advertisement

---

Editors Selection IGR 23-4

Screening and Detection: Two birds with one stone? Piggy-backing on Diabetic Retinopathy Screening

Comment by Anthony Khawaja & Kelsey Stuart on:

112698 Evaluating the outcome of screening for glaucoma using colour fundus photography-based referral criteria in a teleophthalmology screening programme for diabetic retinopathy, Tan R; Teo KYC; Husain R et al., British Journal of Ophthalmology, 2023; 0:

See also comment(s) by Benton Chuter & Linda Zangwill •

Find related abstracts

---

The asymptomatic early stages of glaucoma mean that a substantial proportion of disease in the general population remains undetected, with more than half of all cases worldwide estimated to be undiagnosed or unaware of their diagnosis.¹ Opportunistic screening strategies may reduce the burden of glaucoma-related vision loss and blindness through early identification and treatment of these affected individuals.² Established diabetic retinopathy (DR) screening programs represent one such avenue for opportunistic case detection, as routinely collected color fundus photographs can be additionally screened for features of glaucomatous optic neuropathy.

In this case-control study from Singapore, Tan and colleagues examined the effectiveness of glaucoma screening using referral criteria assessed on imaging from a nationwide teleophthalmology screening program for DR. In total, 5023 diabetic subjects were referred to the Singapore National Eye Centre for further evaluation, and glaucoma suspects (n = 2626, based on vertical cup-to-disc ratio [VCDR] and/or other disc features) were compared to those without features of glaucoma (n = 2398).

Of the glaucoma suspects, 369 (14.1%) were confirmed to have disease, compared with 82 cases (3.4%) in the control group. Screening based on VCDR ≥ 0.65 demonstrated only modest sensitivity (70.5%) and low specificity (55.1%) for detecting glaucoma. Importantly, because of the low prevalence of disease in the general population, this strategy yielded a positive predictive value (PPV) of only 14.1% – meaning that six out of every seven suspects were ultimately found not to have glaucoma. Increasing the referral threshold to VCDR ≥ 0.80 improved the PPV to 46.1% (almost one in two referrals were now found to have disease) but with a substantial reduction in sensitivity (7.8%) – resulting in more than nine out of ten community cases being missed.

These results highlight the difficulty of glaucoma screening strategies based on VCDR and other optic disc characteristics. Lenient thresholds may detect the majority of cases in the community but come at the cost of high false-positive rates, while more stringent cutoffs may fail to detect the bulk of undiagnosed disease. Both approaches have direct clinical and financial implications, and ultimately, a cost-effective balance needs to be struck for such a strategy to be considered feasible in any given healthcare system.

The authors note that an alternative approach may be to use a stringent cutoff, but allow for repeat screening after an interval to enable subsequent detection of cases missed at baseline, however additional data are required regarding the risk of blindness in cases that may be missed at these screening thresholds. The ultimate aim is to identify previously undiagnosed glaucoma cases that have a high risk of blindness if not treated.

In the future, a multimodal approach combining advances in retinal imaging, artificial intelligence and polygenic risk scores, may lead to an appropriate balance of test characteristics to enable effective glaucoma screening on a population level.³

---

• ← Previous
• Next →

---