Visual impairment in aging and cognitive decline: experience in a Memory Clinic

Visual impairment is common in people living with dementia and regular ophthalmological exams may improve their quality of life. We evaluated visual function in a cohort of elderly individuals and analyzed its association with their degree of cognitive impairment. Participants underwent neurological and neuropsychological exams, neuro-ophthalmological assessment (visual acuity, intraocular pressure, rates of past ophthalmological pathologies, use of ocular correction, treatments and surgeries) and optical coherence tomography (OCT) scan. We analyzed differences in ophthalmological characteristics among diagnostic groups. The final sample of 1746 study participants aged ≥ 50 comprised 229 individuals with Subjective Cognitive Decline (SCD), 695 with mild cognitive impairment (MCI) and 833 with Dementia (Alzheimer disease: n = 660; vascular dementia: n = 92, Lewy body dementia: n = 34; frontotemporal dementia: n = 19 and other: n = 28). Age, gender and education were used as covariates. Patients with Dementia, compared to those with SCD and MCI, presented worse visual acuity (p < 0.001), used less visual correction (p = 0.02 and p < 0.001, respectively) and fewer ophthalmological treatments (p = 0.004 and p < 0.001, respectively) and underwent fewer ocular surgeries (p = 0.009 and p < 0.001, respectively). OCT image quality worsened in parallel to cognitive decline (Dementia vs SCD: p = 0.008; Dementia vs MCI: p < 0.001). No group differences in past ophthalmological disorders or abnormal OCT findings were detected. Efforts should be made to ensure dementia patients undergo regular ophthalmological assessments to correct their visual function in order to improve their quality of life.


Methods study subjects.
Fundació ACE is a non-profit organization located in Barcelona, Spain that was declared a welfare institution by the Generalitat of Catalonia in 1995. The present study derives from the Neuro-Ophthalmology Research At Fundació ACE (NORFACE) cohort, which was founded in 2014 to search for retinal biomarkers of AD and examine the relationship between retinal changes and different types of neurodegenerative disorders 29 . Consecutive patients evaluated due to cognitive decline at Fundació ACE between January 2017 and March 2018 were enrolled in the present study. Participants were recruited from three sources: 1) the Memory Clinic, an outpatient diagnostic unit for patients with cognitive decline that has an agreement with the Catalan Public Health System, 2) Fundació ACE's Open House Initiative 30 , a corporate social responsibility program that assesses for free the cognitive status of individuals from the community without the need of a physician's referral, and 3) Fundació ACE Healthy Brain Initiative (FACEHBI) 31 , a research study with the goal of identifying biomarkers of preclinical AD in individuals with Subjective Cognitive Decline (SCD). Inclusion criteria were: age ≥ 50-yo, presence of a consensus-based clinical diagnosis about the participants' cognitive status and ability to complete the full ophthalmological exam and OCT scan (excluding those patients with severe dementia stages, equivalent to a Global Deteriorating Scale [GDS] 32 score > 6).
Clinical diagnostic groups. Study participants completed neurological, neuropsychological and social evaluations at Fundació ACE. For each individual, a consensus-based diagnosis about his/her cognitive status (Dementia vs mild cognitive impairment [MCI] vs subjective cognitive decline [SCD]) was reached at the time of the study recruitment by a multidisciplinary team of professionals that included neurologists, neuropsychologists and social workers 33 . Cognitive assessment consisted of the Spanish version of the Mini-Mental State Examination (MMSE) 34,35 , the memory part of the Spanish version of the 7 Minutes test 36 , the Hachinski Ischemic Scale 37 , the Spanish version of the Neuropsychiatric Inventory Questionnaire (NPI-Q) 38 , the GDS 32 , the Clinical Dementia Rating Score (CDR) 39 , the Blessed Dementia Scale 40 and a comprehensive neuropsychological battery of Fundació ACE (NBACE) 41,42 . Demographical information collected included age, gender and years of education.
