Article Text

Original research
Vision impairment and cognitive decline among older adults: a systematic review
  1. Niranjani Nagarajan1,
  2. Lama Assi1,
  3. V Varadaraj1,
  4. Mina Motaghi2,
  5. Yi Sun2,
  6. Elizabeth Couser3,
  7. Joshua R Ehrlich4,5,
  8. Heather Whitson6,
  9. Bonnielin K Swenor1,7
  1. 1Wilmer Eye Institute, Johns Hopkins University School of Medicine, Baltimore, Maryland, USA
  2. 2Johns Hopkins School of Public Health, Baltimore, Maryland, USA
  3. 3Johns Hopkins University School of Medicine, Baltimore, Maryland, USA
  4. 4Ophthalmology and Visual Sciences, University of Michigan, Ann Arbor, Michigan, USA
  5. 5Institute for healthcare policy and innovation, University of Michigan, Ann Arbor, Michigan, USA
  6. 6Department of Medicine, Geriatrics, Duke University, Durham, North Carolina, USA
  7. 7Epidemiology, Johns Hopkins University Bloomberg School of Public Health, Baltimore, Maryland, USA
  1. Correspondence to Professor Bonnielin K Swenor; bswenor{at}


Objectives There has been increasing epidemiological research examining the association between vision impairment (VI) and cognitive impairment and how poor vision may be a modifiable risk factor for cognitive decline. The objective of this systematic review is to synthesise the published literature on the association of VI with cognitive decline, cognitive impairment or dementia, to aid the development of interventions and guide public policies pertaining to the relationship between vision and cognition.

Methods A literature search was performed with Embase, Medline and Cochrane library databases from inception to March 2020, and included abstracts and articles published in peer-reviewed journals in English. Our inclusion criteria included publications that contained subjective/objective measures of vision and cognition, or a diagnosis of VI, cognitive impairment or dementia. Longitudinal or cross-sectional studies with ≥100 participants aged >50 years were included. The search identified 11 805 articles whose abstracts underwent screening by three teams of study authors. Data abstraction and quality assessment using the Effective Public Health Practice Project Quality Assessment Tool were performed by one author (NN). 10% of the articles underwent abstraction and appraisal by a second author (LA/VV), results were compared between both and were in agreement.

Results 110 full-text articles were selected for data extraction, of which 53 were cross-sectional, 43 longitudinal and 14 were case–control studies. The mean age of participants was 73.0 years (range 50–93.1). Ninety-one (83%) of these studies reported that VI was associated with cognitive impairment.

Conclusion Our systematic review indicates that a majority of studies examining the vision–cognition relationship report that VI is associated with more cognitive decline, cognitive impairment or dementia among older adults. This synthesis supports the need for additional research to understand the mechanisms underlying the association between VI and cognitive impairment and to test interventions that mitigate the cognitive consequences of VI.

  • epidemiology
  • delirium & cognitive disorders
  • ophthalmology
  • geriatric medicine

Data availability statement

All data relevant to the study are included in the article or uploaded as online supplemental information.

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See:

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Strengths and limitations of this study

  • There was heterogeneity in the measurement of cognitive and visual function among all included studies.

  • The quality assessment tool used for assessing quality of included studies penalised longitudinal studies that lost over 40% of participants due to drop-outs/withdrawals, which is common in studies that span over many years.

  • Majority of the included studies were cross-sectional, and these are prone to selection bias.


Dementia is among the most pressing public health challenges of the 21st century.1 In 2015, 46.8 million people were living with dementia, and the number is expected to double every 20 years.2 Vision impairment (VI), another major global health problem, affects at least 2.2 billion people worldwide,3 most of whom are aged 50 years and older.4 Both cognitive and VI are projected to affect an increasing number of people over time, primarily due to population ageing.4 5

Prior work has suggested that cognition and vision are associated,6 7 and while there are shared risk factors (neuropathological/vascular),8 there is also longitudinal evidence that VI is associated with cognitive changes.9 The mechanisms underlying the vision-cognition relationship have not yet been fully characterised, but it is hypothesised that sensory loss, such as hearing impairment and VI, may lead to increased cognitive load, structural and functional changes in the brain, and decreased emotional and social well-being, all of which could potentially increase the risk of cognitive impairment.9 10 While the role of treating hearing loss in preventing cognitive impairment has been acknowledged, VI has not yet been recognised as a potentially modifiable risk factor for cognitive impairment.1 11

