In this study of over 45,000 older adults in the U.S., we examined healthcare utilization patterns for POAG and nAMD among persons newly diagnosed with dementia. We found that compared to those without dementia, POAG and nAMD patients with dementia obtained less eye care and received less monitoring and treatment of their disease. The rate of reduction in care was fairly consistent across both disease cohorts, with patients having dementia averaging about 75% of the number of visits for non-dementia patients. In-office testing and treatment were also similarly reduced across both diseases, with dementia patients receiving about two thirds of the testing and injections that patients without dementia received. Prescription drug coverage, while reduced, was not as impacted, with dementia patients receiving only 17% less coverage. This may be due to the ease of obtaining prescription medications without the requirement of an office visit. It is possible that reduced POAG medication use in patients with dementia was due to an increase in the use of selective laser trabeculoplasty (SLT) or argon laser trabeculoplasty (ALT), but this is unlikely because these procedures would not impact the need for disease monitoring within the first year of diagnosis and treatment, and we adjusted our analyses for baseline glaucoma surgical history. While our study is not determinative of a causal relationship between dementia and reduced treatment rate, these findings suggest that persons living with dementia may be vulnerable to gaps in ophthalmic care.

Our results are consistent with previous studies showing that persons living with dementia are less likely to be seen by an ophthalmologist or receive cataract surgery [9,10,11]. Disparities in anti-VEGF treatment for AMD have been observed on the basis of race [15], but to our knowledge, the effect of dementia on healthcare utilization for nAMD has not previously been examined. This growing body of literature indicates that persons living with dementia are vulnerable to gaps in ophthalmic care. While our study was not equipped to identify the specific reasons why patients with dementia were less likely to be seen by an ophthalmologist, potential reasons include difficulty making, remembering, and arranging transportation to appointments, especially if caregiver support is limited. Hospitalizations and other comorbidities may also limit the continuity of outpatient ophthalmic care. Visual impairment is an important determinant of quality of life and is associated with progression and neuropsychiatric complications of dementia [5,6,7], so it is possible that promoting eye care utilization in this population may improve functional outcomes. However, while there is preliminary evidence that regular eye care encounters and cataract surgery are associated with a lower incidence of dementia and functional decline [16, 17], the effects of eye care utilization on outcomes among those already living with dementia have not been examined.

Of note, not all underutilization of ophthalmic care is necessarily avoidable or harmful. We found that POAG patients with dementia were 34% less likely to have a visual field test than those without dementia, but because of the cognitive demands associated with maintaining fixation and responding quickly and accurately, visual field tests may not be possible or produce reliable results in persons living with dementia [18]. If cognitive concerns were the primary reason that visual field tests were underutilized, we would have expected greater utilization of OCT, which is more rapid and relies less heavily on prolonged fixation and patient participation. However, interpreting retinal nerve fiber layer and ganglion cell layer thinning in the setting of comorbid glaucoma and dementia is challenging, as these changes are known to occur in neurodegenerative diseases such as Alzheimer’s disease independent of glaucoma [19, 20]. Most patients with POAG are asymptomatic in their early stages, and treatment is aimed at reducing the progression of visual field constriction over years or decades. For people with limited life expectancy due to advanced dementia, these treatments may not yield meaningful benefits if death is expected to occur before clinically significant visual impairment due to untreated glaucoma develops. However, such decisions would require a highly personalized assessment of life expectancy and glaucoma progression, and the clinical and health-economic implications of the de-escalation of ophthalmic care at the end of life have not been examined. In the case of nAMD, anti-VEGF injections have the potential to result in more immediate improvements in visual acuity, but these benefits must be weighed against the tradeoffs in quality of life associated with frequent ophthalmology visits and injection-related discomfort and logistical barriers such as ability to consent and availability of legal guardianship.

The results of this study must be considered within the limitations of its design. Due to the deidentified nature of administrative claims data, we lacked information about ocular disease severity such as visual acuity and OCT results, which are important determinants of visit frequency and treatment. The identification of ophthalmic diagnoses and healthcare utilization within administrative claims data is contingent upon insurance billing and reimbursement, so if patients paid for visits, procedures, or medications out of pocket, these would theoretically not be captured. However, this is very unlikely because the outcomes we examined are covered by insurance, and therefore unlikely for patients to ignore their insurance benefit to pay more out of pocket for these services. Another limitation is the reliance on ICD codes to identify dementia, as dementia is frequently underdiagnosed in clinical practice [21]. To counter this, we tested a more expansive definition of dementia, which did not alter the results of the study. Moreover, if the control population contained persons living with dementia who were misclassified as controls, we would expect this to bias results toward the null, suggesting that our results are an underestimate of the impact of dementia on eye care utilization for these disease states. Older adults in this dataset have private health insurance or are primarily covered through Medicare Advantage plans, and while we would expect similar findings in the Medicare fee-for-service (FFS) population, this is difficult to confirm due to known demographic and socioeconomic differences between FFS and Medicare Advantage enrollees. Finally, type 1 error is always a possibility, especially in the context of multiple hypothesis testing.

We found that being diagnosed with dementia leads to lower utilization of eye care in the year following diagnosis. Although the rate of eye care visits was lower, the largest disparities in care were related to the ancillary testing utilization and treatment. Future studies should examine whether eye care underutilization is associated with worse visual, cognitive, and functional outcomes in people living with dementia. If this is the case, then efforts to promote access to eye care may have a positive public health impact in this population.

Summary of inclusion and exclusion criteria