Dementia was defined according to the DSM-V criteria 43 . MCI was defined using Petersen's criteria 44 . SCD refers to the perception of memory or other cognitive problems without impairment on standardized cognitive tests 45 . Specifically, individuals with SCD had a CDR 39 46 , and a strictly normal performance in the NBACE 41,42 . Neuro-ophthalmological evaluation. Study participants underwent a complete neuro-ophthalmological evaluation in parallel to the cognitive evaluation (usually the same day of the initial neurological assessment). The whole visit was performed by an optometrist and was completed in 20 mins. In order to undergo the exam, subjects www.nature.com/scientificreports www.nature.com/scientificreports/ had to be able to cooperate, obey simple instructions and sit still for a few minutes. The evaluation comprised: 1) review of past ophthalmological diseases, treatments and surgeries, 2) monocular visual acuity assessment with the participants wearing their habitual correction for refractive error using a pinhole occluder and the Early Treatment of Diabetic Retinopathy Study (ETDRS) chart 47,48 , 3) intraocular pressure (IOP) measurement by Icare tonometry 49 , and 4) swept source (SS) OCT scan. The visual acuity assessment was done the same way for all participants, regardless of their cognitive status and level of cooperation. Reduced visual acuity was defined as a standard LogMAR fraction scale ≤ 20/50 at 20 ft (equivalent to a fraction scale of 6/15 at 6 m and a decimal scale of 0.4) according to the Snellen scale 50 , which is in line with the North American guidelines of visual loss 51 . High IOP was defined as ≥24 mmHg using Icare Tonometry 49 . All assessments were carried out by a single optometrist and reviewed by a single ophthalmologist. Before beginning the study, the ophthalmologist trained the optometrist in the evaluation of OCT images, with the goal to differentiate normal images from abnormal findings. The ophthalmologist reviewed the ophthalmological history, ocular exam findings and OCT images from those cases in which OCT images were considered abnormal by the optometrist and came up with a diagnosis if appropriate. Of note, the diagnoses of dry and wet AMD were based on the Age-Macular Degeneration Preferred Practice Patterns guidelines from the American Academy of Ophthalmology (2015) 52 , which uses the classification of the Age-Related Eye Disease Study 53 and a more recent classification to define the early and intermediate stages of AMD 54 . The glaucoma category was established based on the image of the head of the optic nerve (ONH), ONH OCT findings and Icare IOP measurements. If a participant presented with what is commonly called a "suspicious papilla", an IOP > 24 mmHg, and signs of thinning of the RNFL in the OCT, the case was assigned a label of "glaucoma" based on OCT-findings. No anterior chamber OCT imaging was available. The ophthalmologist and neurologists were blind to each other's diagnosis. statistical analysis. Data processing and analysis were conducted using SPSS statistical software v. 20. Alpha level was set at 0.05. Frequency analysis and measures of central tendency and dispersion were used to describe the study variables. Bootstrap was used to resample the cohort in order to obtain the 95% confidence interval (CI) for each variable. Demographic data differences among diagnostic groups (SCD vs MCI vs Dementia) were assessed with an ANOVA test for quantitative variables (age, years of education and MMSE scores) and a Chi-Square test for qualitative variables (gender). A Multinomial Logistic Regression (MLR) model including age, gender and years of education as co-variates was used to analyze the differences in ophthalmological pathologies (history of glaucoma and AMD, use of corrective lenses, past ocular surgeries and current ophthalmological treatment), visual exam (low visual acuity and high IOP) and OCT scan findings (suboptimal OCT image quality, abnormal OCT findings and newly detected ophthalmologic pathology) among the three diagnostic groups. We performed a similar MLR model within the Dementia cohort (n = 684) to analyze differences in visual acuity across GDS sub-groups, including age, gender and years of education as co-variates. Additionally, we performed a Linear Regression Analysis adjusted by age, gender and years of education within the Dementia cohort to explore the relationship between MMSE scores and visual acuity.

