages of glaucoma were statistically significant, consistent with results reported by Goldberg et al. . However, after categorization, the mean summary scores for moderate and severe glaucoma in our study (28.73 and 44.55) were significantly higher than those reported by Nelson et al.  (22.5 and 24.9) and Onakoya et al.  (20.58 and 32.65). The trend among the mean summary scores for different stages of glaucoma in our study (20.76, 28.73, and 44.55) was similar to that previously found by Goldberg et al.  (21.7, 29.6, and 40.0). Potential explanations for this discrepancy may lie in cultural and/or social variances, varied sample sizes (China: 508; Nigeria: 132; Australia: 121; England: 47), different methods for categorization of patients, and dissimilar patient selection criteria. With respect to patient selection, different age distributions (China: 18–88 years; Nigeria: ≥40 years; Australia: ≥44 years; England: 53–81 years) and types of glaucoma (China: POAG, NTG, PACG and SG; Nigeria: POAG; Australia: POAG; England: POAG, NTG, PACG and SG) were partially responsible for the discrepancies. Moreover, Nelson et al.  excluded patients with progressive visual fields and visual impairment to reduce the effect of visual acuity on the study; whereas our study, Goldberg et al. , and Onakoya et al.  did not. Although Nelson’s study presented with a good homogeneity, it probably limited the diversity of disease severities.
The results of the factor analysis in our study were not fully conformed to the item distribution of the four dimensions in the original version. The explanations underlying this difference may relate to the following aspects: First and foremost, compared to the Nelson’s patient sample, we had a much larger and heterogeneous patient population. In a certain sense, the finding in our study demonstrated that the original version may not generalize to the patients of more severe glaucomatous impairment. Second, the cultural differences may still exist and result in a deviation in perceiving the items of the questionnaire, although we modified some items to make it more adaptable to Chinese patients. Third, social variance could cause a difference in the response to some specific items, e.g. for the item “crossing the road”, the traffic is crowded in China for the largest population in the world, and this is more pronounced in the metropolitans, such as Shanghai. In addition, item 13 did not fit a specific dimension, while items 2, 3 and 12 were far from being perfect. Consequently, the Chinese patients were more likely to express difficulty in response to this item. Rasch analysis, may be used as a possible method to reengineer the GQL-15-CHI to be a better-structured questionnaire in future research .
Although the psychometric characteristics of the GQL-15-CHI were comparable with those of the original version, the following limitations should be considered when interpreting the results of this study. First, reliance on self-reported visual symptoms may be influenced by recall bias and personality factors. Second, all patients in this study were recruited from a single tertiary institution and belonged to the Chinese Han ethnicity. Although this specialized hospital received patients from all parts of China and Han ethnicity takes up more than 91% of total Chinese population, there may still exists selection bias. In addition, we made six months as the duration cut point when recruiting participants, which can also cause a deviation in selecting the patients. Third, we did not include controls in this study, thereby limiting the conclusions that we can draw from the study regarding whether or not it can discriminate between glaucoma and other visual-impaired disease. Fourth, we did not cover all ocular examinations originally reflected by the original GQL-15 design, such as contrast sensitivity, critical flicker frequency, dark adaptation, glare disability, and stereoacuity. Lastly, this study did not carry out over time for longitudinal observation, and confirmatory factor analysis was not conducted in another sample to validate our findings. Therefore, a larger series with long-term follow-up including control groups and other parameters is needed to further validate the GQL-15-CHI in Chinese populations. Despite these limitations, this study included a large sample size. Moreover, the comprehensive ophthalmic examinations were performed by one expert and therefore avoiding the inter-observer errors.