Association Between Visual Impairment and Depression in Patients Attending Eye Clinics: A Meta-analysis

Key Points

Question  What is the prevalence of depression in patients attending eye clinics?

Findings  In this meta-analysis, depression was found to be common (prevalence of 25%) in patients with visual impairment and those aged older than 65 years.

Meaning  These results suggest that depression in patients with visual impairment is a relatively common health problem in the clinic and in patients with cognitive impairment.

Importance  Given that depression is treatable and some ocular diseases that cause visual loss are reversible, early identification and treatment of patients with visual impairment who are most at risk of depression may have an important influence on the well-being of these patients.

Objective  To conduct a meta-analysis on the prevalence of depression in patients with visual impairment who regularly visit eye clinics and low vision rehabilitation services.

Data Sources  MEDLINE (inception to June 7, 2020) and Embase (inception to June 7, 2020) were searched.

Study Selection  Studies that obtained data on the association between acquired visual impairment and depression among individuals aged 18 years or older were identified and included in this review. Exclusion criteria comprised inherited or congenital eye diseases, review studies, unpublished articles, abstracts, theses, dissertations, and book chapters. Four independent reviewers analyzed the results of the search and performed the selection and data extraction to ensure accuracy.

Data Extraction and Synthesis  Meta-analyses of prevalence were conducted using random-intercept logistic regression models. This study followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.

Main Outcomes and Measures  Proportion of depression.

Results  A total of 27 studies were included in this review, and all but 2 included patients older than 65 years. Among 6992 total patients (mean [SD] age, 76 [13.9] years; 4195 women [60%]) with visual impairment, in 1687 patients with depression, the median proportion of depression was 0.30 (range, 0.03-0.54). The random-effects pooled estimate was 0.25 (95% CI, 0.19-0.33) with high heterogeneity (95% predictive interval, 0.05-0.70). No patient characteristic, measured at the study level, influenced the prevalence of depression, except for the inclusion of patients with cognitive impairment (0.33; 95% CI, 0.28-0.38 in 14 studies vs 0.18; 95% CI, 0.11-0.30 in 13 studies that excluded this with major comorbidities; P = .008). The prevalence of depression was high both in clinic-based studies (in 6 studies, 0.34; 95% CI, 0.23-0.47) and in rehabilitation services (in 18 studies, 0.25; 95% CI, 0.18-0.33 vs other settings in 3 studies, 0.15; 95% CI, 0.05-0.38; P = .17), and did not vary by visual impairment severity of mild (in 8 studies, 0.24; 95% CI, 0.14-0.38), moderate (in 10 studies, 0.29; 95% CI, 0.21-0.39), and severe (in 5 studies, 0.29; 95% CI, 0.12-0.56; P = .51).

Conclusions and Relevance  The results of this meta-analysis suggest that depression in patients with visual impairment is a common problem that should be recognized and addressed by the health care professionals treating these patients.

Low vision negatively affects quality of life and is associated with reduced functional ability, increased disability, falls, social isolation, institutionalization, and mortality. A recent systematic review identified hearing and visual loss as markers of frailty in community-based studies, with depression being a primary associated morbidity.¹ Depression is common in older adults and even more common in those with visual impairment. Clinically significant subthreshold depression has been found in one-third of older adults with visual impairment, approximately twice as high as the lifetime prevalence rates in older adults without visual impairment, where depressive symptoms are present roughly in 15%.^2-7 Depression is a serious medical condition, and even mild symptoms may affect quality of life.⁶ Patients with vision loss may experience a greater burden of their disability, as depression is often accompanied by low energy levels, sleep problems, cognitive problems, or disproportionate worrying. Depression can also affect a person’s learning capacity or the ability to retain information, make decisions, or achieve goals.⁸ Therefore, treatment of depression has increasingly gained attention within eye care settings as shown by numerous mental health care programs that have been tested and often found effective.⁹ A Cochrane systematic review investigated the effect of several interventions to improve the quality of life of patients with low vision and found that psychological therapies or group programs may reduce depression and increase self-esteem in people with low vision.¹⁰