Results
Demographical characteristics of the cohort. 1813 individuals were assessed between January 2017 and March 2018, had a consensus-based clinical diagnosis and underwent ophthalmological exam and OCT scan. Of those, we excluded 22 individuals due to age < 50 and 45 due to incomplete ophthalmological exam and/or OCT scan. Detailed reasons for the exclusions are reported in Supplementary Table S1.
The final study cohort comprised 1746 individuals. Participants were sub-classified into 3 clinical groups according to their cognitive stage: SCD, MCI and Dementia. Due to the low number of cases observed in a few variables and in order to be able to run the appropriate statistical analysis, we simplified the dataset and dichotomized the subcategories within each ophthalmological variable (present/absent or normal/abnormal). A description of the study sample used for statistical analysis is shown in Tables 1-3, while detailed information is shown  in Supplementary Tables 2-4.
Most of the study participants (n = 1583, 90.7%) were currently using corrective lenses (218 (95.2%) with SCD, 643 (94%) with MCI and 722 (86.7%) with Dementia). After adjusting for age, gender and years of education, data showed that patients with Dementia were less likely to be currently using lenses compared to individuals with SCD (OR = 0.37, 95% CI = 0.17-0.84, p = 0.02) and MCI (OR = 0.46, 95% CI = 0.30-0. 70 Ophthalmological exam findings. Data from the participants' ophthalmological exam are detailed in Tables 3 and 5 and Supplementary Table S3. 453 participants (25.9%) suffered from low visual acuity as assessed by the ETDRS chart (14 (6.1%) with SCD, 131 (19.2%) with MCI and 308 (37%) with Dementia). After adjusting for age, gender and years of education, data showed that patients with Dementia were almost 3.5-times more likely to present with reduced visual acuity compared to those in the SCD group (OR = 3.36, 95% CI = 1.77-6.39, p < 0.001) and over 1.5 times compared to those in the MCI group (OR = 1.61, 95% CI = 1.77-6.39, p < 0.001). Additionally, individuals with SCD had half the odds of presenting with reduced visual acuity than MCI patients (OR = 0.48, 95% CI = 0.25-0.90, p = 0.023). We analyzed differences in visual acuity within the Dementia cohort according to patients' GDS status (GDS = 4: mild dementia, GDS = 5: moderate dementia and GDS = 6: moderate-severe dementia) 32 (Supplementary  Table S4). After adjusting for age, gender and years of education, data showed that patients with mild Dementia  www.nature.com/scientificreports www.nature.com/scientificreports/ had 30% lower odds of presenting with reduced visual acuity compared to those with moderate Dementia (OR = 0.68; 95% CI = 0.50-0.93, p = 0.01). Patients with moderate-severe Dementia did not show significant differences in visual acuity compared to those with mild and moderate Dementia. Additionally, we explored the relationship within MMSE 34 scores and visual acuity within the Dementia group adjusted by age, gender and years of education. Visual acuity had a statistical marginal effect on MMSE scores (p = 0.05), such as low visual acuity was associated to lower MMSE scores (data not shown).
Icare tonometry 49 measurements showed that 110 participants (6.3%) had elevated IOP values (12 (5.2%) with SCD, 44 (6.4%) with MCI and 54 (6.5%) with Dementia). No statistical differences in terms of IOP were detected among diagnostic groups after adjusting for age, gender and years of education.
In 173 participants (9.9%) OCT scan images had suboptimal quality (1 (0.4%) with SCD, 42 (6.1%) with MCI and 130 (15.6%) with Dementia). After adjusting for age, gender and years of education, the analysis demonstrated that patients with Dementia were at least 2-times more likely to show worse OCT image quality compared to SCD participants (OR = 14.95, 95% CI = 2.01-111.03, p < 0.008) and at least 1.4-times more likely compared to those with MCI (OR = 2.01, 95% CI = 1.36-2.98, p < 0.001). OCT image quality differences between SCD and MCI groups were borderline significant, with patients in the MCI group showing worse image quality than their peers with SCD (OR = 0.14, 95% CI = 0.02-1.00, p = 0.05).
