Abstract
Purpose :
Large teleretinal screening programs have the potential to reduce the burden of eye disease by detecting and treating conditions early. We performed a retrospective cohort analysis in order to determine the accuracy of teleretinal detection program compared to traditional clinical exam in regards to secondary outcomes. Secondary outcomes were defined as presence of glaucoma or cataract in either eye.
Methods :
Our cohort included patients with type 1 or type 2 diabetes in the Grady Healthcare System in Atlanta, GA who were screened via a teleretinal screening program utilizing nonmydriatic fundus photography from January 2018 to October 9th, 2018. Threshold for clinical referral was the presence of primary outcome defined as positive screen for proliferative diabetic retinopathy (PDR) and/or diabetic macular edema (DME). Teleretinal screening results were compared to clinical exam for presence of secondary outcomes. Agreement between automated teleretinal findings and clinical exam were calculated via Cohen's kappa (k) coefficient.
Results :
Of the 7099 patients screened via teleretinal program, 394 met the threshold criteria for clinical referral. Among those 394 referred patients, 52% (n=205) had positive screening for PDR, 62.4% (n=246) had positive screening for DME, 3.3% (n=13) had positive screening for glaucoma, and 0.8% (n=3) had positive screening for cataract. Agreement between screening result and clinical exam in regards to glaucoma was fair (k=0.23), cataract was slight (k=0.01), and combined secondary outcomes was slight (k=0.04). The predictive values and sensitivity/specificity of the teleretinal screening program in comparison to clinical examination is shown in Table 1.
Conclusions :
The Grady teleretinal screening system demonstrates a low agreement with clinical examination in regards to secondary outcomes. The sensitivity and PPV of all outcomes were low, however specificity and NPV were high across all secondary outcomes with corresponding narrow confidence intervals. This study shows that utilizing a teleretinal screening program in order to detect secondary outcomes such as glaucoma and cataracts has limitations. Further studies with larger sample sizes and greater follow up are needed to better understand the role of teleretinal screening for secondary outcomes.
This is a 2020 ARVO Annual Meeting abstract.