Detection of Drusen and Early Signs of Age,related Maculopathy Using a Nonmydriatic Camera and a Standard Fundus Camera Ronald Klein, MD, MPH, Stacy M. Meuer, BS, Scot E. Moss, MA, Barbara E. K. Klein, MD, MPH Purpose: The study was designed to compare the severity of age-related maculopathy as graded from photographs taken using three different techniques. Methods: Two methods of nonstereoscopic 45 0 retinal photography of the macula (through a non pharmacologically dilated pupil and through a pharmacologically dilated pupil) were compared with results from standard 30 0 stereoscopic photographs in 112 subjects. Corresponding photographic fields were graded by a masked grader for the presence of any drusen, soft drusen, retinal pigment epithelial degeneration, increased retinal pigmentation, and early and late age-related maculopathy. Results: Exact agreement between gradings of the 45 0 photographs taken through non pharmacologically dilated pupils and 30 0 photographs taken through dilated pupils was 75% for any drusen, 72% for soft drusen, 72% for retinal pigment epithelial degeneration, 74% for increased retinal pigment, 85% for pure geographic atrophy, and 89% for exudative macular degeneration. The kappa scores varied from 0.33 for geographic atrophy to 0.60 for exudative macular degeneration. Slightly higher rates of agreement between gradings were found after dilation. Conclusion: These data suggest that 45° non stereoscopic fundus photographs, when graded according to a standard classification scheme, should be considered for detection of age-related maculopathy in situations where the pupils cannot be pharmacologically dilated and retinal specialists are not available to examine the fundus. Ophthalmology 1992;99: 1686-1692
Fundus photography is an objective method of recording retinal pathology (or its absence) that has been useful in screening,1,2 routine care, 3 epidemiologic studies,4-6 and
Originally received: April 20, 1992. Manuscript accepted: May 18, 1992. From the University of Wisconsin, Department of Ophthalmology, Madison. Supported by grant 5 U IO-EY06594 from the National Eye Institute/ NIH, Bethesda, Maryland (Drs. R. Klein and B. E. K. Klein). The authors have no proprietary interest in the development or marketing of the camera used in this study or competing cameras. Reprint requests to Ronald Klein, MD, MPH, University of WisconsinMadison, Department of Ophthalmology, 600 Highland Ave, E5/353 esc, Madison, WI 53792.
1686
clinical trials. 7- 1O Optimaily, 30° color stereoscopic fundus photographs, taken through pharmacologically dilated pupils, are used. However, there are situations in which it is not possible to dilate the pupils with pharmacologic agents or when costs prohibit use of standard 30° cameras. II - 14 In these circumstances, a "nonmydriatic" camera has been used as an alternative means of documenting retinal pathology. Data from earlier studies suggest that evaluation of 45° nonstereoscopic fundus photographs by experienced graders provides a reasonably reliable measure of diabetic retinopathy and is a more sensitive and objective method of detecting retinopathy than direct ophthalmoscopy. 15, 16 No data, however, are available to evaluate the reliability of these methods for detecting age-related maculopathy. This is important, as age-related maculopathy is a leading
Klein et al . Detection of Drusen and ARM cause of visual impairment in older Americans. 17 The purpose of this report is to compare the presence of drusen and signs of early and late age-related maculopathy as determined: (1) by grading of nonstereoscopic 45° retinal photographs taken with the nonmydriatic camera through a nonpharmacologically dilated pupil; and (2) by grading of nonstereoscopic 45° retinal photographs taken with the nonmydriatic camera through a pharmacologically dilated pupil versus grading of stereoscopic retinal photographs taken with a standard 30° camera through a pharmacologically dilated pupil.
Methods Population The study population consisted of subjects photographed at two sites. One group consisted of 80 diabetic subjects (mean age, 61 years) participating in a study to evaluate photography using different cameras as a method of screening for diabetic retinopathy (unpublished data; Pugh et al). The second group consisted of 32 patients with agerelated macular degeneration who were regularly seen in the Retina Clinic at the University of Wisconsin Hospital and Clinics. None had diabetes mellitus. Their mean age was 74 years. The nonmydriatic camera (Canon CR3, Lake Success, NY), which uses infrared light to televise a view of the fundus, was used to photograph the retina of the nonpharmacologically dilated eye of each subject. One photograph was taken of a field centered horizontally and vertically on a point midway between the temporal edge of the optic disc and the fovea. The 45° image produced included an area above and below the temporal arcades and areas just nasal to the disc and temporal to the macula (Fig 1). This image is minified 0.64X (at 0 diopters) compared with that taken with a standard 30° camera (Fig 2) (Zeiss FF3, Carl Zeiss, Thornwood, NY).
