Disparities Between Teleretinal Imaging Findings and Patient-Reported Diabetic Retinopathy Status and Follow-up Eye Care Interval: A 10-Year Prospective Study

Even in the presence of advanced eye disease, diabetic retinopathy (DR) can often be asymptomatic, and vision can remain unchanged beyond the time when intervention is most effective. Although current therapies are effective against vision-threatening complications of diabetes, prevention of visual loss from DR is critically dependent upon patient DR awareness and adherence to appropriately scheduled ophthalmic follow-up and treatment (1–5). As a result, there are well-established national and international guidelines, which generally recommend annual or more frequent ocular examinations for patients, starting either at diagnosis for type 2 diabetes or within 5 years of type 1 diabetes onset (6,7).

In the U.S., only ∼60% of individuals with diabetes receive the recommended minimum annual eye examination (8,9). Despite national initiatives to improve vision health, this percentage remained constant at ∼63% from 1997 to 2010 (9). Unawareness of DR is a major contributor to lack of adherence with eye care guidelines and increased risk of vision loss (10). The extent to which eye care professionals adhere to evidence-based guidelines for monitoring DR is a critical component for prevention of vision loss and blindness (8,11).

Associations between photographically documented DR severity, patient-reported DR awareness, concordance between patient-reported time to next ophthalmic follow-up and recommendations from national guidelines, and provider specialization in a teleretinal screening program have not been fully investigated. In this study, we evaluated patient self-reported awareness of DR and concordance of scheduled follow-up with recommendations based on concurrently obtained image-documented DR severity in a cohort of 25,360 consecutive patients with diabetes over a 10.25-year period in a teleretinal screening program.

Between 1 January 2010 and 31 March 2020, we evaluated all adult patients with diabetes examined at the Joslin Diabetes Center who underwent Joslin Vision Network (JVN) telemedicine program retinal imaging and were not receiving eye care at Joslin (12). JVN assessments were performed as part of routine medical evaluation and not as part of an eye visit. This cohort represented ∼30% of all adult patients evaluated at Joslin for diabetes during this period who received their eye care elsewhere. The study design was consistent with the tenets of the Declaration of Helsinki and approved by the institutional review board of Joslin, which waived the need to obtain informed consent from the patients. The reporting of study data was in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology statement (13).

JVN is a validated American Telemedicine Association category 3 (14) diabetes telehealth eye care program that integrates high-quality nonmydriatic retinal imaging, expert DR severity grading, patient education, and access to specialized diabetes eye care (15–18). As a component of JVN care, certified imagers use a structured interview recorded on standardized templates to determine demographic characteristics and medical, diabetes, and ocular history before retinal imaging. The awareness data obtained in this study represent patient self-reported information at the time of concurrent retinal imaging and reflect the patient’s understanding of DR presence or absence during any prior eye care visit. They do not reflect the accuracy of the self-report as compared with any prior eye care evaluations. The interview questions included the following:

• 1.

  When was your last eye examination performed?

• 2.

  Who performed your last eye examination?

• 3.

  Were your eyes dilated for a retinal evaluation?

• 4.

  When is your next scheduled eye examination?

• 5.

  Have you ever been told that diabetes has affected the back of your eyes or the retina?

Image graders evaluated DR and diabetic macular edema (DME) severity using validated JVN protocols (16,18,19). Eye care provider specialization was categorized as optometrist, nonretinal ophthalmologist, or retinal specialist ophthalmologist after referencing publicly accessible online national and regional databases (20–23). Additional queries using Internet search engines were performed if needed. It is important to emphasize that the awareness information in this study relates only to the understanding the patient had of any past retinopathy at the time of teleretinal imaging and does not reflect its accuracy in terms of what was clinically present at any prior eye examination.

From 1 January 2010 to 31 March 2012, patients underwent imaging using lowlight-adapted nonmydriatic fundus photography (16), and from 1 April 2012 to 31 March 2020, patients underwent imaging using nonmydriatic ultrawide field imaging (18). Both types of imaging were acquired by certified retinal imagers following validated protocols.

Certified graders evaluated images for DR and DME severity using the modified Early Treatment Diabetic Retinopathy (ETDRS) clinical severity scale. Planned follow-up interval was calculated based on patient report of the date of the most recent ocular examination and next scheduled visit with an eye care provider. Target follow-up interval was determined based on American Academy of Ophthalmology (6) guidelines using the clinical ETDRS severity level of DR or DME present on JVN imaging. Follow-up was considered not concordant with the imaging findings if no eye examination was scheduled or if the time to scheduled eye examination was longer than that suggested by American Academy of Ophthalmology Preferred Practice Patterns based on DR severity present on the retinal images.

