Litvin, Taras V.*; Ozawa, Glen Y.*; Bresnick, George H.†; Cuadros, Jorge A.‡; Muller, Matthew S.§; Elsner, Ann E.∥; Gast, Thomas J.**
The International Diabetes Federation recommends that patients with type 2 diabetes mellitus receive a dilated fundus examination by a qualified provider at the time of diagnosis of diabetes and annually thereafter if no retinopathy is present.1 More frequent retinal examinations are indicated if any retinopathy is present. For patients with type 1 diabetes mellitus, a dilated fundus examination for retinopathy screening is recommended from age 11 years and after 2 years after the diagnosis and annually thereafter.2 The current data indicate that, on average, only 60% of patients with diagnosed diabetes comply with these recommendations.3 Even poorer compliance is reported among patients of lower socioeconomic status.4 This underscores the need for a simple and effective screening tool for the detection of sight-threatening retinopathy because early detection and prompt treatment of retinal disease among diabetic patients can prevent vision loss.5,6
Teleophthalmology screening for diabetic retinopathy (DR) has been shown to be effective in detecting DR in a primary care setting.7 Stereoscopic digital retinal photography with pupil dilation has been validated as an acceptable method for detecting and grading the severity of DR and diabetic macular edema (DME).8,9 The international classification of DR developed by the International Council of Ophthalmology and adopted by the American Academy of Ophthalmology uses the presence and severity of retinal lesion types to stratify the risk of progression to sight-threatening complications from DR.10,11 Several organizations throughout the world, including the Canadian Teleophthalmology Network in Alberta and Inoveon DR screening program in Oklahoma, have implemented DR detection programs using stereoscopic retinal photography and the international classification of DR.12,13
Nonmydriatic retinal cameras have been developed to reduce the discomfort and potential hazards of pupil dilation. However, stereoscopic photography without pharmacological pupil dilation is difficult to perform, and it results in images ungradable for retinal thickening in up to 20% of eyes.14,15 On the other hand, determining retinal thickening in nonstereoscopic images is not possible. Therefore, a number of DR detection programs, such as the Scottish Diabetic Retinopathy Screening Program16 and the Veterans Administration Diabetic Retinopathy Screening Program,17 use the presence and location of hard exudates (HEs) close to the center of the macula as a surrogate to detect and stage DME. There has been only limited validation of HEs as a surrogate for DME. Bresnick et al.18 performed a retrospective analysis of the photographic database of the Early Treatment Diabetic Retinopathy Study (ETDRS) using the criterion of HEs within one disc diameter of the foveola and identified CSME with a sensitivity of 94% and a specificity of 54%. Rudnisky et al.19 reported the sensitivity of HE within two disc diameters of the foveola to be 93.9% in detecting ophthalmoscopically confirmed CSME; the specificity was reported to be 81.6%. Retinal images, in both of these studies, were obtained after pupillary dilation.
The purpose of the present study was to test, using a prospective clinical design, the validity of using HEs located within one disc diameter of the foveola in nonmydriatic, nonstereo, digital retinal images as a criterion for referring diabetic patients suspected of having CSME compared with the standard clinical technique of stereo biomicroscopy using a condensing lens or a contact fundus lens.
This study was conducted at the Alameda County Medical Center in Oakland, CA, a DR screening site within the EyePACS telemedicine network.20 This clinic serves uninsured local patients, and the limited clinic budget did not provide optical coherence tomography (OCT). Adult patients with known type 2 diabetes were recruited for the study. The recruitment process was purposely enriched by patients who were deemed likely to have DR based on a greater than 5-year history of diabetes, elevated HbA1c greater than 9.0, older than 40 years, and self-reported comorbidities such as angina or stroke. Written informed consent for the use and disclosure of protected health information was obtained from all subjects before being enrolled in the study. Institutional review board approval was granted by the Alameda County Medical Center, the University of California, Berkeley, and Indiana University.
