# Performance and Effectiveness of Diabetic Retinopathy Screening in Portugal: An Outcome-Based Evaluation

## TL;DR

This study evaluates Portugal's diabetic retinopathy screening program, finding it moderately effective but with a high rate of unnecessary referrals.

## Contribution

The study provides an outcome-based evaluation of diabetic retinopathy screening effectiveness and diagnostic agreement in Portugal.

## Key findings

- The screening program had a 55.9% positive predictive value and 44.1% false positive rate.
- Agreement between screening and hospital diagnoses was highest for pre-proliferative DR (R2) and maculopathy (M1).
- Cohen’s Kappa values indicated only slight agreement between screening and hospital classifications.

## Abstract

Background/Objectives: Diabetic retinopathy (DR) is the leading cause of preventable blindness among working-age adults. Early detection through screening programs is essential for managing the condition and preventing visual impairment. In Portugal, the national DR screening program (DR SP) targets diabetic patients, aiming to detect DR at an early stage and refer patients requiring intervention for an ophthalmology appointment. This study aims to assess the effectiveness of the Portuguese DR SP by analyzing patients referred for a hospital appointment following a positive screening result. Methods: An observational retrospective cohort study was conducted at Unidade Local de Saúde de São João (ULS-SJ), including patients referred to a DR SP hospital appointment between January 2020 and December 2023. Data were collected from hospital records upon approval by the Hospital Ethics Committee. Screening and hospital diagnoses were compared for agreement. The Chi-Square test and Cohen’s Kappa were used to assess the association between screening and hospital diagnoses. Results: A total of 1126 patients (2251 retinographies) were analyzed. The median time from screening to hospital consultation was 63 days (Interquartile Range = 39–99), though referral times varied widely within the same classifications (ranging from 8 to 354 days). The most common screening classifications were R2 (pre-proliferative DR, 47.8%) and M1 (maculopathy, 24.6%). In eyes with DR, agreement between screening and hospital diagnoses was highest for R2 (40.1%) and M1 (32.3%), while proliferative DR (R3) showed 30% agreement. The positive predictive value (PPV) of the screening program was 55.9%, with a false positive rate of 44.1%. A statistically significant association between screening and hospital diagnoses was observed (p < 0.001, Chi-Square test), though Cohen’s Kappa values (0.167 Right Eye, 0.157 Left eye) indicated only slight agreement. Conclusions: Our study found that DR SP effectively identifies patients needing ophthalmologic evaluation with moderate diagnostic agreement and a relatively high false positive rate, leading to unnecessary referrals. While this ensures that sight-threatening cases are not missed, improvements in grader training, classification protocols, and Optical Coherence Tomography (OCT) integration could improve results. Strengthening screening adherence and optimizing referral pathways would further improve the program’s impact on early DR detection and management.