On the basis of retinal image grading, patients were categorized into three groups: 1) patients with diabetes but no DR (ETDRS level 10), 2) patients with minimal or mild nonproliferative DR (level 20 or 35) and no DME, and 3) patients with vision-threatening DR (vtDR) defined as moderate nonproliferative DR or worse (level ≥43) or presence of any level of DME. The threshold for vision-threatening disease was based on the increased risk of retinopathy progression and visual loss in this group and is reflected by a recommended change from annual to more frequent evaluations because of the likelihood of required future treatment (6).

The primary outcome evaluated was patient-reported awareness of DR presence and concordance of follow-up recommendations with national guidelines based on photographically determined DR severity at the time of retinal imaging. Secondary analyses focused on the relationship between eye care provider specialization and patient awareness of DR and concordance of follow-up. In addition, exploratory analyses were performed to understand whether there were changes in patient awareness and concordance of follow-up over time. Nonparametric analyses (Wilcoxon rank sum test) were used to compare distributions of continuous variables between groups. The χ² test was used to compare frequencies of categorical variables. Multivariable adjusted P values were obtained using linear regression models. Severity of DR was evaluated per person, with the more severe level of DR and DME present in either eye determining the severity assigned. If one eye was ungradable, the level of disease present in the gradable eye was considered the severity of DR and DME present for that individual. The Cochrane-Armitage test was used for trend analyses. All statistical analyses were performed with SAS (version 9.4; SAS Institute, Inc., Cary, NC). To account for multiple comparisons, a P value <0.001 was considered statistically significant.

A total of 26,876 consecutive adults with diabetes who had not undergone previous ophthalmic examination at Joslin Diabetes Center were evaluated over 10.25 years (Supplementary Fig. 1). Table 1 summarizes cohort demographics overall and stratified by DR severity. Awareness information was available for 26,384 participants (98.2%), and JVN images were available and gradable in at least one eye for level of DR and DME in 25,742 participants (95.7%). Awareness information and photographic assessment of DR severity were available in 25,360 participants (94.3%). Compared with this group, patients with either missing awareness information or no photographic assessment (n = 1,516; 5.6%) were older (mean age 63.9 vs. 53.3 years; P < 0.001), were non-White (27.7% vs. 18.8%; P < 0.001), and had longer diabetes duration (mean 15.9 vs. 13.4 years; P < 0.001). No statistically significant differences between those with and without awareness and photographic documentation were observed for sex, HbA_1c, or insulin use. Patients without awareness information or gradable DR photographs were excluded from analyses.

Among the 25,360 patients with gradable images and awareness information, no DR was observed in 14,317 (56.5%), mild DR in 6,895 (26.8%), and vtDR in 4,238 (16.7%). Compared with those with no DR, patients with vtDR had longer diabetes duration (9.2 vs. 22.6 years; P < 0.001) and higher HbA_1c (7.8% vs. 8.6%; P < 0.001), and a larger proportion were using insulin (36.5% vs. 66.2%; P < 0.001). Analysis for trend was significant across all three groups (P < 0.001).

More than 96.7% of patients without documented DR on JVN images reported no prior DR (Table 1). Furthermore, 88.5% with mild DR on retinal imaging and 54.9% of those with vtDR did not report any DR in the past. Less severe DR was associated with higher self-reported absence of prior retinopathy (P < 0.001 trend), with 87.5%, 67,9%, 52.5%, and 31.3% of participants with mild, moderate, and severe nonproliferative DR and proliferative DR, respectively, reporting absence.

In patients with photographically documented DR, reporting no previous DR was associated with shorter diabetes duration, lower HbA_1c, lack of previous dilation, last eye examination >1 year before imaging, lack of future scheduled eye examination, no DME, and lower eye care provider specialization (Table 2). In these patients, there were no significant associations between awareness and ethnicity (aware: White vs. non-White 22.1% vs. 23.8%; P = 0.37).

Multivariable regression analyses including age, diabetes duration, HbA_1c, last eye examination within 1 year, future scheduled eye examination, DR severity, and DME severity as covariates demonstrated that unawareness of current DR was independently associated with shorter diabetes duration (odds ratio 1.038; 95% CI 1.033–1.043; P < 0.001), eye examination >1 year ago (1.419; 1.270–1.586; P < 0.001), lack of a future scheduled eye examination (2.176; 1.865–2.540; P < 0.001), milder DR (1.928; 1.841–2.020; P < 0.001), and no DME (1.684; 1.571–1.805; P < 0.001).