The study protocol required two patient visits to the clinic, one for retinal photography alone and the other for a dilated retinal examination. This procedure mimics the actual screening process in the county system where a patient undergoes fundus imaging during the first visit to the clinic, images are reviewed, and the follow-up appointment with the clinician is scheduled based on the outcomes of the image review. The time between the two visits varies according to a number of factors. Such factors include physician and photographer availability issues coupled with reliability and scheduling issues of this indigent population. It was important to mimic that process to assess the utility of HEs as a referral criterion for CSME in a practical setting. If, because of logistical issues, the interval between the first and the second clinic visits exceeded 100 days, the patient was excluded from the study. The actual mean (±SD) interval was 33 ± 31 days. Retinal imaging was performed during the first clinic visit using the nonmydriatic fundus camera CR6-45NM (Canon Inc., Tokyo, Japan) without pupillary dilation. Nonstereoscopic 45-degree images of three retinal fields per eye were obtained in accordance with the EyePACS imaging protocol.21 The primary field included the macula and the optic nerve, the centers of which were located equidistant from the center of the image (default position of the camera; Fig. 1A left). The second field was obtained with the optic nerve at the center of the image (Fig. 1A center). The third field was captured with the optic nerve to the far nasal side of the field, with the macula below and nasal to the center of the picture (Fig. 1A right). An EyePACS image in the primary position is shown with large HEs within one disc diameter of the foveola (Fig. 1B) and an image with smaller HEs (Fig. 1C).
The captured images were uploaded to the EyePACS Web site20 and graded independently for macular edema by two graders (T.J.G. and T.V.L.) according to the EyePACS grading protocol.22 A presumptive diagnosis of clinically significant macular edema (CSME) was made when HEs were noted at or within one disc diameter of the foveola. During the second clinic visit, patients received a dilated fundus examination of the macular region by T.V.L. using a noncontact 90D condensing lens and a biomicroscope. The examiner was masked to the retinal imaging findings. As previously stated, there was typically a delay between imaging and the examination and the patients were not necessarily seen sequentially, lessening the potential bias of results from grading the images to examination results. The presence, extent, and location of retinal thickening were noted, as well as the presence and location of HEs. In cases of uncertainty about the presence of macular edema, Goldmann macular contact lens was also used. The diagnosis of CSME on the dilated fundus examination was made according to the criteria set forth by the ETDRS23: (1) retinal thickening within 500 μm of the center of the macula; or (2) HEs within 500 μm of the center of the macula with adjacent retinal thickening; or (3) retinal thickening of one disc area in size or greater, any part of which is located at or within one disc diameter from the center of the macula.
The sensitivity and specificity values for CSME detection using HEs at or within one disc diameter of the foveola graded in the retinal images were calculated and compared with those for CSME identified during the dilated fundus examination as the standard. The statistical analysis was performed separately for the right and the left eyes.
One hundred forty-three adult diabetic patients were recruited for the study. Forty of these patients were excluded from the study because the time between the first and the second clinic visits exceeded 100 days, reinforcing the scheduling difficulties in this patient population. The mean (±SD) time interval between the first and the second clinic visits of the remaining 103 patients was 33 ± 31 days. Forty-nine percent were females. The mean age of the included patients was 56 ± 17 years. Ethnic composition of the study population is presented in Fig. 2.
For the right eyes, CSME was diagnosed in 16 (15.5%) eyes by biomicroscopy during the dilated examination. Based on retinal images, a presumptive diagnosis of CSME was made independently by the two graders in 28 (26.4%) and in 25 (24.2%) cases, respectively (Table 1). In the right eyes, the sensitivity of HEs located within one disc diameter from the center of the macula as a surrogate for detecting CSME was 93.8% for each of the graders; the specificity values were 88.5 and 85.1%; positive predictive values were 60 and 53.57%; and negative predictive values were 98.7 and 98.7% (Table 2).
