Not Seeing and Seeing Things: Dementia, Visual Function, and Psychiatric Symptoms

The contrasting roles of cognition and vision in the generation of some neuropsychiatric manifestations have always been of interest. Hallucinations, for example, have long been known to occur with dementia.1 However, they can also occur with binocular visual loss as “release phenomena” in the Charles Bonnet syndrome, which is typically diagnosed in the absence of cognitive dysfunction. It may not always be a binary question of one to the exclusion of the other. These two factors—cognition and vision—could interact. For example, although Charles Bonnet syndrome is attributed to deafferenting of cortex by visual loss, there is some evidence that subjects with this syndrome also have mild neuropsychological impairments2,3 and may be at greater risk of developing dementia in the coming months.4 The inference is that mild cognitive dysfunction may be a factor modulating the risk of release hallucinations after visual loss. In this issue of the Journal of Neuro-Ophthalmology, Li and Hamedani5 reverse the question and broaden it. They ask whether visual dysfunction is a modulating factor for hallucinations or other neuropsychiatric symptoms among persons living with dementia. There is already some evidence to support this conjecture. Twenty-five years ago, a small British study of 50 patients with Alzheimer disease found that those with hallucinations had worse visual acuity and were more likely to have cataracts.6 Notably, improvement of vision through refraction resolved hallucinations in 4 of 6 patients. Buried in the results is the fact that the investigators also found a relation between reduced visual acuity and delusions. Li and Hamedani take a different methodological approach. They report a population-based survey of 624 people with cognitive impairment extracted from the large Health and Retirement Study (HRS), a survey that has been repeated every 6 years since 1992. Large population surveys gain in numbers and statistical power but lose in clinical precision. In this study the diagnosis of dementia is based on the Clinical Dementia Rating (CDR) Scale, with dementia defined by a CDR score of 0.5 or more, where 0.5 indicates “questionable dementia,” most typically mild cognitive impairment. Similarly, visual impairment is defined only as either a) a subjective response that near or far vision is “fair or poor” or b) near visual acuity worse than 20/40 with whatever reading glasses they use. This limited visual evaluation thus ignores potential contributions from peripheral vision and other aspects such as contrast sensitivity, which, in some studies, is more predictive of incident dementia.7 Psychiatric symptoms were gleaned from another questionnaire, the Neuropsychiatric Inventory, and the results grouped to create the categories of hallucinations, psychosis, depression, mania, and agitation. The report by Li and Hamedani has 3 key findings. The first is that the subjective reports of visual difficulty were associated with increased odds of hallucinations, depression, and agitation. The second is that visual acuity of worse than 20/40 was associated with hallucinations, psychosis, and mania. It is difficult to gauge the relevance of this observation. However, the third and most important result is that, after adjusting for confounding variables—age, gender, ethnicity, education, work status, hypertension, stroke, and diabetes—these associations were not statistically significant. It is worth mentioning how this contrasts with their prior report on hallucinations in the elderly.8 That analysis combined the same HRS data set and another one (the National Health and Aging Trends Study). It found that in people aged older than 65 years self-reported visual problems carried an odds ratio of about 1.5–2 for hallucinations, even after adjusting for covariates. Despite the fact that the associations in the current article disappeared after adjustment, Li and Hamedani conclude that visual impairment is associated with neuropsychiatric manifestations in patients with dementia though “at least some” of the relationship is due to other factors. Without a significant post-adjustment result we could just as well conclude that all of the relationship is due to those other factors. One possibility, though, is that the confounds contribute to the risk of hallucinations because they increase the risk of visual impairment. Their Table 1 suggests that the people most at risk of neuropsychiatric symptoms are white retired women aged older than 80 years who have diabetes, did not graduate from high school, and are not living in a nursing home. Some of these factors likely increase the risk of visual problems (e.g., age and diabetes), but with others, a link to visual impairment is not so apparent. It would have been of interest to learn which confounds were key. Perhaps it would also have made a more convincing case if they had examined for an effect of the severity of visual impairment. Li and Hamedani pose a plausible hypothesis. Visual degradation and cognitive impairment might combine to increase the frequency of visual hallucinations. Some of the other neuropsychiatric symptoms may also be associated with visual loss, even without dementia. As they note, a study using the National Health and Aging Trends Study data found that subjective visual impairment was associated with depression and anxiety in those aged older than 65 years.9 Similarly, a Brazilian study found a relationship between visual or hearing loss and depression in middle-aged to older adults.10 Thus, it would not be surprising to find a similar relationship between visual loss and depression in those with dementia. As for mania and agitation, visual impairment could lead to agitation by reducing orienting cues for a cognitively impaired person, but a causal role in mania is more difficult to understand. Ultimately, though, the effect of their confounds is to leave us without a definite answer.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.