Figure 1. Forty-five degree fundus photograph taken with the nonmydriatic camera through nonpharmacologically dilated pupil shows standard field taken for study.
Figure 2. Thirty degree fundus photograph taken of Diabetic Retinopathy Study Field 2. Notice the relative magnification of the image compared with the 45° photograph.
One drop of 2.5% phenylephrine and one drop of 1% tropicamide were then instilled in the cul-de-sac of each eye to obtain dilation. Another photograph of the same retinal field was taken with the 45° camera in 103 of the 112 subjects. Shortly thereafter, stereoscopic color retinal photographs were taken of the macula with a standard 30° (Zeiss) fundus camera (Diabetic Retinopathy Study [DRS] Standard Field 2). All photographs for a given subject were taken with either color Kodachrome 25 or Ektachrome 100 slide film and were processed and returned as 2 X 2 inch color slides. The slides were mounted in clear plastic mounting sheets and graded at the University of Wisconsin by a single experienced masked grader. First, nonmydriatic camera photographs taken through a nonpharmacologically dilated pupil were graded, then (at least 1 week later) nonmydriatic photographs taken through a pharmacologically dilated pupil were graded, and finally (at least 1 week later) photographs taken with the standard 30° camera were graded. The grader, masked to subject information, used the National Health and Nutrition Examination Survey III grading protocol, described in detail elsewhere. 14 Briefly, after assessing the quality of the photographs (clarity, field definition, presence of artifacts, and the ability to grade lesions), the severity of specific lesions associated with diabetic retinopathy was determined using a modification of the Airlie House Classification. 7 ,18,19 Next, the presence of any drusen, soft drusen, retinal pigment epithelial degeneration, increased retinal pigment, geographic atrophy, and signs of exudative macular degeneration (subretinal hemorrhage, subretinal fibrous scar, retinal pigment epithelium detachment and/or serous detachment of the sensory retina) were determined using a modification of the Wisconsin Age-Related Maculopathy grading system. 20 ,21 The location oflesions within the temporal retinal vascular arcades in field 2 was assessed. To examine between-grader variation, a second experienced grader graded 44 sets of 45° and 30° photographs from the orig-
1687
Uphthalmology
VoLume 99, Number 11, November 1992 photographs: 45° photographs taken with the non mydriatic camera through a nonpharmacologically dilated pupil; and 30° photographs taken through a dilated pupil. Exact agreement was 75.0% (84 of 112). The unweighted kappa score was 0.60 (standard error [SE], 0.06). Gradings of 45 ° photographs and 30° stereoscopic photographs taken through pharmacologically dilated pupils were in agreement 89.3% (92 of 103) of the time. The unweighted kappa score was 0.82 (SE, 0.07). There was a greater likelihood of "underestimating" early age-related maculopathy when grading 45° photographs compared to when grading 30° stereoscopic photographs. Of the 45° photographs taken through a nonpharmacologically dilated pupil, 9.8% (11 of 112) could not be graded. Of the 45° photographs taken through the pharmacologically dilated pupil, 1.9% (2 of 103) could not be graded. Of the 112 30° photographs taken with the standard camera through a dilated pupil, drusen were found in the macular area in 87, soft drusen in 42, retinal pigment epithelial degeneration in 25, increased retinal pigment in 29, exudative macular degeneration in 11, and geographic atrophy in 7. Analyses were performed to compare gradings of photographs for specific age-related maculopathy lesions. The results are summarized in Table 2. Exact agreement between gradings of photographs taken with the standard 30° camera ranged from 72% to 89% compared with gradings of 45° photographs taken through nonpharmacologically dilated pupils and ranged from 79% to 98% compared with gradings of 45 ° photographs taken through pharmacologically dilated pupils. The kappa scores were generally in the moderate to substantial agreement cate-
inal set using the same grading procedure. To examine within-grader variation, the original grader regraded 20 sets of 45° and 30° photographs after an interval of 6 months.