For the left eyes, CSME was diagnosed in 16 (15.5%) cases by biomicroscopy during the dilated examination. Based on retinal images, CSME was diagnosed by grader 1 in 26 (25.2%) cases, whereas grader 2 diagnosed CSME in 23 (22.3%) (Table 1). Based on the results of the left eye analysis, the sensitivity values of HEs located within one disc diameter from the center of the macula as a surrogate for detecting CSME were 93.8 and 75% for each of the two graders, respectively; the specificity was 87.4% for both graders; positive predictive values were 57.7 and 52.17%; and negative predictive values were 98.7 and 95%, respectively (Table 2).
When HEs were present at or within one disc diameter from the foveola in undilated nonstereoscopic fundus photos, CSME was detected (determined by a dilated biomicroscopic fundus examination) with good sensitivity, with an acceptable level of overdiagnosis for a screening situation. These results indicate that HEs are a valid surrogate marker for the detection of CSME when stereophotography and OCT are inadequate or simply unavailable for financial reasons.
Having a valid inexpensive alternative for the detection of CSME is important for several reasons. It ensures that DR screenings can successfully detect sight-threatening macular edema without dilation, stereoscopic photos of the macula, or OCT. In addition, DR screenings that rely on nonmydriatic stereoimaging can result in a high proportion of ungradable images for the detection of retinal thickening.14,15 When one image of a macular stereopair is unusable, a surrogate marker for CSME is very useful. The results of this study also suggest that HEs near the center of the macula can be used by primary care providers to screen for CSME using direct ophthalmoscopy, with some consequent overreferral to retinal specialists. We do not suggest that direct ophthalmoscopy is a substitute for a dilated retinal examination.
Our sensitivities are comparable to results published by Bresnick et al.,18 comparing HEs within one disc diameter of the center of the macula versus the then current “gold standard for CSME” graded in ETDRS stereoscopic photographs. Our specificities are somewhat higher than those reported by Bresnick et al.,18 although both sensitivities and specificities are similar to data published by Rudnisky et al.19 in 2006, reporting the ability of HEs within two disc diameters of the fovea to detect CSME that was confirmed by a dilated fundus examination using a retinal contact lens. However, our study is different from these other two studies because we obtained our retinal images undilated, more closely matching the common screening condition of nonmydriatic nonstereo retinal imaging.
With regard to the quality of the grading, the lower sensitivity demonstrated by grader 2 in detection of CSME in the left eyes prompted further investigation. Of the three CSME cases that grader 2 missed, one case showed unmistakable exudates well within one disc diameter of the foveola and therefore may have been a case of data entry error. The other two cases showed a single small exudate at the border of one disc diameter from the foveola. The difference in grading in those cases may be attributed to variability in judgment between graders for the threshold of HE detection by each grader. This highlights the importance of testing grading systems for intragrader and intergrader repeatability and of performing quality control of image grading in DR screening programs.
Low positive predictive values point to a relatively high rate of “overreferral,” although it may be acceptable in a screening setting especially in a patient population with generally poor access to health care. This study shows the utility of the use of HE surrogate for detection of DME by screening programs using nonstereoscopic images. In our opinion, this is important for rapid screening of large diabetic populations with relatively low equipment expense, low false-negative rates, and reasonable false-positive rates. It is better to err on the positive side and send the patient to a retina specialist who does not find serious pathology than miss an actually active disease and lead to blindness on the part of the patient.
