The benefits of screening for diabetic retinopathy have been clearly demonstrated. However, adherence with regular eye care is less than optimal. Digital retinal imaging using store-and-forward technology (teleretinal imaging) is being used throughout the VHA to enhanced adherence with eye care. Teleretinal imaging is assumed to be sufficient to identify diabetic retinopathy but we do not know its ability to diagnose concomitant non-diabetic eye diseases such as glaucoma and cataract. These other diagnoses might be missed among individuals enrolled solely in a teleretinal imaging screening program. Also, other non-invasive testing that match components of the complete eye exam (e.g. visual acuity, intra-ocular pressure) may supplement results of retinal imaging, increasing its diagnostic agreement to the level of a complete eye exam. We hypothesized that teleretinal imaging supplemented with a non-invasive eye care assessment (hereafter referred to as a Technology Assisted Ophthalmic [TAO] exam) can serve as an alternative to a complete eye exam (CEE) in screening patients with diabetes.
The study employed a prospective, observational design in which participants had both a TAO exam and a CEE performed by an eye care professional. The two exams were compared for level of agreement for the following outcome measures: eye care referrals (for any ocular findings), level of diabetic retinopathy, and the presence of referable non-diabetic ocular disease.
The protocol required participants to have both a TAO exam and CEE on the same day. The TAO exam was completed by a trained technician and results were interpreted remotely by an optometrist. The CEE was performed by a separate optometrist. Both were masked to the results of the alternative exam. The optometrists completed a results template that captured referable eye diseases, level of diabetic retinopathy, and non-diabetic ocular pathology (e.g. cataract, glaucoma suspect, macular degeneration and reduced visual acuity). If any component of the TAO exam was ungradeable, the patient received a referral recommendation. A subset of participants with a normal baseline CEE had a follow-up visit one year later to assess for incidence of new ocular findings. Since the CEE served as the clinical "gold standard" for the TAO exam, we performed a sub-study in which 45 participants had a second CEE by an independent examiner within 30 days, to validate the original findings.
Level of agreement was determined by both percent agreement and by kappa statistic for inter-rater agreement. Outcomes of interest were overall need for eye care referral and for agreement on individual ocular findings including diabetic retinopathy. Since each instance (baseline and follow-up) served as an independent evaluation, we determined level of agreement for the combined baseline and follow-up exams.
Three hundred and seventeen (317) participants were enrolled, all of whom had baseline studies; 72 participants with a normal baseline CEE had a follow-up exam one year later. There was substantial agreement for eye care referral recommendations for both baseline (level of agreement - 80%, kappa - 0.63) and the combined baseline and follow-up exams (level of agreement - 85%, kappa - 0.67). In the combined groups, individual ocular diagnoses also had very high level of agreement, ranging from 83-97% (kappa - 0.38-0.78), with cataract having the highest agreement and the category of "other referable findings" (which encompassed more rare eye findings) having the lowest agreement. The overall prevalence of ocular findings varied from 3-20% (macular degeneration- 3%, diabetic retinopathy- 4%, cataract- 4% and suspicion of glaucoma- 20%).
The TAO exam had more false positive referral recommendations since it conservatively referred in the presence of any ungradeable findings (24%). However, many of these patients were appropriately referred since other ocular findings were often present and graded appropriately on the exam.
There was a high level of agreement for the presence or absence of diabetic retinopathy (96%, kappa - 0.65). Examiners also had substantial agreement in grading levels of diabetic retinopathy (from mild retinopathy to proliferative retinopathy), but the TAO exam over-called 12 cases and under-called 4 cases.
The incidence of retinopathy or other ocular findings in the 72 patients with follow-up exams was very low. Overall, 6 patients (8%) developed referable eye findings as determined by the CEE. Again, there was substantial agreement between the TAO and CEE for both the overall referral recommendation (93%, kappa - 0.63) and for referral related to diabetic retinopathy (99%, kappa - 0.66).
While the CEE served as the standard to which the TAO exam was compared, even these exams had variations in findings. Of 45 participants who had two CEE performed within one month, there was 93% agreement (kappa - 0.73) for the presence of referable ocular findings. Five patients had referable findings confirmed on the second CEE. However, three patients with initial findings were not confirmed at the second CEE.
Teleretinal imaging has been deployed in the care of patients with diabetes based on small validation studies evaluating its ability to diagnose diabetic retinopathy. This study provides strength to the reliability and impact of teleretinal imaging as a tool in the management of patients with diabetes. We have also added to the field by including an assessment of non-diabetic ocular findings. We found a similarly high level of agreement for entities such as glaucoma, age-related macular degeneration, and cataract. The addition of other ocular assessment tools such as visual acuity and intra-ocular pressure added minimally to the results.
The TAO exam did have more false-positive referrals (often due to ungradeable images) and under-called a few retinopathy diagnoses. But we also revealed that the CEE, which serves as the basis for clinical decision-making, is also subject to variation. There were a few cases of disagreement when two CEE exams were performed over a short interval.
Thus, these findings validate the use of teleretinal imaging in patients with diabetes and mitigate any concerns about its ability to identify patients with vision-threatening conditions including diabetic retinopathy and other ocular diseases.
External Links for this Project
- Cavallerano AA, Conlin PR. Teleretinal imaging to screen for diabetic retinopathy in the Veterans Health Administration. Journal of Diabetes Science and Technology. 2008 Jan 1; 2(1):33-9. [view]
- Wakefield BJ. Nursing workload measurement in remote monitoring
Remote monitoring pre-conference. Paper presented at: American Telemedicine Association Annual Meeting; 2008 Apr 8; Seattle, WA. [view]