A simple score-based strategy to improve equity of the UK biennial diabetic eye screening protocol among people deemed as low risk

Aims/hypothesis

Biennial, as opposed to annual, screening for diabetic retinopathy was recently introduced within England for those considered to be at ‘low risk’. This study aims to examine the impact that annual vs biennial screening has on equitable risk of diagnosis of sight-threatening diabetic retinopathy (STDR) among people at ‘low risk’ and to develop an amelioration protocol.

Methods

In the North East London Diabetic Eye Screening Programme (NELDESP), 105,083 people without diabetic retinopathy were identified on two consecutive screening visits between January 2012 and September 2023. Data for these individuals were linked to electronic health records (EHR). Characteristics associated with subsequent STDR diagnosis were identified (including age, gender, ethnicity and diabetes duration), and logistic regression was performed to identify people who require annual screening, using variables available to the NELDESP and data from EHR. Simulations of the biennial screening protocol, and of protocols incorporating the outcomes of the logistic models and a simplified points model, were implemented, and the relative risk of STDR calculated at each screening appointment was compared amongst various population subgroups. The results were validated using data from the South East London DESP.

Results

Among the low-risk participants, there were 3694 incident STDR cases over a mean duration of 5.0 years (SD 3.4 years). Under the biennial screening protocol, almost all groups had a significantly higher risk of STDR diagnosis compared with people aged 41 years or older who were of white ethnicity and had been living with diabetes for <10 years. Compared with biennial screening, a simplified screening protocol based on age, diabetes duration and ethnicity reduced the number of delayed STDR diagnoses from 39% to 25%, with a more equitable performance across population groups, and a modest impact on screening appointment numbers (46% vs 57% reduction in annual screening appointments, respectively).

Conclusions/interpretation

A simple, clinically deliverable, personalised protocol for identifying who should be screened annually or biennially for diabetic eye disease would improve equity in risk of delayed STDR diagnosis per appointment.

Graphical Abstract