Pupillary dilation rates reported for prior eye examination were high (96.9%). In patients with DR, lack of previous dilation was associated with increased unawareness of current DR (not dilated vs. dilated 84.6% vs. 74.7%; P < 0.001). Over the 10.25-year period, awareness among patients with photographically documented DR (n = 11,043) remained unchanged from January 2010 to June 2017 (January 2010 to June 2012 21.7%; June 2012 to December 2014 21.5%; January 2015 to June 2017 22.5%; P = 0.95) but seemed to be improving from June 2017 to March 2020 (34.3%; P < 0.001) (Supplementary Fig. 2A). Follow-up concordance during these quartiles was similar, except for vtDR, which did not improve in any time period (Supplementary Fig. 2B).

The time distributions since the last reported ophthalmic examination and the follow-up interval recommended at that visit for the next eye examination are shown for the whole cohort by DR severity subgroup in Table 1 and among those aware of the presence of DR in Table 3. More than 22.3% (n = 5,648) of patients either reported not having an upcoming scheduled eye examination or did not recall approximately when they needed to return. With worsening DR severity, this percentage decreased (no DR 21.1% [n = 3,568]; mild DR 18.8% [n = 1,278]; vtDR 18.9% [n = 802]; P < 0.001).

Among only those patients reporting scheduled eye care follow-up (no DR 75.1% [n = 10,749]; mild DR 81.2% [n = 5,527]; vtDR 81.1% [n = 3,436]), patient-reported provider-prescribed follow-up agreed with recommended practice patterns in 88.5% (n = 17,450) overall and in 96.5% (n = 10,368), 97.4% (n = 5,386), and 49.4% (n = 1,696) of those with no DR, mild DR, and vtDR, respectively. Compared with patients with no or mild DR, the proportion without concordant follow-up increased among those with vtDR (no DR 96.5% vs. 49.4%; P < 0.001; mild DR 97.4% vs. 49.4%; P < 0.001).

Overall, 0.93% (n = 237) of the cohort reported no prior eye care provider, 20.4% (n = 5,164) reported an optometrist, 28.7% (n = 7,267) reported a nonretinal ophthalmologist, and 21.9% (n = 5,560) reported a retinal specialist ophthalmologist. Provider type could not be determined for 7,132 patients (28.1%), because either the patient could not recall provider details (n = 645; 2.5%) or the reported provider type could not be determined from available data resources (n = 6,487; 25.6%). With increasing DR severity, the percentage of patients with unknown provider type decreased (no DR 30.6%; mild DR 27.0%; vtDR 21.7%; P < 0.001). Failure to report a provider or not knowing the provider was more common among patients who reported being unaware of current DR (unaware vs. aware 30.4% vs. 14.2%; P < 0.001). Table 1 summarizes the distribution of eye care provider specialization for patients with a previously reported eye examination. There was a trend toward receiving care from more specialized providers as DR severity increased (P < 0.001). Individuals seeing retinal specialists increased from 15.1% for those with no DR to 45.3% for those with vtDR.

On the basis of DR severity identified at retinal imaging, increasing awareness and closer adherence to recommended follow-up were associated with increasing provider specialization (Supplementary Fig. 2C and D). This trend was most pronounced among patients with vtDR. Among patients with vtDR, retinal imaging recommended an earlier eye examination than the patient-reported scheduled follow-up in 74.6% when seen by an optometrist, 58.7% when seen by a nonretinal ophthalmologist, and 33.3% when seen by a retinal physician. In this group with vtDR, patients seeing a retinal physician had the highest DR awareness rate at 60.32% and the highest concordance with follow-up at 66.7%. The association between increased awareness in those with DR and increasing eye care provider specialization remained significant (P < 0.001) after adjusting for baseline differences in age, diabetes duration, DR severity, and DME severity in each of the groups. Table 4 lists the data on patients with vtDR with prior examination and scheduled future examination and who were aware of the presence of DR in their eyes. A sensitivity analysis was conducted to evaluate the potential impact of the 28.1% for whom provider type could not be determined. The association between increasing specialization of provider type and patient awareness of DR remained significant even when all (100%) of the unaware patients with undetermined provider type were assumed to have seen a retinal specialist (P < 0.001).