In 315 participants (18%) OCT images were considered to be pathological (17 (7.4%) with SCD, 123 (18%) with MCI and 175 (21%) with Dementia) but no significant differences were detected among groups after adjusting for age, gender and years of education. The most frequent OCT-based diagnosis was posterior pole deformation due to posterior staphyloma/myopia magna (n = 70) followed by epiretinal membrane (n = 63), image  Table 4. Diagnostic group differences in past ophthalmological pathologies and treatments. A Multinomial Logistic Regression model including age, gender and years of education as co-variates was used to analyse group differences in past ophthalmological pathologies and treatments. For each comparison, the first listed group (Dementia, MCI or SCD) acted as reference. Statistical significance was set-up at p < 0.05. AMD = agerelated macular degeneration; CI = confidence interval; MCI = mild cognitive impairment; OR = odds ratio; SCD = subjective cognitive decline.
After completing the study's neuro-ophthalmological exam and OCT scan, a new ophthalmological pathology was detected in 213 participants (12.2%) (15 (6.6%) with SCD, 73 (10.7%) with MCI and 125 (15%) with Dementia). No statistical differences were detected among groups after adjusting for age, gender and years of education. The types of new ophthalmological pathologies identified were as follows: posterior pole deformation due to posterior staphyloma/myopia magna (n = 56), epiretinal membrane (n = 42), retinal pigmented epithelium abnormality (n = 27), dry AMC (n = 23), other maculopathies (n = 13), image suggestive of glaucomatous pathology (n = 12) and wet AMD (n = 10). Those participants in whom a new pathology was detected were recommended to consult with an ophthalmologist for a full evaluation.

Discussion
In the present study we analyzed visual function, ophthalmologic pathologies and care received and OCT scan findings in a cohort of 1746 individuals evaluated at a Memory Clinic. We detected that patients living with Dementia presented with worse visual acuity, were less likely to receive adequate visual health care and exhibited suboptimal OCT scan image quality compared to those with MCI and SCD.
First, we examined the prevalence of common age-related ocular pathologies in the three study groups. After adjusting for age, gender and years of education, we did not detect a higher prevalence of open-angle glaucoma or AMD in patients with Dementia compared to those with SCD and MCI. These two ocular pathologies are leading causes of visual loss in older adults of developed countries and share with AD common risk factors, a chronic decline of function and histopathologic features, although their relationship with dementia is not fully understood [59][60][61][62] . Several studies have detected a two-to-three fold increase of glaucoma prevalence in patients with AD 59,63 but there are conflicting results about whether patients with glaucoma have a higher risk of developing AD 61,64 . Most of the studies that analyze the relationship of retinal thickness changes assessed by OCT and cognitive decline consistently exclude cases with a previous diagnosis of open-angle glaucoma or AMD, as retinal thinning has been reported in these entities and could interfere with the measurement of that associated with AD neurodegeneration itself. It is important to note that in our cohort, as opposed to other studies, the Dementia group was not only comprised of AD patients (79.23%) but it also included patients with a diagnosis of Vascular Dementia, DLB or FTD, among others. The relation of non-AD dementia types with glaucoma is relatively understudied compared to AD and there is little data about glaucoma prevalence among those disorders. Also, our control group consisted of individuals with SCD who consulted for cognitive problems at the Memory Clinic or Open House Initiative program, instead of healthy subjects from the general population. Although individuals with SCD show unimpaired performance on formal neuropsychological testing, there is increasing evidence that SCD may represent the first symptomatic manifestation of AD 45 .  Table 5. Diagnostic group differences in ophthalmological exam and OCT findings. A Multinomial Logistic Regression model including age, gender and years of education as co-variates was used to analyse group differences in ophthalmological exam and OCT scan findings. For each comparison, the first listed group acted as the reference one. Statistical significance was set-up at p < 0.05. CI = confidence interval; MCI = mild cognitive impairment; OCT = optical coherence tomography; OR = odds ratio; SCD = subjective cognitive decline.