Definitions Early age-related maculopathy was defined as either the presence of soft drusen or the presence of any drusen type with retinal pigment epithelial degeneration or increased retinal pigment in the macular area in the absence of late age-related maculopathy as defined below. 6 Late age-related maculopathy was defined as the presence of signs of exudative age-related macular degeneration or pure geographic atrophy.
Statistics Wisconsin Storage and Retrieval, a data processing software system, was used for processing all subject files. 22 Percentages of exact agreement were calculated, and unweighted kappa scores were computed. 23 For interpreting kappa values, we used guidelines set forth by Landis and Koch,23 which describe the relative strength of agreement associated with different values of kappa: fair, 0.21 to 0.40; moderate, 0.4 I to 0.61; substantial, 0.61 to 0.80; and almost perfect, 0.81 to 1.00.
Results Table 1 compares severity of age-related maculopathy (none, early, or late) determined by grading two types of
Table 1. Comparison of Severity of Age-related Maculopathy by Grading Photographs Taken with Both a Nonmydriatic 45 0 Camera and a Standard 30 0 Camera Standard 30° Camera (dilated pupil) Nonmydriatic 45° Camera Nonpharmacologically Dilated Pupil No age-related maculopathy Early age-related maculopathy Late age-related maculopathy Cannot grade Total Pharmacologically Dilated Pupil No age-related maculopathy Early age-related maculopathy Late age-related maculopathy Cannot grade Pictures not taken Total
1688
No Age-related Maculapathy
Early Age-related Maculapathy
Late Age-related Maculapathy
47
12
0
3 0
22
2 15
0
7 57
4
0
38
17
52 2
7 24
0 0
0
0
16
1 2 57
1 6
0
38
1 17
Total
59 27 15 11 112 59 26 16 2 9 112
Klein et al . Detection of Drusen and ARM Table 2. Percent of Eyes with Agreement between Gradings of Specific Age-related Maculopathy Lesions 45° Photographs Taken Through Nonpharmacologically Dilated Pupil and 30° Photographs Taken Through Pharmacologically Dilated Pupil
Characteristic Any drusen Soft drusen Retinal pigment epithelial degeneration Increased retinal pigment Exudative macular degeneration Geographic atrophy SE
=
45° Photographs and 30° Photographs Taken Through Pharmacologically Dilated Pupil
Exact Agreement (%)
Kappa (SE of Kappa)
Exact Agreement (%)
(SE of Kappa)
74.9 72.3
0.42 (0.08) 0.50 (0.07)
78.6 79.6
0.44 (0.09) 0.62 (0.08)
72.3 74.1 89.3 84.8
0.39 (0.07) 0.42 (0.07) 0.60 (0.07) 0.33 (0.06)
85.5 82.5 98.0 97.1
0.54 0.51 0.91 0.79
Kappa
(0.08) (0.08) (0.09) (0.09)
standard error.
gories. There was a greater likelihood of "underestimating" the presence of soft drusen (31.3% versus 37.5%), retinal pigment epithelial degeneration (13.4% versus 22.3%), and increased retinal pigment (15.2% versus 25.9%) when grading 45° photographs taken through a nonpharmacologically dilated pupil compared with grading 30° stereoscopic photographs. There were no photographs taken with the 45 0 camera in which the grader misclassified drusen and hard exudate or soft exudate as determined by grading 30° photographs. Between- and within-grader agreement for detection of specific age-related maculopathy lesions from photographs taken with the 45 ° camera (through a nonpharmacologically dilated and a pharmacologically dilated pupil) and/or the 30° camera are presented in Tables 3 and 4. The kappa scores were generally in the moderate to substantial agreement categories. The quality of the photographs was judged to be fair to good in 81 (72.3%), borderline in 26 (23.2%), and poor in 5 (4.5%) of the 45° photographs taken through a nonpharmacologically dilated pupil. All 112 30° stereoscopic photographs were considered to be offair to good quality. Focus problems were present in 37 (33.0%) and illumination problems were present in 20 (17.9%) of the 112 45° photographs taken through a non pharmacologically dilated pupil. Agreement with 30° stereoscopic photographs was significantly higher (P < 0.05) for 45° photographs taken through a nonpharmacologically dilated pupil that were of fair to good quality than for those of borderline to poor quality for the following lesions: presence of drusen (82.7% versus 58.0%), soft drusen (77.7% versus 58.0%), retinal pigment epithelial degeneration (80.2% versus 51.6%), increased retinal pigment (81.4%
versus 54.8%), exudative macular degeneration (98.7% versus 64.5%), and geographic atrophy (85.1 % versus 54.8%).