This study has limitations. First, the mean (±SD) time interval between imaging and dilated eye examinations was 33 ± 31 days, whereas ideally imaging and examination would have been done at the same visit. In the framework of population screening, it is common to have a screening visit and then a follow-up examination of those subjects showing clinically significant retinopathy so some interval is largely unavoidable. For the specific situation of DME, the ETDRS reported that macular edema tended to be chronic and that spontaneous visual recovery was unusual.23 In addition, a recent study by Kwon et al.24 followed two groups of patients with mild and moderate DME measured by OCT for a period of 6 months without treatment and found no significant changes in the 250- to 300-μm central macular thickness group. There was significant progressive decrease in retinal thickness in the 300- to 500-μm group, but the eyes still were at an average of 318 μm at 1 month and 284 μm at 6 months. Thus, although it is possible that there were some changes in degree of edema during this 1-month period between imaging and examination, it is unlikely that there were significant changes having a major impact on our results. In addition, our findings are similar to those reported by Rudnisky et al.19 In their study, fundus photos were obtained on the same day as a dilated fundus examination and the presence of HEs within 2DD of the fovea was used as a surrogate for CSME detection. They report that HE within 2DD of the foveola has a sensitivity of 93.9% and a specificity of 81.6% for CSME detection. Thus, it may be inferred that any significant changes in retinal thickening occur slowly, and that the elapsed time in our study likely did not have a major impact on our results.
Second, the historically accepted gold standard for detecting CSME is the use of 30-degree, film-based, stereo macular photos performed and graded according to the ETDRS protocol.25 However, our use of the dilated biomicroscopic examination as the standard in the present study is supported by the high correlation reported between CSME detected by contact lens biomicroscopy compared with CSME detection by the ETDRS protocol.26
In addition, further validation studies to compare the HE surrogate for macular thickening with more objective means such as OCT are planned. Although OCT is widely accepted as an objective method for detecting diabetic maculopathy, it is currently too costly and technically challenging to integrate into existing retinopathy detection programs in primary care settings. Low-cost and reliable methods of detecting CSME, such as the use of an HE surrogate marker described here, are needed to meet the challenge of widespread screening for this vision-threatening condition.
Taras V. Litvin
School of Optometry
University of California, Berkeley
200 Minor Hall
Berkeley, CA 94720
This study was supported by grant EY020017 from the National Eye Institute, National Institutes of Health. Involved in design and conduct of study (TVL, MSM, AEE, GYO, GHB); data collection and management of data (TVL, TJG, JAC, MSM); analysis and interpretation of data (TVL, GHB, GYO) and preparation, review, or approval of the manuscript (TVL, GHB, GYO, JAC, AEE, MSM, TJG). We also thank our photographers Lai Pang, Nathan Louie, and Jessica Archambault.
This study was presented in the form of a Scientific Poster (115858) at the annual meeting of the American Academy of Optometry in 2011.
Received July 16, 2013; accepted January 17, 2014.
3. Hazin R, Barazi MK, Summerfield M. Challenges to establishing nationwide diabetic retinopathy screening programs. Curr Opin Ophthalmol 2011; 22: 174–9.
4. Brechner RJ, Cowie CC, Howie LJ, Herman WH, Will JC, Harris MI. Ophthalmic examination among adults with diagnosed diabetes mellitus. JAMA 1993; 270: 1714–8.
5. Han Y, Schneck ME, Bearse MA Jr., Barez S, Jacobsen CH, Jewell NP, Adams AJ. Formulation and evaluation of a predictive model to identify the sites of future diabetic retinopathy. Invest Ophthalmol Vis Sci 2004; 45: 4106–12.
6. Gar S, Davis R. Diabetic retinopathy screening update. Clin Diabetes 2009; 27 (4): 140–5.
7. Andonegui J, Serrano L, Eguzkiza A, Berastegui L, Jimenez-Lasanta L, Aliseda D, Gaminde I. Diabetic retinopathy screening using tele-ophthalmology in a primary care setting. J Telemed Telecare 2010; 16: 429–32.
8. Hubbard LD, Sun W, Cleary PA, Danis RP, Hainsworth DP, Peng Q, Susman RA, Aiello LP, Davis MD, Diabetes C, Complications Trial/Epidemiology of Diabetes I, Complications Study Research G. Comparison of digital and film grading of diabetic retinopathy severity in the diabetes control and complications trial/epidemiology of diabetes interventions and complications study. Arch Ophthalmol 2011; 129: 718–26.
