Prevalence of the burden of diseases causing visual impairment and blindness in South Africa in the period 2010–2020: a systematic scoping review and meta-analysis

Authors: Xulu-Kasaba, Z.N.; Kalinda, C.

Geographical coverage: South Africa

Sector: Burden of disease

Sub-sector: Epidemiology

Equity focus: None – interested in the population of South Africa as a whole.

Study population: People living in South Africa.

Review type: Other review

Quantitative synthesis method: Meta-analysis

Qualitative synthesis method: Not applicable

Background: The prevalence of visual impairment (VI) continues to rise, despite efforts to reduce it. The burden of disease negatively impacts the quality of life, education opportunities and other developments in various communities.

Objectives: This study aimed to determine and quantify the major causes of VI in South Africa, to ensure accurate interventions in addressing them and to reduce the burden of ocular disease in that context.

Main findings:

Overall, the review found a prevalence of VI of 2% for blindness and 12% for moderate and severe visual impairment (MSVI).

A total of 10 studies met the inclusion criteria for the final review; however, nine studies were eligible for quality assessment. The quality index for the included studies was 71.1%. The prevalence of VI was 2% for blindness and 12% for moderate and severe visual impairment (MSVI). Pooled prevalence identified uncorrected refractive error (URE) (43%), cataract (28%), glaucoma (7%), and diabetic retinopathy (4%) as major causes of MSVI. The leading causes of blindness were untreated cataracts (54%), glaucoma (17%), and diabetic retinopathy (57%). Ocular diseases causing VI are avoidable and similar to those of low to middle income countries. MSVI were caused by URE, cataract, glaucoma, and diabetic retinopathy. Blindness was mainly caused by cataracts, glaucoma, and diabetic retinopathy.

The high level of heterogeneity measured as I² in most pooled estimates ranged from 24 to 98. Assessment of the funnel plot and Doi plot ruled out a significant publication bias.

More than 50% of the included studies came from KwaZulu-Natal, one of the nine provinces of South Africa, a factor that may have possibly skewed the findings of the study. The demographics and population types in KwaZulu-Natal are rather different from those of areas like the Western Cape, for example, a far greater Indian population – which may influence susceptibility to certain conditions.

Auhtors note the importance of addressing the identified conditions by enabling HReH frameworks and systems that will seek to alleviate these causes of unnecessary blindness in South Africa. A strategic plan to manage these conditions would largely reduce the burden of VI in the country. Early screenings and interventions to maximise care at primary health levels would decrease the burden of avoidable blindness in the country significantly.

Given the geographical concentration of studies in this review noted above, more population-based studies should be conducted in provinces such as the Eastern Cape, which has the largest poverty-stricken population in South Africa, and Gauteng, the most densely populated province in the country. Future research should also explore the availability of human resources for Eye Health, to explore the extent to which VI and blindness can be managed in the country.

Methodology:

The research question guiding the study was: What are the major causes of VI and blindness in South Africa? The study search was guided by the Population Concept Context (PCC) model.

Inclusion criteria for this review consisted of studies on human subjects across all ages, studies on eye diseases, ocular conditions, refractive error, and visual impairment, primary studies, published in English in the period 2010 to 2020 and conducted within South Africa. Primary studies that quantified disease prevalence were included in this systematic review.

Databases searched for suitable articles published in English included PubMed, Google Scholar, and EBSCOhost databases: Health Source: Nursing/Academic Edition, Health source-Consumer, CINAHL, and Academic Search Complete. Following this search, unpublished rapid studies were further searched for in relevant repositories.

Two independent co-screeners conducted title, abstract, and full article screening using standardised, piloted tools. Data extraction was conducted by one reviewer, followed by data charting using a tool developed, piloted and finalised for this study. Information was extracted to give a quantitative account of the prevalence and major causes of VI and blindness in South Africa.

Nine of the 10 included articles were quality-indexed. This was performed by two independent reviewers, using the 2018 version of the Mixed-Method Appraisal Tool (MMAT). One study did not have a methodology section and so was excluded.

The Inverse Variance Heterogeneity (IVhet) model in MetaXL was used to calculate the Pooled Prevalence Estimates (PPE) and their 95% Confidence Intervals (CIs) for selected studies. This model was chosen over the Fixed-Effect (FE) or Random-Effect (RE) models due to its accurate coverage probability. PPEs were determined for each ocular problem where at least four studies provided prevalence data. Forest plots graphically displayed these estimates. Cochran’s Q statistic and I² were used to measure study heterogeneity. The Luis Furuya-Kanamori (LFK) index and funnel plots were used to assess publication bias.

Applicability/external validity: Authors noted that more than 50% of the included studies were conducted in one province (KwaZulu-Natal), which had unusual population characteristics. This potentially reduces the extent to which the estimates obtained from this study are more widely applicable.

Geographic focus: The article highlights the distinct characteristics of South Africa, with minimal discussion on applying the findings to other low- and middle-income countries.

Summary of quality assessment:

While the methods used to identify, include and critically appraise studies were generally sound, included studies were limited to those written in English and there was no evidence of relevant experts being conducted. The data analysis did not take sufficient account of differences between studies and how these might contribute to the overall prevalence estimates (in particular, the fact – flagged by the authors – that 50% of the included studies were conducted in one province of South Africa). For these reasons, we attributed low confidence in the results of this review.