Discussion Our data suggest that photographs taken through a nonpharmacologically dilated pupil with a nonmydriatic camera may be of value in the detection of age-related maculopathy in some epidemiologic studies. However, if the objective of the study is to use age-related maculopathy as an outcome variable for a controlled clinical trial or to reliably document the appearance of specific age-related maculopathy lesions, grading of stereoscopic 30 0 fundus photographs would be the first choice. Photographs taken with either camera would probably be oflimited value in the routine screening of asymptomatic healthy older patients in areas where ophthalmologic care is limited because of the relatively low prevalence of eyes with exudative macular degeneration that would benefit from laser ablation of extrafoveal subretinal new vessels. 6 ,24,25 Our data suggest that while agreement between gradings of fundus photographs taken with different cameras is very good for detection of drusen and late age-related macular degeneration, the 45 ° nonstereoscopic fundus photographs are less sensitive as a means of detecting early signs of age-related maculopathy, such as the presence of pigment epithelial abnormalities. This is not surprising, as the latter may be subtle, requiring excellent focus, magnification, and stereopsis to detect. In addition, in some cases increased retinal pigment may become more ap-
1689
Ophthalmology
Volume 99, Number 11, November 1992
Table 3. Percent of Agreement for Age-related Maculopathy Lesions and Age-related Maculopathy Severity between Two Graders for 44 Eyes 45° Photographs Taken Through Nonpharmacologically Dilated Pupil
Exact Agreement Characteristic
(%)
Kappa (SE of Kappa)
Any drusen Soft drusen Retinal pigment epithelial degeneration Increased retinal pigment Exudative macular degeneration Geographic atrophy Overall age-related maculopathy severity"
84.1 79.5 72.7 86.3 88.6 79.5 90.9
45° Photographs Taken Through Pharmacologically Dilated Pupil
Exact Agreement (%)
Kappa (SE of Kappa)
0.68 (0.12) 0.68 (0.11) 0.54 (0.10) 0.74 (0.12) 0.77 (0.11) 0.54 (0.10)
81.8 75.0 77.2 77.2 95.4 86.3
0.87 (0.09)
84.1
30° Photographs Taken Through Pharmacologically Dilated Pupil
Exact Agreement (%)
Kappa (SE of Kappa)
0.58 (0.13) 0.59 (0.11) 0.56 (0.12) 0.49 (0.13) 0.89 (0.13) 0.55 (0.13)
88.6 72.7 77.2 88.6 100.0 95.4
0.71 (0.15) 0.54 (0.11) 0.59 (0.12) 0.79 (0.13) 1.00 (0.15) 0.83 (0.15)
0.77 (0.10)
93.2
0.90 (0.11)
SE = standard error . • None, early, or late.