www.nature.com/scientificreports www.nature.com/scientificreports/ Further analysis uncovered that patients with Dementia were less likely to receive ophthalmologic health care than their peers with SCD and MCI: despite being significantly older, they were less likely to use corrective lenses, received fewer ophthalmological treatments and had lower rates of ocular surgery. Of note, unadjusted data suggested that rates of surgeries increased parallel to the degree of cognitive impairment (SCD: 25.8% vs MCI: 40.4% vs Dementia: 47.2%), but after accounting for age, gender and years of education, the analysis revealed that Dementia patients had in fact undergone fewer surgeries compared to the MCI and SCD groups. Cataract extraction was the most frequent ocular surgery performed in our sample (>89%) and it is known that it can often correct refractive errors. We acknowledge that this fact could have some impact in our data about the use of corrective lenses. Overall, these data suggest that despite the high prevalence of visual impairment in elderly patients with dementia, it is diagnosed or treated less often than in people without dementia 65 . As health care is universal and free of cost in Spain, this fact mainly excludes economic reasons for the differences detected among groups. One possible explanation could be the communication difficulties that patients with dementia experience as their cognition worsens, making them less able to report or ask support for their visual deficits. Regarding this issue, in Spain there are no specific guidelines for the ophthalmological evaluation and treatment of people with dementia. Studies show that most patients with dementia residing in long-term care facilities do not regularly use their prescription glasses, or those are no longer accurate 28 . In general, as dementia progresses, caregivers tend to adopt a non-invasive attitude about patients' healthcare, to avoid surgical procedures and to seek less medical consultations not directly related to dementia, especially in those patients with more advanced stages of the disease and baseline psychotic symptoms 66 . In specific comorbidities, such as sensory deficits, this is known to be counter-productive, as the association of visual impairment and dementia in elderly population leads to poor quality of life, anxiety 67 , isolation 68 and increased mortality rate 69 . Lastly, the phenomenon of "diagnostic overshadowing" could be affecting the visual care received by patients with dementia 70 . This term refers to the tendency of clinicians to overlook and underdiagnose medical co-morbidities in patients with mental retardation, psychiatric illnesses and dementia.
In the ophthalmological exam performed to the study participants, we detected that almost 40% of Dementia patients presented low visual acuity despite wearing their habitual correction for refractive errors. Visual acuity worsened in parallel to cognitive decline, as patients with Dementia were 3.4 and 1.6 times more likely to present worse visual acuity than those with SCD and MCI, respectively. Interestingly, this was not associated with previously diagnosed ocular pathologies or abnormal OCT findings. In our cohort, patients with Dementia were less likely to wear corrective lenses, to take ophthalmologic treatments and to undergo ocular surgery, as discussed above, and this could be influencing their visual acuity. Besides, complex visual deficits associated to parietal and occipital cortices damage that occur in AD 20,22 , such as impairment of reading abilities, could interfere with their performance in the ETDRS chart. Several studies have detected an inverse association between cognitive decline and visual acuity [71][72][73] . Recently, the Salisbury Eye Evaluation Study reported that visual impairment measured by the ETDRS chart was inversely associated with cognitive function quantified by MMSE scores both cross-sectionally and longitudinally in a large population-based sample of older US adults without dementia at baseline 73 . The Prevalence of Visual Impairment in Dementia (PrOVIDe) study is a large cross-sectional research project in the United Kingdom among 700 patients with dementia aged 60-89 years that reported a prevalence of visual impairment of 16.3% (as defined by standard logMAR < 6/18) or 32.5% (logMAR < 6/12) while participants were using their habitual visual correction. 45% of cases of visual impairment were correctable with new glasses, and almost 50% of the remaining cases were associated with cataracts (and thus potentially remediable). Moreover, undercorrected/uncorrected visual impairment was higher in residents of nursing homes compared to those living in the community. 22% of participants reported not having had a sight test in the previous two years (including 19 participants that had not been tested in the last ten years). 16% of participants could not read standard newspaper-size print with their current visual correction. The study, though, found no evidence of differences in management of visual impairment in people with dementia compared to elderly population in general 4 .
