Effectiveness of screening and monitoring tests for diabetic retinopathy – a systematic review

Methodological quality of the review: Low confidence

Author: Hutchinson A, McIntosh A, Peters J, O’Keefe C, Khunti K, Baker R, Booth A.

Region: Details not provided

Sector: Diabetic retinopathy

Sub-sector: Screening and monitoring tests

Equity focus: None specified

Review type: Effectiveness review

Quantitative synthesis method: Narrative analysis

Qualitative synthesis methods: Not applicable

Background       

Diabetic retinopathy is a common complication of diabetes and the leading cause of blindness in people under the age of 60 in industrialized countries. There is inconsistent and inconclusive evidence available on the most effective methods for the screening of diabetic retinopathy.

Research objectives

To determine which screening and monitoring tests for diabetic retinopathy are most effective and under what circumstances.

Main findings

The authors identified 22 cohort studies for inclusion that specifically addressed screening for and early management of diabetic retinopathy.

Authors noted that available studies were limited in their ability to answer key questions on the effectiveness of tests for early detection of diabetic retinopathy. Included studies supported the conclusion that retinal photography under mydriasis was identified as the most effective test for diabetic retinopathy, with the majority of studies included reporting levels of sensitivity in excess of 80%. Nevertheless, this effectiveness reduced when photographs are ungradable. Ophthalmoscopy as a screening test could also reach acceptable standards of sensitivity and specificity.

Authors concluded from the study that the most effective strategy for testing diabetic retinopathy is the use of mydriatic retinal photography with the additional use of ophthalmology for cases where photographs are ungradable.

Authors noted that future research should focus on effective implementation methods to support the decision-making process and/or monitor the effectiveness of any screening programme.

Methodology

Cohort studies that specifically addressed screening for and early management of diabetic retinopathy were included in the review. The review included studies investigating both type 1 and type 2 diabetes. Studies were included if they were prospective in design and compared at least one screening method in a blinded fashion with a reference standard and all patients included in the study had the reference standard examination. No randomized control trials were identified, hence not included in the review.

Authors searched a number of databases including CINAHL, Cochrane Trials Register, EMBASE, Healthstar, MEDLINE, and OpenSIGLE from 1983 onwards. Only studies written in English were included in the review. The authors felt that the studies selected used a variety of reference standards and outcomes that made comparison between the studies difficult and therefore a meta-analysis of the data was not conducted. Two authors independently screened full texts of identified studies and extracted data of included studies. However, authors did not assess the quality and risk of bias of included studies.

Applicability/external validity

The authors noted that the review offers a ‘general trend’ of the result, since the majority of studies included in the review contained a small number of cases and ‘potential selection bias could limit any generalizability between them’.

Geographic focus

Authors did not report the geographical focus of studies included in the review.

Quality assessment

This review was attributed low confidence in the conclusions about the effects as major limitations were identified. This review was based on a thorough search of the literature on relevant databases, nevertheless authors did not report searching references lists in included studies and contacting authors/experts for additional studies. Additionally, language bias was not avoided as only English-language studies were eligible for inclusion. Authors did not report quality assessing studies included and they also do not clarify which evidence is subject to low risk of bias.

The authors felt that the studies selected used a variety of reference standards and outcomes that made comparison between the studies difficult and therefore a meta-analysis of the data was not conducted. Authors recognized certain limitations of the review, including the small sample sizes of the majority of studies included. Therefore authors concluded that the review offers a ‘general trend’ of results rather than robust conclusions.