Depression screening has been recommended as a part of low vision services, with appropriate training of rehabilitation professionals and the use of standardized questions in both high-income and low-middle–income countries.^11,12 In low vision services in Wales, barriers to optimal depression screening have been identified, including perceived patient reluctance to discuss depression, time constraints, and lack of confidence in addressing depression.¹¹ On the other hand, low vision professionals were found to lack confidence in their knowledge and skills to address depression.¹³ Because recent projections show a large increase in people with vision loss due to population growth and demographic aging,¹⁴ it is also important to understand the increasing burden of mental health needs in eye clinics and low vision and mental health centers. To our knowledge, no recent reviews or meta-analyses have been published on the prevalence of depression in patients visiting eye clinics and low vision services, regardless of ocular causes or age ranges. Our study aimed to investigate this prevalence with the goal of developing public health strategies for people accessing eye services.

This meta-analysis included a search of MEDLINE (via PubMed) and Embase databases from inception to June 7, 2020. This study followed the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) reporting guideline.

This meta-analysis included cross-sectional studies of adult patients (18 years or older) attending primary or secondary care eye services and low vision rehabilitation services. We included baseline data from randomized controlled trials only if they were considered representative of populations with visual impairment.

We accepted the diagnostic category of visual impairment as applied by the authors of each study, and we used the International Classification of Diseases, 11th Revision (ICD-11)¹⁵ to form study subgroups according to distance visual impairment severity, defined according to the best-corrected visual acuity in the better-seeing eye as mild (20/40 or better), moderate-severe (better than 20/200 but worse than 20/40), or blindness (worse than 20/400).

We accepted the diagnostic category of depression as applied by the authors of each study, including validated tools and questionnaires, depression according to the Diagnostic and Statistical Manual of Mental Disorders (Third Edition; DSM-III) or Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition; DSM-IV) or standard psychiatric diagnostic criteria, and self-reported depressive disorder. Original research had to be reported in English with adequate information on the prevalence of depression. We used the following exclusion criteria: inherited or congenital eye diseases only, review studies, unpublished articles, abstracts, theses, dissertations, and book chapters.

The MEDLINE and Embase search strategies are shown in detail in the eMethods in the Supplement. Four independent reviewers (E.M., F.M., D.P., and M.P.) analyzed the results of the search and performed the selection, classification of literature, and data extraction to ensure accuracy. Discrepancies were addressed by discussion or with a fifth reviewer (G.V.).

All included studies were subject to methodologic critical appraisal using an adapted risk of bias assessment for prevalence studies.¹⁶ The risk of bias assessment used 5 domains, which included lack of generalizability bias, record bias, attrition bias, detection bias, and reporting bias. Every domain received a maximum of 2 points (eMethods in the Supplement).

Study prevalence was pooled using the metaprop command in R software (R Foundation) with the following specifications, according to Schwarzer et al¹⁷: we fitted a random-intercept logistic regression model, used a maximum-likelihood estimator for tau², a logit transformation of proportions, and a Clopper-Pearson CI for individual studies.

Between-study heterogeneity was assessed graphically by inspecting the overlap of study CIs and also estimating and presenting the 95% predictive interval, which comprises between-study variance. A preplanned heterogeneity investigation was conducted, adding study-level categorical covariates to the model: age (median, 75 years or older vs younger than 75 years), inclusion of patients with major chronic conditions (vs exclusion), types of validated questionnaires or scales used to diagnose depression (most common tools in studies), and severity of visual loss (mild, moderate, or severe according to the reported better-seeing eye visual acuity distribution). Post-hoc subgroup analyses were conducted by sample size (median, 125 patients or more vs 125 patients or fewer). Further subgroup analyses by causes of low vision were not conducted, because studies generally included people with multiple diseases. Significance was set at P < .05, and all P values were 2-sided.

