Trends in yield and effects of screening intervals during 17 years of a large UK community-based diabetic retinopathy...

To describe changes in risk profiles and yield in a screening programme and to investigate relationships between retinopathy prevalence, screening interval and risk factors.

We analysed a population of predominantly Type 2 diabetic patients, managed in general practice, and screened between 1990 and 2006, with up to 17 years' follow-up and up to 14 screening episodes each. We investigated associations between referable or sight-threatening diabetic retinopathy (STDR), screening interval and frequency of repeated screening, whilst adjusting for age, duration and treatment of diabetes, hypertension treatment and period.

Of 63 622 screening episodes among 20 788 people, 16 094 (25%) identified any retinopathy, 3136 (4.9%) identified referable retinopathy and 384 (0.60%) identified STDR. The prevalence of screening-detected STDR decreased by 91%, from 1.7% in 1991-1993 to 0.16% in 2006. The prevalence of referable retinopathy increased from 2.0% in 1991-1993 to 6.7% in 1998-2001, then decreased to 4.7% in 2006. Compared with screening intervals of 12-18 months, screening intervals of 19-24 months were not associated with increased risk of referable retinopathy [adjusted odds ratio 0.93, 94% confidence interval (CI) 0.82-1.05], but screening intervals of more than 24 months were associated with increased risk (odds ratio 1.56, 95% CI 1.41-1.75). Screening intervals of < 12 months were associated with high risks of referable retinopathy and STDR.

Over time the risk of late diagnosis of STDR decreased, possibly attributable to earlier diagnosis of less severe retinopathy, decreasing risk factors and systematic screening. Screening intervals of up to 24 months should be considered for lower risk patients.