Since the majority of VI is due to correctable conditions, namely refractive error and cataract,12 establishing the existence of an association between vision and cognitive impairment could present an additional opportunity to prevent cognitive impairment and dementia through interventions that optimise vision. In this systematic review of the literature, we examined the association between cognitive and vision impairment among older adults in existing observational studies. This qualitative review summarises the existing research examining the vision–cognition relationship, providing insight on data gaps and areas for continued investigation, as well as highlighting differences in methodological approaches that may impact the interpretation of results across studies.

Methods/literature search

We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (online esupplemental file 1).

Cross-sectional and longitudinal studies reporting a measure of association between visual function and cognitive impairment were included if they had ≥100 participants aged of ≥50 years (mean) at baseline. Reasons for exclusion of studies were: (1) Outcome measure was not vision or cognition, (2) Association between vision and cognition was not explored, (3) Sample size <100, (4) Publication not in English, (5) Mean age <50 years, (6) No cognitive measure, (7) No vision measure and (8) Outcome was not part of inclusion criteria.

An academic librarian searched: Ovid Medline, Embase, Cochrane and PubMed from their inception to March 2020, and developed a search strategy that combined controlled vocabulary and keywords related to geriatrics, cognition and vision (online esupplemental file 2-complete Ovid Medline search strategy). Searches were limited to human studies published in English. Conference and poster abstracts, and short oral presentations were also included.

Search results were exported to Covidence (Veritas Health Innovation, Melbourne, Australia). Three teams of two reviewers each worked independently and in duplicate to screen titles, abstracts and full-text articles to determine inclusion (NN and MM; EC and YS; VV and LA). Disagreements were adjudicated by a member of the other study team.

Data were extracted from the included publications by one author (NN), and another (LA) extracted data from a random sample of 10% to compare results. Any discrepancies were adjudicated by a third author (VV). Data collected for each publication included: study design, participant characteristics, vision and cognition assessment methods and the summary measure that described the vision–cognition association.

The methodological quality of included studies was assessed by one author (NN) using the Effective Public Health Practice Project Quality Assessment Tool (EPHPP)13 and the global quality ratings and findings were summarised qualitatively. A random 10% sample was reviewed by another author (VV) to ascertain consistency in quality assessment (QA).


Study Selection

Online esupplemental file 3 is a PRISMA flow chart that describes the results of the search strategy of articles that examined the association between VI and cognitive impairment or decline. Of the 11 805 studies that were imported for screening, 110 articles were included in our final systematic review.7 14–122

Description of included studies

Table 1 describes the characteristics of the studies included. The total number of participants in this review was 9 799 329 (range: 112–7 210 535 per study), with a mean age of 73.0 years, (range: 50.0–93.1). Of the total 110 studies included, 53 were cross-sectional, 43 were longitudinal and 14 had a case–control study design. The range of follow-up time for the longitudinal studies was 2 months to 10 years.

Table 1

Patient demographics and study characteristics

Of the 110 studies included, 51 reported findings from participants enrolled in population-based studies. There were five studies each from the following large population based longitudinal studies: English Longitudinal Study of Aging (ELSA) and The Three-City Study. Three studies each from Fujiwara-Kyo Study, Salisbury Eye Evaluation Study, Irish Longitudinal Study on Aging and Singapore Epidemiology of Eye Diseases. Two studies each came from The Newcastle 85+ study, Study of Osteoporotic Fractures, Blue Mountain Eye Study, Australian Longitudinal Study on Aging, Health and Retirement Study, National Health and Aging Trends Study (NHATS), Leiden 85+ study, Health ABC study and the Singapore Malay Eye Study. Additionally, 10 studies used insurance claims data from different countries.