OCT image quality was also inversely related to the degree of cognitive decline. These data are in agreement with our group's recent study 29 which quantified peripapillary RNFL thickness in 930 individuals including cognitively healthy, amnestic-type MCI 74 and AD, and showed that OCT image quality was the most influential factor determining RNFL variability, while there were no significant differences in RNFL thickness among clinical groups. Although OCT scan is considered a non-invasive and fast test, its completion requires minimum cooperation. Patients with moderate/advanced stages of Dementia are less able to follow directions and remain still, thus affecting the image quality achieved. This result may have relevant implications for OCT studies in aging and dementia, and suggests that data correction for image quality should be implemented in OCT studies.
Finally, we assessed OCT scan pathologic findings and newly detected ophthalmologic pathologies and found that there were no significant differences among groups. Unexpectedly, posterior pole deformation due to posterior staphyloma/myopia magna was the most frequent OCT-based diagnosis detected in our cohort (n = 70). There are several factors that could explain it. First, we did not systematically record myopia magna or alterations in the eyeball shape as a diagnosis in the participant's past ophthalmological history. Secondly, we only recorded the main OCT-based diagnosis, and given the great deformation of the posterior pole that this condition implies, it tends to cancel out other possible diagnoses, such as epiretinal membrane or AMD. Lastly, cases with major posterior pole alterations that do not fall within other more specific categories would also be included in this group. Among the 70 participants in which posterior pole deformation was detected, 37 of them had low visual acuity and 59 were using visual correction. Additionally, it is important to emphasize that a clinical diagnosis of glaucoma would require a face-to-face visit with an ophthalmologist, supported by functional type tests (e.g. visual fields evaluation), examination of the eye fundus by indirect ophthalmoscopy and/or stereoscopic pictures, and often several longitudinal visits to reach a firm diagnosis. Given the design of the study, we could only establish a suspicion of glaucoma, but not determine if it was primary open-angle glaucoma or a secondary glaucoma.
www.nature.com/scientificreports www.nature.com/scientificreports/ We cannot rule out that if we had performed such additional tests, the number of glaucoma diagnoses would have been higher in our cohort.
The main strengths of our study are the large sample size, the consensus-based clinical diagnosis and the double-blind nature of the study. Our study had also several limitations. First of all, we did not perform a thorough ophthalmological testing covering the assessment of eye movements, reading speed or color vision, features that are known to be abnormal among patients with dementia. Diabetes status and review of signs of diabetes retinopathy were also not recorded. Furthermore, the distribution of cognitive stages within our Dementia group does not reflect the reality of the Memory Clinic or the general population, as patients with advanced stages of dementia (GDS 32 > 6) were excluded from the study as they were not able to cooperate in the ophthalmologic exam and OCT scan. Thus, our study consists of a convenience sample instead of a pure observational/ecological one. On the other hand, this makes the study results even stronger, as even though our Dementia group included mostly mild and moderate cases, we were still able to uncover significant differences compared to MCI and SCD groups. Another limitation of the study is that we did not have access to private medical records and in some cases we may have missed relevant clinical information. Additionally, we did not perform a follow-up of those cases in which we detected abnormal OCT scan findings and were recommended to consult with an ophthalmologist.
To conclude, our study showed that in a large cohort of elderly individuals evaluated due to cognitive decline in a Memory Clinic, those diagnosed with Dementia, compared to their peers with SCD and MCI, presented with worse visual function, were less likely to receive visual healthcare and had suboptimal OCT image quality, independently of their age, gender and years of education. Both visual impairment and cognitive decline are major health issues in older adults that have significant adverse effects in their quality of life and threaten the sustainability of public health care systems. Thus, efforts should be made to guarantee regular ophthalmological assessments and optimize visual function in elderly patients living with dementia in order to improve their quality of life, avoid isolation and reduce their caregivers' burden.

Data Availability
The datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.