parent in one member of the stereoscopic 30° photograph pair because of variations in illumination. There are few data available regarding the ability to detect signs of age-related maculopathy using ophthalmoscopy. In the pilot phase of the Visual Acuity Impairment Survey, Sperduto et aJ26 reported exact agreement of84.2% (112 of 133; kappa, 0.44; SE, 0.09) for detection of signs of macular degeneration between ophthalmologists using ophthalmoscopy and masked ophthalmologists grading stereoscopic fundus photographs. In the Framingham Eye Study, retinal drusen and pigmentary abnormalities were significantly (P < 0.05) less likely to be detected by the examiners (third year ophthalmology residents or staff ophthalmologists) compared with detection by grading of stereoscopic fundus photographs. However, agreement between examiners and graders was similar for detection of exudative macular degeneration. 25 In the Table
Beaver Dam Eye Study, the exact agreement for detection of soft drusen by trained nonophthalmologist examiners using direct and indirect ophthalmoscopy and graders examining stereoscopic 30° fundus photographs taken with a standard fundus camera was 85.0% (kappa, 0.34; SE, 0.02) (Klein R, unpublished data). These data suggest that grading of 45° fundus photographs, taken with the nonmydriatic camera, may be of similar sensitivity for detecting signs of age-related maculopathy as ophthalmoscopy done by non ophthalmologists or ophthalmologists. However, fundus photography has several advantages. First, changes over time can be more readily recognized. In addition, the reliability of these assessments can be evaluated by replicate gradings. The quality of the photographs and gradings can be monitored, and remedial action can be taken as needed. Graders can be masked with respect to subject characteristics. Photographs also
4. Percent of Agreement for Age-related Maculopathy Lesions within Grader for 20 Eyes 45° Photographs Taken Through Nonpharmacologically Dilated Pupil
45° Photographs Taken Through Pharmacologically Dilated Pupil
Characteristic
Exact Agreement (%)
Kappa (SE of Kappa)
Exact Agreement (%)
Any drusen Soft drusen Retinal pigment epithelial degeneration Increased retinal pigment
90.0 85.0 85.0 70.0
0.80 (0.17) 0.74 (0.17) 0.71 (0.14) 0.40 (0.13)
90.0 85.0 95.0 95.0
SE
=
standard error.
1690
Kappa (SE of Kappa)
0.73 0.73 0.83 0.81
(0.22) (0.20) (0.19) (0.17)
30° Photographs Taken Through Pharmacologically Dilated Pupil
Exact Agreement (%)
Kappa (SE of Kappa;
90.0 90.0 85.0 85.0
0.69 (0.22) 0.80 (0.19) 0.70 (0.18) 0.73 (0.19)
Klein et al . Detection of Drusen and ARM permit an opportunity to add new items to the grading scheme. One concern when grading 45° nonstereoscopic fundus photographs is the misclassification oflesions such as retinal drusen and hard exudate. Such misclassification may result in erroneous conclusions regarding relationships of prevalence of hard exudate or in decisions to refer such patients for evaluation of possible macular edema. Our data suggest that this may not be a problem, as we found no such misclassification. This may have resulted from recognition of differences of the lesions in the 45° photographs by the experienced grader. The sharper margins, the pattern, and the juxtaposition to retinal microaneurysms are suggestive of the appearance of hard exudate material in contrast to individual round flat spots, pale yellowish to white in color, which are typical of small drusen. Similarly, there was no misclassification of soft drusen as soft exudates. There were fewer ungradable 45° fundus photographs (9.8% of those taken through a nonpharmacologically dilated pupil and 1.9% through a pharmacologically dilated pupil) compared with an earlier study (12.7% and 6.8%, respectively), which suggests an improvement in the technology and/or the photographers. IS In our study, better photographic quality was associated with increased ability to detect the lesions associated with age-related maculopathy. These data suggest that the optimal method of detecting the presence and severity of age-related maculopathy is grading of 30° stereoscopic fundus photographs. In older people, a higher frequency of poorer quality photographs taken through a nonpharmacologically dilated pupil is expected because of the high prevalence of media opacities such as cataract and a decrease in the ability of the pupil to dilate. In the Beaver Dam Eye Study, people 75 years of age or older had a higher frequency of late cataract (52.2% versus 0.5%) and smaller pupils after pharmacologic dilation (less than 5 mm in diameter) (9.6% versus 1.4%) than people 43 to 54 years of age.27 For these reasons, pharmacologic dilation of the pupil is important in studies of age-related maculopathy in people 75 years of age or older. In summary, the use of the nonmydriatic camera for detection of age-related maculopathy should only be considered in situations where the pupils cannot be pharmacologically dilated and retinal specialists are not available to examine the fundus. In these situations, it may offer a more sensitive and objective method of detecting maculopathy than direct ophthalmoscopy. Acknowledgment. The authors thank Jacqueline A.
Pugh, MD, for permitting them to use fundus photographs taken in her study, which was funded by the V.A. Health Services Research and Development Office.
