Comparison of Photographic Screening Methods for Diabetic Retinopathy - A Meta-analysis

Area of Science:

• Ophthalmology
• Medical Imaging
• Public Health

Background:

• Diabetic retinopathy (DR) is a primary cause of irreversible vision loss globally.
• Standardized DR screening practices, particularly imaging modalities, show significant variation.
• Effective screening is crucial for preventing vision impairment and blindness in diabetic patients.

Approach:

• A systematic review and meta-analysis adhering to PRISMA Guidelines.
• Searched MEDLINE, Web of Science, and other databases (1990-2019).
• Included 62 studies comparing mydriatic and non-mydriatic fundus photography against dilated fundus examination (DFE) as the reference standard.

Key Points:

• Sensitivity for detecting any DR (ADR) ranged from 81% (single field) to 99% (4-7 fields/wide-angle).
• Sensitivity for referable DR (RDR) ranged from 76% (single field) to 93% (wide-angle).
• Specificity ranged from 91% (wide-angle) to 99% (three fields); significant heterogeneity was observed.

Conclusions:

• DR screening sensitivity and specificity are positively correlated with the number of photographic fields used.
• Pooled data suggest that non-mydriatic two-field photography may be adequate for ADR and RDR screening.
• Further standardization of DR classification across studies is needed to reduce heterogeneity.