Nonmydriatic Retinal Image Review at Time of Endocrinology Visit Results in Short-Term HbA1c Reduction in Poorly Controlled Patients with Diabetic Retinopathy

Introduction

Poor control of diabetes mellitus (DM), as reflected by elevated HbA1c, is a risk factor for development and progression of diabetic retinopathy (DR) in both type 1 and type 2 DM. ^{1 –3} Efforts at improving blood glucose control, including pharmacologic combination therapy, lifestyle changes, and reduction of body weight, have been the subject of extensive study. ^{4 –7} A comprehensive team approach comprising medical, educational, and behavioral components may help to optimize diabetes control and thereby reduce the development of complications. ⁸ Nonmydriatic retinal imaging approaches have been documented to provide accurate indication of DR severity. ⁹

We evaluated the effect of personalized diabetes education on DM control by discussing retinal findings after nonmydriatic retinal imaging at the time of an endocrinologist examination in patients with DR and poorly controlled DM and subsequently assessing change in HbA1c. To our knowledge, no previous study has integrated nonmydriatic retinal imaging as an educational tool to help patients understand the gravity of their complications during the course of an endocrine visit and assessed the effect on subsequent glycemic control independent of ophthalmic intervention.

Materials and Methods

The study protocol was approved by the Institutional Review Board of the American University of Beirut Medical Center (AUBMC) and was in accordance with the Helsinki declaration. Written, informed consent was obtained for all subjects. The study was registered at the ACTRN Web site with trial registration number ACTRN 12609000362291. This prospective, controlled trial enrolled 120 diabetic subjects seen by a single endocrinologist (I.S) in the AUBMC diabetes clinic. Inclusion criteria were age 18 years or older, type 1 or type 2 DM, as defined by the Expert Committee on the Diagnosis and Classification of Diabetes Mellitus, ¹⁰ documented poor glycemic control (HbA1c ≥8.0%) within 1 month of baseline, and DR documented in the medical record (early treatment diabetic retinopathy study [ETDRS] level>20). ¹¹ Exclusion criteria included gestational DM, media clarity or pupil size inadequate for nonmydriatic retinal imaging, and poor visual acuity (ETDRS VA <20/100). Sample size was chosen to provide 95% power to detect a 1% difference in mean HbA1c between the groups with an alpha error level of 5%, which would require a total sample size of 112. Additional subjects were recruited to account for potential loss to follow-up.

All eligible patients presenting to the clinic were approached for study participation. HbA1c was performed at the AUBMC laboratory on the day of enrollment to confirm a value of >8.0%. Consecutively enrolled subjects were alternately assigned to the intervention group or control group. The “intervention” group had nonmydriatic retinal imaging following their endocrinologist examination performed by a single technician, followed by viewing of the images and a brief discussion by a single retina specialist (H.S.) regarding the findings. The retina specialist showed the subjects their images on a computer monitor highlighting clinical findings. Discussion explicitly included (1) presentation demonstrating a range of DR including a normal fundus, mild-moderate nonproliferative DR, active proliferative DR, and tractional retinal detachment (Fig. 1); (2) a brief summary of the results of the Diabetes Control and Complications Trial (DCCT) and the United Kingdom Prospective Diabetes Study (UKPDS) regarding HbA1c control and DR; and (3) emphasis on importance of blood glucose control to delay onset or slow progression of DR. The “control” group did not undergo either retinal imaging or discussion of retinal findings by the retinal specialist. The endocrinologist was masked to whether the subjects were assigned to the intervention or control group during care and follow-up. Three months after the endocrinologist visit, HbA1c measurement was repeated for each subject.

OPEN IN VIEWER

Fig. 1.

The photographic set shown to the intervention group illustrating the spectrum of retinopathy from very mild nonproliferative diabetic retinopathy to tractional retinal detachment.

Results

Of the 120 recruited subjects, 113 (94%) completed the study (59 [98%] from the intervention group and 54 [90%] from the observation group). Three subjects were dropped, because the confirmatory HbA1c was <8.0% (one in the intervention group and two in the control group). Four subjects could not be reached for follow-up at study closure (none in the intervention and four in the control groups). Table 1 shows baseline characteristics of the cohort. There was no statistical difference in mean age, HbA1c, DM duration, or presence of hypertension between the intervention and control groups at baseline. DR levels were similar for both groups.

Of the 113 subjects who completed the study, 62% (70/113) had improved HbA1c at 3 months. Figure 2 shows the baseline and 3-month HbA1c for each subject identified by DR level for both the intervention and control groups. A total of 78% (46/59) had HbA1c improvement in the intervention group compared with 44% (24/54) in the nonintervention group (p=0.0002). The mean±standard deviation for HbA1c decreased by 1.35%±1.62% in the intervention group, ranging from a decrease of 5.9% to an increase of 2.7%. In the control group, the HbA1c increased by 0.26%±1.59%, ranging from a decrease of 2.5% to an increase of 4.1%. The difference between the intervention and control groups was statistically significant (p<0.0001). Controlling for gender, age, duration of disease, hypertension, and use of insulin, the difference between groups remained significant (p=0.0002).

OPEN IN VIEWER

Fig. 2.

HbA1c taken at study entry (0 months) and at study closure (3 months) in the intervention (right) and nonintervention (left) groups.