9. Li HK, Danis RP, Hubbard LD, Florez-Arango JF, Esquivel A, Krupinski EA. Comparability of digital photography with the ETDRS film protocol for evaluation of diabetic retinopathy severity. Invest Ophthalmol Vis Sci 2011; 52: 4717–25.
10. Wilkinson CP, Ferris FL 3rd, Klein RE, Lee PP, Agardh CD, Davis M, Dills D, Kampik A, Pararajasegaram R, Verdaguer JT, Global Diabetic Retinopathy Project Group. Proposed international clinical diabetic retinopathy and diabetic macular edema disease severity scales. Ophthalmology 2003; 110: 1677–82.
12. Ng M, Nathoo N, Rudnisky CJ, Tennant MT. Improving access to eye care: teleophthalmology in Alberta, Canada. J Diabetes Sci Technol 2009; 3: 289–96.
13. Fransen SR, Leonard-Martin TC, Feuer WJ, Hildebrand PLThe Inoveon Health Research Group. Clinical evaluation of patients with diabetic retinopathy: accuracy of the Inoveon diabetic retinopathy-3DT system. Ophthalmology 2002; 109: 595–601.
14. Ahmed J, Ward TP, Bursell SE, Aiello LM, Cavallerano JD, Vigersky RA. The sensitivity and specificity of nonmydriatic digital stereoscopic retinal imaging in detecting diabetic retinopathy. Diabetes Care 2006; 29: 2205–9.
15. Bursell SE, Cavallerano JD, Cavallerano AA, Clermont AC, Birkmire-Peters D, Aiello LP, Aiello LM Joslin Vision Network Research Team. Stereo nonmydriatic digital-video color retinal imaging compared with Early Treatment Diabetic Retinopathy Study seven standard field 35-mm stereo color photos for determining level of diabetic retinopathy. Ophthalmology 2001; 108: 572–85.
17. Cavallerano AA, Conlin PR. Teleretinal imaging to screen for diabetic retinopathy in the Veterans Health Administration. J Diabetes Sci Technol 2008; 2: 33–9.
18. Bresnick GH, Mukamel DB, Dickinson JC, Cole DR. A screening approach to the surveillance of patients with diabetes for the presence of vision-threatening retinopathy. Ophthalmology 2000; 107: 19–24.
19. Rudnisky CJ, Tennant MT, de Leon AR, Hinz BJ, Greve MD. Benefits of stereopsis when identifying clinically significant macular edema via teleophthalmology. Can J Ophthalmol 2006; 41: 727–32.
20. EyePACS, LLC. Eye Picture Archive Communication System; 2010. Available at: www.eyepacs.com
. Accessed January 3, 2014.
22. Cuadros J, Bresnick G. EyePACS: an adaptable telemedicine system for diabetic retinopathy screening. J Diabetes Sci Technol 2009; 3: 509–16.
23. Early Treatment Diabetic Retinopathy Study Research Group. Photocoagulation for diabetic macular edema. Early Treatment Diabetic Retinopathy Study report number 1. Arch Ophthalmol 1985; 103: 1796–806.
24. Kwon SI, Baek SU, Park IW. Comparison of natural course, intravitreal triamcinolone and macular laser photocoagulation for treatment of mild diabetic macular edema. Int J Med Sci 2013; 10: 243–9.
25. Early Treatment Diabetic Retinopathy Study Research Group. Grading diabetic retinopathy from stereoscopic color fundus photographs—an extension of the modified Airlie House classification. ETDRS report number 10. Ophthalmology 1991; 98: 786–806.
26. Kinyoun J, Barton F, Fisher M, Hubbard L, Aiello L, Ferris F 3rdThe ETDRS Research Group. Detection of diabetic macular edema: ophthalmoscopy versus photography—Early Treatment Diabetic Retinopathy Study Report Number 5. Ophthalmology 1989; 96: 746–50.