Our search found 7054 potentially related studies; 6909 articles were rejected as obviously unsuitable because they were unrelated to depression in people with visual impairment. Of the 145 remaining articles, 100 were rejected for a variety of reasons, including articles not written in English; those with unspecified data, ie, the number of people with depression and visual impairment was not available; subthreshold depression was used as a target disease definition; there was no representative sampling; the study population was identical to that of other included studies; or a longitudinal design was used. Forty-five remaining studies included patients with low vision in whom depression status was investigated. Of these, 18 studies were community- or population-based and will be considered in a future review. The remaining 27 studies were conducted in eye clinics or low vision centers and were included in this review, and all but 2 included patients older than 65 years (Figure 1).

A summary of the main characteristics of the included 27 studies^4,18-43 is presented in Table 1 and Table 2. Seven studies were conducted in Europe,^{19,20,25,26,30,33,35} 8 in the US,^{4,22,27,31,34,38,39,42} 3 in Asia,^18,21,43 8 in Oceania,^{23,24,28,29,32,37,40,41} and 1 in Africa.³⁶ The number of participants ranged from 53 to 990.

Studies were conducted in clinical eye care settings(6 studies, 796 patients),^{21,23,28,34-36} rehabilitation settings (18 studies, 5615 patients), or other health care settings (3 studies, 581 participants). Rehabilitation settings included low vision clinics and blindness-related society services,^{4,18-20,25,27,29-32,37-39,41,42} registered and nonregistered patients with visual impairment,³³ patients included in a randomized controlled trial on fall prevention,⁴⁰ and patients from the Society for the Blind database.²⁶ Other health care settings included residents with visual impairment living in a care facility for older adults²⁴ or in a nursing home for blind people,⁴³ and the assessment of drivers with visual impairment who had been in a car crash.²²

Various definitions of visual impairment were adopted in the studies (Table 1); 12 studies (44.4%) did not provide a definition of visual impairment.^{4,19,20,25-27,31,33,36,38,39,42} In all but 1 study (3.7%),²² patients were attending or were scheduled to attend low vision rehabilitation and were consistent with our target population. Six studies (22.2%) defined visual impairment as best corrected visual acuity of 20/40 or worse in the better-seeing eye.^{23,24,28,32,37,41} Three studies (11.1%) expressed visual acuity in logMAR, considering visual impairment to be visual acuity worse than 20/80.^18,34,40 One study (3.7%) used a visual impairment diagnosis based on the International Classification of Diseases, Ninth Revision (ICD-9) or International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10).²¹

Depression was assessed using a number of different tools: the Geriatric Depression Scale-15,⁴⁴ used in 8 studies (29.6%)^{20,21,33,37-40,43}; the Patient Health Questionnaire-9,^45,46 used in 3 studies (11.1%)^23,28,34; the Patient Health Questionnaire-2,⁴⁷ used in 2 studies (7.4%)^32,41; the Hospital Anxiety and Depression Scale,⁴⁸ used in 4 studies (14.8%)^19,25,26,36; and the Center for Epidemiologic Studies Depression Scale (CESDS),⁴⁹ used in 4 studies (14.8%).^4,22,31,42 Two studies (7.4%)^27,42 used a depression diagnosis based on the Diagnostic and Statistical Manual of Mental Disorders (DSM-III or DSM-IV).^50,51 Other tools used for depression diagnosis were the Beck Depression Inventory, the Cornell Scale for Depression in Dementia, and the Mini International Neuropsychiatric Interview. Self-reported depression was accepted in 1 study (3.7%).²⁹

Table 2 summarizes the risk of bias assessment for all included studies. Most studies had 1 or 2 points in the major domains of the quality scale used. Nine of the 27 studies (33.3%) reached a total score between 8 and 9 points,^{18,21,23,27,28,32,34,39,43} whereas 18 (66.7%) reached a total score between 6 and 7 points.^{4,19,20,22,24-26,29-31,33,35-38,40-42}

