Abstract
Purpose :
To evaluate the efficacy of the Teleretinal Imaging (TRI) program at the VA Tennessee Valley Health System (VA TVHS) in the detection of diabetic retinopathy (DR) and to evaluate for redundancy in the system..
Methods :
This study consisted of a retrospective chart review of 100 TRI screens at the VATVHS from January 1, 2014 through December 31, 2014. The data collected included age, findings on screening examination, need for referral, prior and/or follow-up eye exam and proximity to TRI screen, findings on eye exam, and need for therapeutic intervention.
Results :
25 patients had abnormalities on TRI (11 with DR and 14 with other retinal pathology). Of these, 4 did not have follow-up. Of the remaining 21 patients, 11 (52%) had screening findings that did not correlate with findings on follow-up exam. 4 patients (16% of abnormal screens) were referred for urgent follow-up. 9% of screened patients had already seen their eye care specialist within the 10 months prior to TRI and 52% had eye exams within 10 months after regardless of screening findings. Of the 72 normal screens, 32 (43%) had an eye exam within the next 10 months. 3% of the images were unreadable requiring referral. The sample of TRIs that had eyecare specialist examinations showed a sensitivity of 58.2% with a 95% confidence interval of 32.92% to 81.56% and a specificity of 75.61% with a 95% confidence interval of 59.70% to 87.64%, giving a 24.39% false positive rate. The positive predictive value was 50.00% with a 95% confidence interval of 33.85% to 66.15%.
None of the abnormal findings evaluated by an eye care provider required therapeutic interventions
Conclusions :
There was poor correlation between abnormal findings on TRI and identified pathology by the eye care specialist. Patients who were referred were more likely to not have the pathology identified on the screen than to have it. This evidence suggests that the TRI screening is not accurately identifying patients needing follow-up eye care. This also showed redundancy of care in the system with 54% of patients having an eye exam within the 10 months before or after the screen regardless of results. Finally, none of the patients identified with DR on the screen required treatment. Our findings identify several areas of inefficiency and redundancy in the current execution of this TRI program. This study is limited by the retrospective nature and sample size.
This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.