These data from >25,000 patients demonstrate a substantial disparity between photographically documented DR severity at the time of patient visit and self-reported awareness of prior retinopathy. The same is true for concordance of self-reported scheduled follow-up ophthalmic examination with evidence-based national guidelines. These two components are fundamental to appropriate management of patients at risk of diabetic eye disease (1–5). Unawareness of DR and its adverse effects on vision is a major contributor to lack of adherence with recommended eye care guidelines and increased risk of vision loss (10).

In this study, even among those with photographically documented vtDR, >54% reported being unaware of having any prior DR, and >51% did not report ophthalmic follow-up concordant with national standards. Conversely, among those without DR on photographic imaging, only 3.2% self-reported DR. It is not surprising that self-reported awareness of DR would be higher in those with vtDR compared with those with mild or nonexistent disease; however, the fact that fewer than half of patients with vtDR reported being aware of any retinopathy suggests a major information, understanding, and/or communication deficit.

These deficits exist despite increased national efforts to emphasize medical care standards (24–29) and despite evaluation at a tertiary-care diabetes-specific academic medical center. Given same-day photographic documentation in >25,000 patients, our study provides robust evidence that patient self-report of DR cannot be used reliably to determine absence of vtDR or appropriateness of scheduled ophthalmic follow-up, regardless of DR severity, scheduled future ophthalmic appointment, or type of provider delivering the prior DR assessment. Even among patients with image-documented vtDR, under the care of a retinal specialist, with a prior dilated eye examination in the past year, and with a scheduled ophthalmic appointment, >26% were not aware of any DR in the past, and >33% did not report concordant follow-up.

The implications of these findings for the care of patients with diabetes are substantial and affect primary care, diabetes/endocrinology specialists, and eye care providers. Because patient self-report alone cannot be used as a foundation from which to determine DR presence or concordance of planned ophthalmic follow-up with national guidelines (30), there is a critical need for non–eye care providers to ensure that eye care has been received by the patient and obtain ophthalmic documentation of DR severity and the eye care plan (7). Generally, patients with no or minimal DR require annual eye examinations, increasing to every 3–4 months or more frequently for patients with vtDR. However, care for a specific patient can vary as a result of issues unique to that patient. Therefore, it is essential that non–eye care providers obtain timely documentation of the ophthalmic follow-up plan from their eye care colleagues.

For these communications to be successful, improved methods to deliver, review, and update medical information between providers who are often located in disparate locations and practices are probably required. Eye care providers must identify DR severity accurately and provide appropriate follow-up recommendations based on the individual patient. These data need to be communicated in a clear and timely manner to all pertinent medical care providers. Interestingly, awareness and follow-up concordance with national guidelines increased with greater eye care provider specialization, remaining statistically significant even after accounting for age, diabetes duration, and DR/DME severity. This finding suggests that different approaches to improve patient awareness and appropriate follow-up may be needed for different provider types.

Eye care information should be reviewed during non–eye care visits, ensuring that actual DR severity is known and that the patient has follow-up consistent with the eye care plan. Otherwise, arrangements should be made as necessary. The non–eye care visit is also an excellent opportunity to reinforce the importance of knowing DR status and maintaining timely follow-up eye care. Indeed, education of patients with DR during endocrinology visits has been reported to improve subsequent control of HbA_1c (31,32), although additional DR education at the time of ophthalmology visits did not (29). Given that low adherence to recommended eye care guidelines has been observed in other ocular diseases, such as glaucoma (33) and age-related macular degeneration (34), understanding how to improve medical knowledge transfer and retention between providers and patients may have broad importance.

This study highlights the effectiveness of retinal telemedicine screening programs to identify gaps in knowledge and care among patients at risk of vision loss resulting from diabetes. Telemedicine programs may provide an entry point for interventions to address these issues. Advances in high-resolution undilated imaging technology and regulatory approval of artificial intelligence methods for automated identification of referable DR are highly promising developments with regard to future scalability and efficiency of such programs.

In our study, provision of adequate ophthalmic care was suggested by the finding that ∼97% of patients with a previous examination were dilated as recommended by national guidelines (30), ∼86% had been evaluated within 2 years, and <6% did not have a prior eye examination. In addition, patients with more advanced eye disease were seen more frequently and by more specialized providers, as would be appropriate according to diabetic eye care guidelines. These data suggest that eye care examinations were generally performed in a manner and frequency suitable for obtaining quality information and providing appropriate care.