Generalizability bias reached a score of 2 points in 7 of the 27 studies (25.9%)^{18,23,28,34,35,41,43} and 1 point in 20 (74.1%).^{4,19-22,24-27,29-33,36-40,42} The main problems related to this quality domain were lack of clarity regarding the definition of visual impairment and/or depression. All studies were prospective and gained 2 points for record bias, although it was sometimes unclear whether consecutive patients were included. Regarding attrition bias, we assigned a score of 0 points to 3 studies (11.1%) in which the percentage of patients enrolled was not specified.^4,19,37 Seventeen studies (63.0%) gained 1 point because more than 10% of the eligible patients were not included.^{20,22,24-26,29-36,38,40-42} Only 1 study (3.7%)²⁹ obtained 2 points for detection bias, because all the other studies provided no information about masking with respect to vision status and were assigned 1 point. Regarding reporting bias, we assigned 1 point to 4 studies (14.8%),^29,33,35,36 and 23 studies (85.2%)^{4,18-28,30-32,34,37-43} gained 2 points. More details can be found in the eAppendix in the Supplement.

We included 27 studies with a median sample size of 125 patients (range, 42-990 patients). Among 6992 total patients (mean [SD] age, 76 [13.9] years; 4195 women [60%]) with visual impairment, in 1687 patients with depression, the median proportion of depression was 0.30 (range, 0.03-0.54).

Figure 2 presents the meta-analysis of the prevalence of depression in all studies. Overall, the random-effects pooled estimate was 0.25 (95% CI, 0.19-0.33), with high heterogeneity (95% predictive interval, 0.05-0.70).

We investigated whether study-level variables were associated with differences in prevalence of depression (forest plots and data are presented in eFigures 1-6 the Supplement).

Regarding patients’ characteristics, the proportion of depression did not differ according to median age of 75 years or older (in 16 studies,^{4,19,20,22-25,27,33,37-43} 0.25; 95% CI, 0.18-0.34) vs younger than 75 years (in 11 studies,^{18,21,26,28-32,34-36} 0.26; 95% CI, 0.15-0.39; P = .96) (eFigure 1 in the Supplement). Depression prevalence did not vary significantly with increasing severity of visual impairment of mild (in 9 studies,^{22-24,32,36-38,41,42} 0.24; 95% CI, 0.14-0.38), moderate (in 10 studies,^{18,20,21,26-28,30,34,39,40} 0.29; 95% CI, 0.21-0.39), and severe (in 5 studies,^{25,29,33,35,43} 0.29; 95% CI, 0.12-0.56). The severity of visual impairment was unclear in 3 studies (0.15; 95% CI, 0.08-0.26; overall P = .51) (eFigure 2 in the Supplement).^4,19,31 Depression was significantly higher in 14 studies with no exclusion on comorbidities (0.33; 95% CI, 0.28-0.38)^{4,18,20,21,25,27,31,33,34,38-41,43} compared with 13 studies that excluded patients with major comorbidities, mainly cognitive impairment (0.18; 95% CI, 0.11-0.30; P = .02)^{19,22-24,26,28-30,32,35-37,42} (eFigure 3 in the Supplement).

Regarding study design features, we found that smaller studies with a sample size below the median 125 patients showed a higher depression prevalence (0.39; 95% CI, 0.34-0.45) than larger studies (0.16; 95% CI, 0.11-0.24; P < .001) (eFigure 4 in the Supplement). Prevalence was slightly higher in clinic-based studies (in 6 studies,^{21,23,28,34-36} 0.34; 95% CI, 0.23-0.47) compared with those conducted in rehabilitation services (18 studies,^{4,18-20,25-27,29-33,37-42} 0.25; 95% CI, 0.18-0.33) and other settings (3 studies,^22,24,43 0.15; 95% CI, 0.05-0.38; overall P = .17), but overall these differences were not significant (eFigure 5 in the Supplement).

There were significant differences in depression rates using different questionnaires or criteria to diagnose depression (eFigure 6 in the Supplement). Higher levels of depression were found in 2 studies^18,35 using the Beck Depression Inventory (0.42; 95% CI, 0.26-0.59) compared with 17 studies^{19-21,23,25,26,28,32-34,36-41,43} using the Geriatric Depression Scale-15 (0.33; 95% CI, 0.27-0.40), Patient Health Questionnaire-2 (0.31; 95% CI, 0.21-0.42), and Hospital Anxiety and Depression Scale (0.31; 95% CI, 0.19-0.46) (overall P < .001). Lower levels of depression were found when the CESDS (in 4 studies,^4,22,31,42 0.13; 95% CI, 0.06-0.24) or other tools (in 3 studies,^24,27,30 0.16; 95% CI, 0.04-0.44) were used.

