The association between vision impairment and incidence of dementia and cognitive impairment: a systematic review and meta-analysis

Author: Shang X, Zhu Z, Wang W, Ha J, He M.

Geographical coverage: North America, Asia, Europe, and Australia

Sector: Burden of disease

Sub-sector: Epidemiology

Equity focus: None

Study population: Adults with dementia and adults with vision impairment.

Review type: Other review

Quantitative synthesis method: Meta-analysis

Qualitative synthesis method: Not applicable

Background: There’s a potential window to delay the onset of clinical dementia, making it crucial to identify modifiable risk factors for dementia prevention in an aging global population. Sensory impairment, especially midlife hearing loss, is a significant dementia risk factor. Vision impairment, much of which is preventable or treatable, is also important, as it’s linked to accelerated cognitive decline and conditions like age-related macular degeneration and glaucoma, which are associated with Alzheimer’s disease. While the link between vision impairment and dementia has been explored in numerous studies since 2017, the results are inconsistent. No systematic reviews or meta-analyses on this topic have been published yet.

Objectives: To determine whether vision impairment was a predictor for the development of dementia and cognitive decline – and to determine the magnitude of this association and the global burden of dementia associated with vision impairment.

Main findings:

Overall, the pooled RR associated with vision impairment was 1.47 (95% confidence interval [CI], 1.36e1.60), while the pooled RR associated with cognitive impairment was 1.35 (95% CI, 1.28e1.41).

A total of 38 prospective cohort studies (36 plus two additional studies retrieved through the citations or references) reported the association of vision impairment with incident dementia or cognitive impairment. These 38 articles, with the addition of the unpublished results of the China Health and Retirement Longitudinal Study, HRS, Mexican Health and Ageing Study, and the UK Biobank, were included in the systematic review.

Seven cohort studies reporting only the association between vision impairment and cognitive decline and 11 investigating the association between causes of vision impairment (age-related macular degeneration, cataract, glaucoma and diabetic retinopathy) and dementia or Alzheimer’s disease were excluded from the meta-analysis. Two studies were excluded from the meta-analysis because the quality score was 5 or less. Overall, 22 cohort studies were included in the meta-analysis. Seven studies were conducted in North America, three were conducted in Asia, three were conducted in Europe, and one was conducted in Australia. The mean duration of follow-up for the analysis of dementia ranged from 3.7 to 14.5 years. Nine of 14 studies showed a high overall risk of bias (Newcastle-Ottawa quality score). The most common limitations in the studies were loss to follow-up, ascertainment of exposure, and short duration of follow-up.

For the meta-analysis of the association between vision and cognitive impairment, data for 45,313 individuals (57.8% women) from 12 studies were available, among which 13,350 cases of cognitive impairment were newly identified. The number of participants in each study ranged from 625 to 19,618. Ten studies were conducted of both women and men and two studies were conducted of women only. Eight studies were conducted in North America, two studies were conducted in Australia, one study was conducted in Asia, and one study was conducted in Europe. The mean duration for the analysis of cognitive impairment ranged from 3 to 18 years. Nine of 12 studies showed high overall risk of bias with Newcastle-Ottawa quality scores of 7 or less. The most common limitations in the studies were loss to follow-up, ascertainment of exposure, and non-representativeness.

In the meta-analysis of 14 prospective cohort studies with 6,204,827 participants and 171,888 dementia patients, the pooled RR associated with vision impairment was 1.47 (95% confidence interval [CI], 1.36e1.60). In the meta-analysis of 12 prospective cohort studies with 45,313 participants and 13,350 patients with cognitive impairment, the pooled RR was 1.35 (95% CI, 1.28e1.41). Stratified analyses conducted by the authors showed that the associations of vision impairment with incident dementia and cognitive impairment were similar across methods of vision assessment, length of follow-up and study quality. The global number of people with dementia associated with moderate or severe vision impairment in 2016 was 2.1 million (80% uncertainty interval, 1.0e3.3 million), which accounted for 4.7% (95% CI, 2.3%e7.5%) of the global burden of dementia.

Overall, authors note that screening and treating vision impairment, especially in low and middle income countries, may help to alleviate the global burden of dementia. However, further research is needed that addresses the gaps in the evidence noted under the limitations above.

Methodology:

Inclusion criteria consisted of all prospective cohort studies reporting the association between vision impairment and dementia or cognitive impairment were included in the systematic review. The criteria for inclusion of studies in the meta-analysis included prospective design; quality score of more than 5; end points of dementia, Alzheimer’s disease, or cognitive impairment; presentation of quantitative point estimates (HRs, relative risks [RRs], or odds ratios) and variance of the estimates of the association between vision impairment and outcomes; and description of adjustment for potential confounders.

Two authors conducted a comprehensive literature search on PubMed, EMBASE, and Web of Science on September 15, 2020. They also reviewed references and citations of the included studies on Google Scholar to find other relevant studies. The search was limited to adult subjects but had no language restrictions. They also identified seven unpublished large-scale cohort studies with relevant data. They obtained access to unpublished data from four studies: the China Health and Retirement Longitudinal Study, the Health and Retirement Study (HRS), the Mexican Health and Ageing Study, and the UK Biobank. Two reviewers independently screened the titles and abstracts to identify articles relevant to vision impairment and dementia or cognitive impairment.

Two reviewers independently extracted data into a customised database using a predefined data extraction sheet. Study quality was assessed independently by two members of the research team using the Newcastle-Ottawa Scale. The overall certainty of the body evidence was evaluated using the Grading of Recommendations Assessment, Development and Evaluation approach.

Pooled estimates across studies were obtained using a random-effects meta-analysis after log-transforming the study-specific estimates. Study weights were calculated from the inverse variance of the log Relative Risk (RR). Pooled RR was used for studies that reported RRs of dementia only for subgroups. The most fully adjusted models were used to estimate the pooled RR. Random-effects models were used to estimate pooled RRs for the association of vision impairment with incident dementia and cognitive impairment. Secondary analyses examined the association across subgroups of important variables. Stratified analyses were performed for variables like follow-up length, vision assessment methods, geographical region, and study quality. The I² statistic was used to test between-study heterogeneity. Methods like Egger’s test, funnel plots, trim and fill, and removing one study at a time were used to investigate publication bias and conduct sensitivity analysis.

The Population-Attributable Risk (PAR) of dementia linked to vision impairment was calculated for each country using the prevalence of vision impairment data from 2015 and dementia data from 2016, both estimated by the Global Burden of Disease Study. The meta-analysis was conducted using Stata software.

Applicability/external validity: Authors identify a number of differences between included studies which may limit the applicability of their findings (for example, different definitions used of vision impairment or cognitive decline).

Geographic focus: The review includes studies produced in LMICs and involved the production of a global estimate regarding the burden of dementia, based on country-specific estimates.

Summary of quality assessment:

The methods used to identify, include, and critically evaluate studies were quite robust, with an extensive search and two individuals performing all essential tasks. The data analysis approach was also comprehensive, but there was no subgroup analysis based on the quality of the included studies to determine its impact on the results. Given these factors and the overall low quality of evidence in this review, we assign a medium level of confidence to its findings.