Impact of Diabetic Retinopathy Screening

A systematic review 1 aimed at quantifying case detection and blindness prevention attainable through screening for diabetic retinopathy in a district population. Of the screening tests, mydriatic camera was the most sensitive (88%), and the other tests did not differ significantly from each other (point estimates from 50% to 65%). Published specificity estimates were between 90% and 100%. The relative risk of blindness in a person would lie between 0.15 and 0.39, with a midpoint estimate of 0.23. Of those treated, about 6% would be prevented from going blind within a year of treatment and 34% within 10 years of treatment.

The review has a thorough literature search and clear inclusion criteria. The effectiveness of treatment was estimated by using a previous meta-analysis and one additional RCT. The results of the RCTs are based on eyes and not people, and effectiveness in both eyes is unlikely to be independent.

Another systematic review by the same authors 3 abstracted in DAREincluded 10 prevalence and incidence studies, 14 studies on validity and reliability of screening, 5 studies on feasibility of screening programmes, and 6 studies on the effectiveness of treatment. In general-practice-based British population, the rate of proliferative retinopathy was 6% for the first round of screening, and about 2% at subsequent annual screening rounds. Retinal photography is more sensitive than direct ophthalmoscopy. Feasibility of the screening programmes has been demonstrated in Britain and other countries. The relative risk of visual loss or blindness was 0.39 (95% CI 0.28 to 0.55) after treatment with laser photocoagulation as compared to control eyes. For eyes wity non-proliferative or early proliferative retinopathy, early laser photocoagulation resulted in a relative risk of 0.77 (95% CI 0.56 to 1.06). For patients with both macular oedema and mild to moderate non-proliferative retinopathy, treatment is more effective (relative risk 0.44).

Comment: The quality of evidence is downgraded by limitations in review methodology.

The following decision support rules contain links to this evidence summary:

• Screening for diabetic retinopathy http://www.ebmeds.org/ebmeds/ebmeds_home.asp?mode=scripts&submode=view&id=scr00491&country=UK

References

• Bachmann MO, Nelson SJ. Impact of diabetic retinopathy screening on a British district population: case detection and blindness prevention in an evidence-based model. J Epidemiol Community Health 1998 Jan;52(1):45-52. [PubMed]
• Bachmann M, Nelson S. Screening for diabetic retinopathy: a quantitative overview of evidence, applied to the populations of health authorities and boards. Health Care Evaluation Unit 1996;1-46. [DARE]