What level of visual impairment is required to cause disability? Typically, this question has been answered by relating disease severity to disability measures, or simply by using convenient numerical cut-offs. Jammal and colleagues take a different approach: they used latent class analysis (LCA) on NEI-VFQ questionnaire data from 263 individuals and found their analysis supported a separation of these individuals into two groups. The quality of this separation, judged by model fit indices, was high. It should be noted that LCA does not have a preconceived notion about how it creates groups; it simply creates groups with individuals as similar to each other as possible with regards to their pattern of questionnaire responses.
These two groups created differed primarily with regards to their vision, with one group demonstrating more better-eye (MD = -6.0 vs. -2.5 dB) and worse-eye (MD = -13.4 vs. -6.1 dB) VF damage, and also worse VA (logMAR of 0.05 vs. -0.02). The two groups also differed with regards to their age, with the older group showing greater VF damage. The authors label these groups as disabled and non-disabled, though the trait described by LCA is latent (hidden), such that the groups may differ by the amount of disability, the pattern of disability, or both. To this point, the amount of VF damage overlapped substantially across the two groups, with several eyes in the group labeled as disabled showing little VF damage, and several eyes in the group labeled as non-disabled showing significant VF damage. Thus, while it is very intriguing that responders could be separated into groups so well, further work is required to clarify exactly what separated these groups, whether these groups can be used to classify disability in glaucoma, and whether similar groupings would have been noted in a population with more advanced glaucoma patients.