References 1. Javitt JC, Canner JK, Sommer A. Cost effectiveness of current approaches to the control of retinopathy in type I diabetics. Ophthalmology 1989;96:255-64.
2. Dasbach EJ, Fryback DG, Newcomb PA, et al. Cost-effectiveness of strategies for detecting diabetic retinopathy. Med Care 1991;29:20-39. 3. American Academy of Ophthalmology Quality of Care Committee, Retina Panel. Diabetic Retinopathy. Preferred Practice Pattern. San Francisco: The Academy, 1989. 4. Klein R. Retinopathy and other ocular complications in diabetes. In: Olefsky JM, Sherwin R, eds. Diabetes Mellitus: Management and Complications. New York: Churchill Livingstone, 1985; 10 1-58. 5. Klein R, Klein BEK, Moss SE, et al. The Wisconsin Epidemiologic Study of Diabetic Retinopathy.II.Prevalence and risk of diabetic retinopathy when age at diagnosis is less than 30 years. Arch Ophthalmol 1984;102:520-6. 6. Klein R, Klein BEK, Linton KLP. The prevalence of agerelated maculopathy. The Beaver Dam Eye Study. Ophthalmology 1992;99:933-43. 7. Diabetic Retinopathy Study Research Group. Report 7. A modification of the Airlie House classification of diabetic retinopathy. Invest Ophthalmol Vis Sci 1981 ;21 :210-26. 8. 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. 9. Davis MD, Hubbard LD, Trautman J, Klein R. Studies of retinopathy. Methodology for assessment and classification with fundus photographs. Diabetes 1985;34(SuppI3):42-9. 10. The Diabetes Control and Complications Trial Research Group. Color photography vs fluorescein angiography in the detection of diabetic retinopathy in the Diabetes Control and Complications Trial. Arch OphthalmoI1987;105:134451. 11. Jones D, Dolben J, Owens DR, etal. Non-mydriatic Polaroid photography in screening for diabetic retinopathy: evaluation in a clinical setting. Br Med J 1988;296: 1029-30. 12. Rogers D, Bitner-Glindzicz M, Harris C, Yudkin JS. Nonmydriatic retinal photography as a screening service for general practitioners. Diabetic Med 1990;7: 165-7. 13. Klein R, Barrett-Connor EL, Blunt BA, Wingard DL. Visual impairment and retinopathy in people with normal glucose tolerance, impaired glucose tolerance and newly diagnosed NIDDM. Diabetes Care 1991;14:914-8. 14. National Center for Health Statistics. National Health and Nutrition Examination Survey III Data Collection Forms. Hyattsville, MD: U.S. Department of Health and Human Services, 1990. 15. Klein R, Klein BEK, Neider MW, et al. Diabetic retinopathy as detected using ophthalmoscopy, a nonmydriatic camera and a standard fundus camera. Ophthalmology 1985;92: 485-91. 16. Ryder REJ, Vora JP, Ateia JA, et al. Possible new method to improve detection of diabetic retinopathy: Polaroid nonmydriatic retinal photography. Br Med J 1985;291:1256-7. 17. National Society to Prevent Blindness. Vision Problems in the U.S.: Data Analysis. New York: The Society, 1980. Available from: National Society to Prevent Blindness, 500 E. Remington Rd., Schaumburg, IL 60173. 18. Klein BEK, Davis MD, Segal P, et al. Diabetic retinopathy: assessment of severity and progression. Ophthalmology 1984;91: 10-7. 19. Klein R, Klein BEK, Magli YL, et al. An alternative method of grading diabetic retinopathy. Ophthalmology 1986;93: 1183-7.
1691
Ophthalmology Volume 99, Number 11, November 1992 20. Klein R, Davis MD, Magli YL, et al. The Wisconsin agerelated maculopathy grading system. Ophthalmology 1991;98:1128-34. 21. Klein R, Davis MD, Magli YL, Klein BEK. Wisconsin AgeRelated Maculopathy Grading System. Madison: Department of Ophthalmology, Univ of Wisconsin School of Medicine 1991. Available from: National Technical Information Service (Accession #PB91-184267/AS). 22. Entine S, Holladay D, Olscheske T. WISAR: Wisconsin Storage and Retrieval System. Madison: University ofWisconsin Clinical Cancer Center, 1981 . 23. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977;33:159-74.