The studies in this review included participants from over 17 different countries (table 1), 30 studies (27%) from the USA, followed by 25 studies from Europe (23%) including the UK, Germany, Ireland, Finland, Switzerland, France and Netherlands. Ninety of the studies were published between 2009 and 2020. All papers provided a description of sampling methods. 16 studies were included which were either conference abstracts or short oral presentations.24 32 45 49–51 53 56 92 93 96 112–116

Assessment of cognitive function

To assess cognitive function, 89 studies used objective assessments, 13 used other assessment methods such as self-report and diagnosis codes, and 7 studies used a combination of both (table 2); one study did not provide information about cognitive function assessment.92 Mini-Mental State Examination (MMSE) was the most commonly used objective method to assess cognitive function (42 studies). Other methods used to objectively measure cognition included: Montreal-Cognitive Assessment test,28 32 34 53 93 Addenbrooke’s Cognitive Examination-Revised,45 61 79 Cognitive Performance Scale,42 49 65 Blessed-Orientation-Memory-Concentration test,25 62 Abbreviated Mental Test,72 74 89 100 110 119 Blessed Dementia Scale,86 Digit Symbol Substitution Test 22 122 and Cambridge Cognitive Examination test.88 For the studies that used other assessment methods, four used self-reported cognitive measures,18 22 39 105 and nine used diagnostic codes to define cognitive decline and/or dementia.19–21 51 69 99 103 115 117 Among the studies that used a combination of objective and subjective methods, four used self-reported cognitive function along with an objective measure.22 30 48 55

Table 2

Measures of vision and cognition assessed in studies

Assessment of visual function

In order to assess visual function, 66 studies used objective assessments, 34 used other assessment methods such as self-report and diagnosis codes, and 8 studies used a combination of both (table 2); no information was available from two studies.24 116 Visual acuity (VA) was the most commonly measured visual function (42 studies), of which the Snellen acuity chart was the most commonly used method (18 studies). VA was also measured in combination with other visual functions, including: visual fields (VF) (six studies),20 51 70 90 102 118 contrast sensitivity (CS) (eight studies),29 52 86 93 94 102 118 122 macular pigment optical density (two studies)14 34 and fundus photography (two studies).63 118 Other methods used to objectively measure visual functions included: colour vision,17 VF only,28 CS only,35 76 fundus photo with grading38 100 and autorefraction.74 Other assessment methods included: self-reported vision (24 studies) and diagnostic codes or patient records to define VI (10 studies).19 21 46 69 83 99 104 115 117 121 The studies that used a combination of methods, eight studies used self-report along with an objective measure of visual function.22 40 53 59 68 71 107 113

Quality of studies

The methodological quality of included studies was assessed using the EPHPP.13 The tool assessed each study on five domains: (1) Selection bias, (2) Study design, (3) Confounders, (4) Data collection methods and (5) Analysis. For each included study the five relevant domains were ranked on a three-point Likert scale with three representing a low risk of bias (‘strong’), two a possible risk of bias (‘moderate’) and one a high risk of bias (‘weak’). An overall rating was derived following the EPHPP methodology. A study consisting of at least one ‘weak’ rating in a domain received an overall rating of ‘moderate,’ while those with two or more domains with ‘weak’ ratings were automatically classified as ‘weak’ overall. We present our studies in three different tables which is categorised based on the overall ratings, with ‘strong’, ‘moderate’ and ‘weak’ studies in tables 3–5, respectively. In our sample, 17 studies received a rating of strong, 70 moderate and 23 weak.

Table 3

Studies with a ‘strong’ rating

Table 4

Studies with a ‘moderate’ rating

Table 5

Studies with a ‘weak’ rating

Study findings

Of the 110 studies included (tables 3–5), 91 found a significant positive association between VI and cognitive decline, cognitive impairment or dementia, and 13 studies found no significant association.26 30 43 44 60 61 68 81 82 90 94 115 117 There were six studies that were inconclusive.23 37 67 70 79 83 Of the 91 studies that found a significant association, 77 used objective methods to assess their vision or cognitive outcome. Of the 43 longitudinal studies, 35 found a significant association between VI and cognitive decline, cognitive impairment or dementia. The most commonly presented statistical measures were ORs and HRs. The random 10% of the study sample that was separately extracted by an independent author (LA) was found to be similar to elements from the primary extraction.