The pooled prevalence estimate of depression did not change (0.25; 95% CI, 0.19-0.33) after excluding 6 studies^{4,19,29,33,36,37} with a risk of bias score of less than 7 points and only slightly increased to 0.27 (95% CI, 0.21-0.34) after the exclusion of 3 studies^22,24,43 conducted in other health care settings.

The results of this meta-analysis indicate that 1 in 4 patients with visual impairment who attended eye care services were affected by depression. Studies in this review included mostly patients aged 65 years or older. The finding of depression was similar, or even more common, among patients in clinical services compared with rehabilitation services, which could reflect patients’ initial shock of receiving a diagnosis of an irreversible eye disease.⁵² Alternatively, the lower rates of depression associated with rehabilitation services could be due to the fact that initiating low vision rehabilitation has beneficial effects on perceived depression, or it could be due to self-selection of patients who were less depressed seeking help in low vision services. The high rates of depression in adults with visual impairment are often overlooked or underestimated in primary care offices and eye clinics. Although the high prevalence of depression among patients with visual impairment is more likely to be recognized by low vision rehabilitation specialists, these patients are commonly not assessed with appropriate diagnostic procedures nor referred for adequate treatment.^53-56

Further study-level factors associated with lower depression rates were investigated, but a similarly high prevalence was found across subgroups of age and visual impairment severity. Depression was more common in studies that included patients with both visual and cognitive impairment, in studies with a smaller sample size with the primary aim of detecting depression rather than multiple disabilities, and in studies that varied by type of diagnostic tool, with the CESDS commonly used in low prevalence studies. The CESDS may have yielded lower prevalence estimates because it is multidimensional and also has items related to anxiety, sleep, and loneliness. Despite differences in depression, these subgroup outcomes did not support the exclusion of specific patient subgroups from depression screening because a significant proportion of patients with visual impairment were depressed in all subpopulations and settings.

Clinic-based studies did not always report visual impairment in their inclusion criteria, but the setting was consistent with low vision services and eye clinics, suggesting that our results may be widely applicable to these settings. However, nearly all studies were conducted in high-income countries, to which the applicability of our results should be restricted.

These results suggest that all primary and specialized eye care professionals, not only those working in low vision services, should have an appropriate knowledge of this topic and adequate clinical experience to decide when and how to investigate the presence of depression in people with visual impairment and eventually refer patients with depression for appropriate treatment. Treatment and follow-up visits of integrated care, which coordinates ophthalmologists and psychiatric or psychological referrals, may maximize efficiency and lead to effective patient-centered care.^9,10,53-56

Our review of observational cross-sectional studies did not aim to prove a causal relationship between visual impairment and depression. Moreover, the definitions of visual impairment and the cutoffs used for depression diagnosis differed between studies. Nonetheless, the results of our review suggest the need for depression screening in patients attending eye clinics who are 65 years or older and have mild to severe visual loss, regardless of comorbidities.

Longitudinal studies have shown that visual impairment is associated with incident depression, which can be due to difficulties with reading, mobility, and driving.^57-59 However, longitudinal studies have also found that baseline depression is a predictor of the development of visual impairment.⁶⁰ The complexity of the interrelationship between depression and visual impairment should also be seen in the framework of the multiplicity of associations between ocular diseases and systemic diseases, specifically neurologic disease. The existence of common causal factors is also key to understanding such complex relationships. Previous studies have demonstrated that diseases of the eye, particularly age-related macular degeneration (AMD), may be associated with higher rates of Alzheimer and Parkinson disease.^61,62 On the other hand, smoking, alcohol intake, and lack of exercise are all independent risk factors for AMD, Alzheimer disease, and Parkinson disease,^63-67 as well as for cardiovascular disease and cancer. Regardless of the difficulty in assessing causal associations between different systemic diseases, our study results suggest that visual impairment may be used as a cross-sectional marker of multimorbidity and frailty in older patients.

