Predicting the need for diabetic macular oedema treatment from photographic screening in the Singapore Integrated...

To identify diabetic maculopathy features from photographic screening that are predictive of treatment on referral to a tertiary care centre.

Retrospective review of participants who underwent screening by Singapore Integrated Diabetic Retinopathy Programme from 2015 to 2019. Participants underwent visual acuity (VA) test and non-stereoscopic retinal photographs. Maculopathy features include haemorrhages, microaneurysm and hard exudates (HE), stratified by inner and outer zone (1 and 1-2 disc diameter from fovea respectively) and VA of 6/12. Diabetic macular oedema (DMO) treatment was defined as intravitreal injection or macular photocoagulation up to 540 days from point of referral.

16,712 patients screened had referable eye disease. Out of 3518 maculopathy suspects, 281 (8.0%) received DMO treatment within 540 days. Those treated for DMO had shorter duration of diabetes (6.90 vs. 9.13 years, p < 0.001), higher total cholesterol (4.65 ± 1.20 vs. 4.36 ± 1.13 mmol/L, p = 0.001) and LDL cholesterol (2.59 ± 1.05 vs. 2.37 ± 0.93 mmol/L, p < 0.05) than those without treatment. High-risk features, including inner zone haemorrhages with VA ≤ 6/12 (HR 12.0, 95% CI: 5.5-25.9) and inner zone hard exudates (HR 7.4, 95% CI: 3.4-15.8), significantly increased the likelihood of requiring DMO treatment compared to low-risk features. Higher body mass index is protective of DMO treatment in mild non-proliferative diabetic retinopathy (HR 0.84, 95% CI: 0.73-0.97).

Haemorrhages, microaneurysms and HE within inner zone are important photographic features predictive of DMO treatment. VA is an important stratification for screening especially in patients with only visible haemorrhages.