In this systematic review, we evaluated and synthesised the literature examining the association between VI and cognitive function among older adults, and found strong agreement that VI is associated with cognitive impairment, cognitive decline or dementia. Results from the longitudinal studies that found a positive association between vision and cognition supports our hypothesis that VI may be a risk factor for cognitive impairment, cognitive decline or dementia.

Ninety-one studies reported associations between decline in visual and cognitive functions. Garin et al,40 who received a ‘moderate’ rating in the QA, performed a cross-sectional analysis in a representative sample of Spanish population and measured cognition objectively. They also measured distance and near vision and found that objective and subjectively measured poor distance and near VA were associated with worse cognitive functioning. Lin et al47 used data from a large longitudinal cohort study of older women and found that VI was associated with greater odds of cognitive and functional decline over 2 years. This study used objective measures of assessment for both vision and cognition and received a ‘strong’ rating in the bias assessment. Luo et al,48 who received a ‘moderate’ rating in QA, performed a cross-sectional analysis on a large population sample from China. They reported that those with VI and Dual Sensory Impairment (DSI) were more likely to have severe to extremely severe dementia compared with those without any sensory impairment. Another longitudinal study that received a ‘moderate’ rating in QA from Germany by Hajek et al58 with a large sample size (n=2394) showed that the onset of severe VI was associated with a decline in cognitive function scores. Uhlmann et al64 in their paired case–control study between VI and dementia patients concluded that VI was associated with both an increased risk and an increased clinical severity of AD. Although Frost et al38 found a strong association between early age-related macular degeneration and AD, their study was cross-sectional, and the sample size was too low to derive an inference. Both these studies received ‘moderate’ rating in QA. Davies-Kershaw et al95 in their longitudinal analysis using the ELSA wave 2 and wave 7 data found that individuals in the younger group (50–69 years) and with moderate and poor self-rated vision were at greater risk of developing dementia than those with normal self-rated vision. Hamedani et al99 used Medicare claims data from 2014 consisting of 472 871 participants and concluded that blindness/low vision was associated with a greater odd of Alzheimer’s disease and all-cause dementia. Both these studies also received ‘moderate’ rating in QA. Soto-Perez-de-Celis et al62 in their cross-sectional study case–control study found that DSI was significantly associated with possible CI. However, the study received an overall ‘weak’ rating in QA.

Of the 91 studies that found an association between VI and cognitive function, 35 were longitudinal, 46 were cross-sectional and 10 were case–control studies. Of the 13 studies that found no association between VI and cognitive function, 6 were longitudinal, 5 were cross-sectional and 2 were case–control studies. Ihle et al60 performed a cross-sectional analysis using a sample of 2812 participants from Switzerland. They objectively measured cognition and vision and concluded that their data did not support an increased relation of cognitive and sensory abilities in old age. This study received a ‘weak’ rating in QA. Hong et al82 used data from Blue Mountain Eye Study, a longitudinal study from Australia that studied associations between VI and a decline in MMSE scores over a duration of 10 years. The study concluded that VI was not associated with cognitive decline over 5 years or 10 years. Although the study included a large number of participants overall (n=2334), only 152 individuals with VI were included in this analysis, suggesting that there may have been survival bias. Brenowitz et al94 in their longitudinal study using the Health ABC data concluded that VA and CS independently were not significantly associated with incident dementia. However, Swenor et al122 used data from the same study and found that impaired VA, CS and stereo acuity had a greater risk of incident cognitive impairment. This could be due to the different outcome measures assessed, that is, dementia94 vs cognitive impairment.122 These three studies received a ‘moderate’ rating in QA. Michalowsky et al,117 who received a ‘strong’ rating in their case–control study concluded that VI was not significantly associated with dementia, a combination of both visual and hearing impairments was significantly associated with the risk of dementia.

There was considerable heterogeneity in the measurement and reporting of cognitive function. Studies measured cognitive function using a variety of instruments with the most common being MMSE. The MMSE is a paper‐based test with a maximum score of 30, with lower scores indicating more severe cognitive impairment. A score of 24 is often used as a threshold to define ‘normal’ cognitive function.123 The MMSE has been found to be a valid and reliable tool as assesses by many studies.123 124 Several studies used self-report, diagnosis codes and data from existing records to define cognitive status. Similarly, visual function was also assessed by various methods including self-report. While VA was assessed most commonly, there was significant variation in the charts and tools used to assess it. The parameters used to define cognitive decline and VI may have impacted results across and within studies.