We found few other reviews summarizing the association between depression and visual impairment in clinical settings.⁶⁸ Among older adults with visual impairment, patients with AMD seem to be at particularly high risk of depression compared with patients with other eye diseases.⁶⁹ Popescu et al⁷⁰ compared rates of depression in older adults with AMD, glaucoma, and Fuchs corneal dystrophy and found that the highest rate of depression (39%) was in the AMD group. In their review, Nyman et al⁷¹ found that working-age adults were more likely to report mental health issues; however, inconsistent results were found specifically for depression.

This study has some limitations. Our review did not aim to explore the relationship between visual impairment and depression. Moreover, the definitions of visual impairment and the cutoffs used for depression diagnosis differed between studies. A further limitation of our review is that studies yielded heterogeneous results, as is common for meta-analyses of prevalences, which limits the power of heterogeneity investigations.

Given that depression is treatable and some ocular diseases that cause visual impairment are reversible, early identification and treatment of people most at risk for depression could be associated with their overall well-being.⁷² The guidelines of the National Institute for Health and Care Excellence suggest that practitioners working in primary care and general hospital clinics should be aware that patients with a chronic physical health problem, especially those with a functional impairment such as visual impairment, are at high risk of depression. The National Institute for Health and Care Excellence recommends that patients with positive screening results to at least 1 of 2 standard questions (“During the last month, have you often been bothered by feeling down, depressed, or hopeless? During the last month, have you often been bothered by having little interest or pleasure in doing things?”) be referred to their general practitioner for appropriate assessment. The results of this meta-analysis suggest that screening should occur in both low vision settings, such as rehabilitation clinics, and in primary care and general clinical settings, where a high prevalence of visual impairment has been reported.⁷³

These findings also suggest that further research should address the clinical effectiveness of depression screening and treatment among patients with visual impairment. This would mean integrating a screening protocol to detect and treat depression in general eye clinics, low vision, and rehabilitation settings. In cases of major depression, a patient should be referred to their general practitioner or directly to a psychiatrist for appropriate care.⁷⁴

Accepted for Publication: April 8, 2021.

Published Online: May 27, 2021. doi:10.1001/jamaophthalmol.2021.1557

Corresponding Author: Mariacristina Parravano, MD, IRCCS-Fondazione Bietti, Via Livenza 3, Rome 00198, Italy (mcparravano@gmail.com).

Author Contributions: Drs Virgili and Lucenteforte had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Maurutto, Menchini, Lanzetta, Varano, Virgili.

Acquisition, analysis, or interpretation of data: Parravano, Petri, Maurutto, Lucenteforte, Menchini, van Nispen, Virgili.

Drafting of the manuscript: Petri, Maurutto, Menchini, Varano, Virgili.

Critical revision of the manuscript for important intellectual content: Parravano, Lucenteforte, Menchini, Lanzetta, van Nispen, Virgili.

Statistical analysis: Petri, Lucenteforte, Virgili.

Obtained funding: Virgili.

Administrative, technical, or material support: Petri, Maurutto, Virgili.

Supervision: Parravano, Menchini, Lanzetta, Varano, Virgili.

Conflict of Interest Disclosures: Dr Parravano reported receiving personal fees from Allergan, Novartis, Bayer, and Zeiss outside the submitted work. Dr Varano reported receiving personal fees from Allergan, Novartis, Bayer, Biogen, and Sifi, all for work on the advisory board. Dr van Nispen reported receiving consultation fees from MeiraGTx UK II Ltd and grants from Bayer BV for investigator-initiated research outside the submitted work. Dr Menchini reported receiving personal fees from Novartis Pharma AG. Dr Lanzetta reported serving as a consultant to Allergan, Bayer, Centervue, Novartis, and Roche outside the submitted work. No other disclosures were reported.

Funding/Support: This work was supported in part by the Italian Ministry of Health and Fondazione Roma under the Aging Network of Italian Research Hospitals.

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.