In this cohort, regardless of the underlying causes, patient awareness of current DR was low. Investigating changes in patient awareness of DR over time is crucial for addressing disparities in knowledge and may provide insight into the effectiveness of public health strategies contributing to long-term policy development in DR care. Exploratory analysis suggested that during the last 3 years of this 10-year study, statistically significant improvements were seen in awareness among all patients with DR and follow-up concordance with national guidelines. Nevertheless, a vast majority (>85%) of patients with mild DR and nearly 50% of patients with vtDR at the time of retinal imaging still reported no prior diabetic eye disease.

There was an association of increasing awareness, disease severity, and concordant follow-up with greater eye care provider specialization. This association remained statistically significant even after accounting for age, diabetes duration, and DR/DME severity and after sensitivity analysis accounting for all providers whose category could not be determined. These data suggest that approaches to remedy awareness and follow-up deficits may need to be specifically tailored to provider type. The relative effectiveness of establishing DR awareness and timely follow-up among different eye care professional specialists has not been extensively studied.

A limitation of this study is patient self-report, with its inherent potential for recall bias and factual error. Studies have shown that self-reported adherence with dilated eye examination may overestimate the occurrence based on medical record review (35,36) If true, some DR severity could have been underestimated at the patient’s prior examination, despite high reported dilation rates, suggesting that the deficit in concordance with national guidelines may actually have been underestimated in the current study. Furthermore, this study did not assess the accuracy of DR findings from prior examinations and thus cannot identify the underlying causes of current retinopathy unawareness, lack of follow-up concordance with national standards, or association with provider specialization. Patients self-select their physicians based on many factors, including proximity, insurance coverage, knowledge, and referrals from physicians. Lack of patient-reported awareness of DR may reflect patient recall and does not necessarily reflect improper patient care or inadequate discussion by the health care team. Additionally, at the time of last eye examination, severity of DR and follow-up may have been assessed appropriately, with subsequent DR worsening in the interval before study imaging. Self-reported follow-up intervals that were not concordant with best practices could also have resulted from inadequate patient understanding or limited information retention rather than inaccurate ocular assessment or guidance.

In conclusion, despite current medical standards and a population of patients seeking diabetes-specific care at a tertiary medical center, >75% of patients with photographically documented DR reported being unaware of having had any DR, and 55% of patients with vtDR were similarly unaware. Additionally, self-reported ophthalmic follow-up was concordant with national ophthalmic guidelines for only 54% of patients with vtDR. Although this study does not address the underlying reasons for these disparities, it indicates that patient self-report alone cannot currently be used as a foundation from which to determine DR presence or concordance with suggested ophthalmic follow-up. These data suggest a critical need to improve the education of patients with diabetes and eye care professionals, improve the effectiveness of information transfer and retention, and ensure efficient interaction between ophthalmic and medical care providers. Furthermore, it suggests that these efforts may need to differ among various types of eye care providers.

Acknowledgments. The authors acknowledge the support of the Joslin Vision Network Clinical Team (Jessica Rodriguez, Sashida Rodriguez, Cassondra Marcy, Mina Sehizadeh, Bina Patel, Komal Thakore, Jeffrey Ho, and Joseph Kane). L.M.A. is deceased.

Funding. Research support was provided by the Massachusetts Lions Eye Research Fund (L.P.A., J.K.S., P.S.S.) and the Joslin Diabetes Center (Diabetes Research Center Enrichment Core and Clinical Research Center; grant P30DK036836). The JVN technology was developed at the Joslin Diabetes Center, and a single ultrawide field imaging device was provided on temporary loan to the Joslin Diabetes Center by Optos and was used during part of this study.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. P.S.S. researched data and wrote manuscript. J.D.C. researched data, reviewed/edited the manuscript, and contributed to discussion. J.K.S. reviewed/edited the manuscript and contributed to discussion. A.M.T researched data and reviewed/edited the manuscript. D.T. reviewed/edited the manuscript and contributed to discussion. M.J.A. reviewed/edited the manuscript and contributed to discussion. L.M.A. contributed to the initiation and design of the study and contributed to discussion. L.P.A. researched data, reviewed/edited the manuscript, and contributed to discussion. P.S.S. and J.D.C. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Presented in part at the Annual Meeting of the Association for Research in Vision and Ophthalmology, Baltimore, MD, 10 May 2017.

Handling Editors. The journal editors responsible for overseeing the review of the manuscript were Cheryl A.M. Anderson and Jennifer B. Green.