Our systematic review has found that there is a strong consensus in the literature that VI is associated with cognitive decline, cognitive impairment or dementia. Two hypotheses may help explain this association. The first one is that a common pathological process (eg, vascular disease) might be responsible for both the sensory and cognitive impairment in older adults. The second one is that by increasing cognitive load, sensory impairments such as VI might cause cognitive impairment.125 Literature also suggests that vision rehabilitation in the form of cataract surgery slows the rate of cognitive decline, and therefore, early vision interventions could potentially reduce risk of dementia.126

Our review evaluated bias for all of the 110 included studies. The majority of studies included in our review were cross-sectional, and according to EPHPP guidelines, cross-sectional studies can only receive a low or moderate rating in the bias assessment. Cross-sectional studies are also prone to selection bias, thus yielding estimates that may not reflect true associations in the target population. Studies receiving a strong rating were all longitudinal. However, the tool penalises longitudinal studies that lose >40% of participants due to dropouts/withdrawals. This may, perhaps unfairly, affect longer longitudinal studies to a greater extent since they collect data over many years and can have more drop-outs due to deaths since they are conducted among older adults.

This review has several important implications. First, it highlights the need for standardised methods to assess and define both visual and cognitive function that will aid future research on these emerging public health issues. Second, it brings into focus the consistent association of VI with cognitive impairment in older adults and the need to better understand the mechanisms underlying this relationship. Third, as the longitudinal results support the sensory consequence theory, and suggest that VI may be a risk factor for cognitive decline, this points to a need for formulating preventive measures and vision rehabilitation models, such as prescription glasses, cataract surgery, low vision rehabilitation, etc, that could have the potential to improve overall health and well-being of older adults.


Given the large number of studies included in this review and the heterogeneity of measures used to assess the outcome, it was not possible to compare and meta-analyse results across studies. Although 35 longitudinal studies found a positive association between VI and cognitive decline, we cannot establish temporality between this relationship due to the heterogeneous nature of the studies. The studies included diverse populations, with different disease processes, and variation in definitions of both cognitive and VI. There is also potential bias associated with studies that used different protocols for cognitive and sensory measurements. The MMSE, which was the most commonly used assessment method for testing cognition is sensory dependent and therefore one can argue that the results may be confounded with VI.127 Further studies should examine the impact of using vision independent cognitive tests on the vision–cognition relationship. Our review examined all cause VI and dementia, and further study is needed to examine the vision–cognition relationship by dementia subtype and by different vision pathology. However, despite the heterogeneity in studies and assessment methods, we synthesised the evidence qualitatively and by taking into account study quality assessed using a validated tool. While our search strategy was robust, it may have been limited by the exclusion of studies that were not published in English.


The number of older adults with VI and dementia is increasing globally, and therefore, the elucidation of the relationship between vision and cognition is of particular public health importance. This systematic review found that the positive association of VI with cognitive decline, cognitive impairment or dementia is largely consistent across studies using different measures of vision and cognition, as well as between countries and cohorts. This overall agreement in the literature suggests that poor visual and cognitive function are associated, and that additional research is now needed to move beyond documenting these associations. The focus of this area of research should now turn to identifying the factors and strategies that mediate the vision–cognition relationship and identifying potential interventions, such as vision rehabilitation models and strategies tailored to people with VI, that may mitigate the cognitive implications of VI.

Data availability statement

All data relevant to the study are included in the article or uploaded as online supplemental information.

Ethics statements

Patient consent for publication


BKS and NN had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.


Supplementary materials


  • Twitter @LamaAssi3

  • Contributors Study concept and design: BKS and HW. Collection, management, analysis and interpretation of the data: BKS, NN, EC, MM and YS. Drafting of manuscript: NN and LA. Preparation, review or approval of the manuscript: NN, VV, LA, BKS, JRE and HW. Decision to submit the manuscript for publication: BKS, JRE and HW.

  • Funding BKS is supported by funding from the National Institute on Ageing (K01AG052640). JRE is supported by the National Eye Institute (K23EY027848).

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.