Because of limited numbers of certain DR severity levels, the groups were evaluated in terms of those who had mild-to-moderate NPDR (level 20–43: intervention 35/59, control 32/54) compared with those with more severe DR (Table 2). In this subpopulation analysis, there was a significant reduction in HbA1c at 3 months in the intervention group regardless of DR severity. In contrast, the nonintervention group had increases in HbA1c at 3 months that were statistically significant in the mild-moderate group. The severity of DR did not appear to have substantial impact within either of the study groups.

Discussion

Vision loss from DR remains a major problem in all developed countries. It has been well established that improved glycemic control can markedly reduce the risk of onset and progression of DR. ^1,3 In addition, studies have demonstrated that vision loss is one of the most feared of all complications from DM. ^12,13 We postulated that a single intervention at the time of endocrinology evaluation using a subject's retinal fundus images and brief education directly associated with these images might be effective in improving DM control as measured by subsequent HbA1c. Digital imaging technology allows retinal image acquisition at the time of a patient's endocrinologist visit, which is feasible and accurate. ¹⁴ Our results indicate that 3 months after an endocrinology visit with associated retinal imaging and brief retinopathy education, the intervention group HbA1c was reduced by 13.4%, whereas the observation group HbA1c increased 2.7%. The improvement in HbA1c after retinal education was evident even after correcting for gender, age, duration of disease, hypertension, and the use of insulin (p=0.0002). We did not address treatment modalities separately in this study, because we were evaluating DM control and not the risk of progression of DR.

Our data suggest that during the endocrinology visit a better understanding of how diabetes affects the eye and vision may have a substantial impact on glycemic control over the next 3 months in patients with DR and poor glycemic control. As the fear of vision loss is so pervasive, it is likely that this discussion may alter patient behavior at least for the short term. In addition, the use of nonmydriatic digital imaging allows patients to see any changes in their own retinas, personalizing the experience and perhaps providing additional impetus for subsequent improvement in glycemic control. Additional benefits include a simple, noninvasive, and cost-efficient intervention requiring minor additional effort from the diabetes care provider team. The imaging is also well tolerated, ¹⁵ because, unlike traditional dilated fundus examinations, vision recovers rapidly after the nonmydriatic imaging, allowing patients to view their own images during the subsequent discussion.

Fonda and associates reported a retrospective decrease of HbA1c in subjects who had retinal imaging using the Joslin Vision Network telehealth program. ¹⁶ Benefits were also found with regard to blood pressure control and LDL levels when compared with traditional eye care follow-up. This study differs from the present report in that it was performed retrospectively and as part of a telehealth program wherein subjects were referred for special ophthalmic evaluation according to the severity of detected disease. The results of our present study demonstrate that the imaging and educational experience alone can have a substantial impact on near-term glycemic control.

Klein and associates have demonstrated that younger onset insulin-dependent subjects with overt signs or symptoms from complications of DM (proteinuria or decrease in vision from DR) experience a more significant drop in HbA1c with time, compared with subjects who are asymptomatic. ¹⁷ We observed similar results in subjects with DR who are not necessarily symptomatic from their eye disease. Although the treatment approach for type 1 and type 2 DM is different and HbA1c goals may vary, both the DCCT and the UKPDS showed a beneficial effect of intensive control (decrease of HbA1c by 2% and 1%, respectively). Thus, we did not limit our study to either type 1 or type 2 DM alone, although different outcomes with type 1 versus type 2 DM would be interesting to evaluate in subsequent studies.

Also of note is the fact that, in our present study, both the control and intervention groups did not undergo any specific additional counseling for other DM complications and there was no second encounter with another provider. It would be important to know whether directed education regarding some of the other DM complications would have a similar and possibly additive effect. This study did not address this issue and was not designed to determine the relative contributions of the subjects observing their own retinal images when compared with only seeing the illustration of the spectrum of disease severity combined with physician discussion. Although there was a substantial effect on reduction of HbA1c, the study also did not address the particular behaviors that were changed by this intervention. Further, the study only evaluated outcomes at 3 months. It is possible that without additional intervention, this beneficial effect may be lost over time. Even with additional intervention there is the possibility that patients may become tolerant to the education and have less future benefit. Conversely, the encounter may have led to sustained or even further reduction in HbA1c.

Nevertheless, the substantial reduction observed in HbA1c, and the significant associated reduction in morbidity that would occur if this benefit could be maintained, clearly supports additional studies focusing on this type of intervention.

It should be noted that this imaging approach does not replace a comprehensive retinal ophthalmologic examination. The goal was to evaluate the impact of reviewing retinal images at the time of endocrinology visit; thus, key aspects of a comprehensive ocular examination, including visual acuity, were intentionally not measured as part of the intervention, because most endocrinology practices would not have the facilities, equipment, or trained personnel required to perform quality refractions and best-corrected visual acuity.

Given the effect observed in this study, it is evident that even limited focused education may have a beneficial impact. In type 1 DM, the DCCT demonstrated that for each 10% decrease in HbA1c from baseline there is a 39% decrease in risk of progression of DR. ¹⁸ We demonstrated an overall 13.4% reduction in HbA1c equating to more than a 52% reduction in the risk of development or worsening of DR. In type 2 DM, the UKPDS showed a 35% reduction in risk of onset or progression of DR when the HbA1c decreased by 1 percentage point. ¹⁹ In our series, dropping the HbA1c by 1.35 percentage points would equate to more than a 47% reduced risk of onset or progression of DR. The potential benefit to the worldwide population of patients with DM could therefore be enormous. Further studies to rigorously evaluate the most effective and most durable approach to such an intervention are warranted.