1692
24. Bressler NM, Bressler SB, West SK, et al. The grading and prevalence of macular degeneration in Chesapeake Bay watermen. Arch Ophthalmol 1989;107:847-52. 25. Leibowitz HM, Krueger DE, Maunder LR, et al. The Framingham Eye Study Monograph. Surv Ophthalmol 1980;24(Suppl): 1-610. 26. Sperduto RD, Hiller MR, Podgor MJ, et al. Comparability of ophthalmic diagnoses by clinical and Reading Center examiners in the Visual Acuity Impairment Survey Pilot Study. Am J Epidemiol 1986;124:994-1003. 27. Klein BEl(, Klein R, Linton KLP. Prevalence of age-related lens opacities in a population. The Beaver Dam Eye Study. Ophthalmology 1992;99:546-52.
Fundus photography is an objective method of recording retinal pathology (or its absence) that has been useful in screening,1,2 routine care, 3 epidemiologic studies,4-6 and
Originally received: April 20, 1992. Manuscript accepted: May 18, 1992. From the University of Wisconsin, Department of Ophthalmology, Madison. Supported by grant 5 U IO-EY06594 from the National Eye Institute/ NIH, Bethesda, Maryland (Drs. R. Klein and B. E. K. Klein). The authors have no proprietary interest in the development or marketing of the camera used in this study or competing cameras. Reprint requests to Ronald Klein, MD, MPH, University of WisconsinMadison, Department of Ophthalmology, 600 Highland Ave, E5/353 esc, Madison, WI 53792.
1686
clinical trials. 7- 1O Optimaily, 30° color stereoscopic fundus photographs, taken through pharmacologically dilated pupils, are used. However, there are situations in which it is not possible to dilate the pupils with pharmacologic agents or when costs prohibit use of standard 30° cameras. II - 14 In these circumstances, a "nonmydriatic" camera has been used as an alternative means of documenting retinal pathology. Data from earlier studies suggest that evaluation of 45° nonstereoscopic fundus photographs by experienced graders provides a reasonably reliable measure of diabetic retinopathy and is a more sensitive and objective method of detecting retinopathy than direct ophthalmoscopy. 15, 16 No data, however, are available to evaluate the reliability of these methods for detecting age-related maculopathy. This is important, as age-related maculopathy is a leading
Klein et al . Detection of Drusen and ARM cause of visual impairment in older Americans. 17 The purpose of this report is to compare the presence of drusen and signs of early and late age-related maculopathy as determined: (1) by grading of nonstereoscopic 45° retinal photographs taken with the nonmydriatic camera through a nonpharmacologically dilated pupil; and (2) by grading of nonstereoscopic 45° retinal photographs taken with the nonmydriatic camera through a pharmacologically dilated pupil versus grading of stereoscopic retinal photographs taken with a standard 30° camera through a pharmacologically dilated pupil.
Methods Population The study population consisted of subjects photographed at two sites. One group consisted of 80 diabetic subjects (mean age, 61 years) participating in a study to evaluate photography using different cameras as a method of screening for diabetic retinopathy (unpublished data; Pugh et al). The second group consisted of 32 patients with agerelated macular degeneration who were regularly seen in the Retina Clinic at the University of Wisconsin Hospital and Clinics. None had diabetes mellitus. Their mean age was 74 years. The nonmydriatic camera (Canon CR3, Lake Success, NY), which uses infrared light to televise a view of the fundus, was used to photograph the retina of the nonpharmacologically dilated eye of each subject. One photograph was taken of a field centered horizontally and vertically on a point midway between the temporal edge of the optic disc and the fovea. The 45° image produced included an area above and below the temporal arcades and areas just nasal to the disc and temporal to the macula (Fig 1). This image is minified 0.64X (at 0 diopters) compared with that taken with a standard 30° camera (Fig 2) (Zeiss FF3, Carl Zeiss, Thornwood, NY).
Figure 1. Forty-five degree fundus photograph taken with the nonmydriatic camera through nonpharmacologically dilated pupil shows standard field taken for study.
Figure 2. Thirty degree fundus photograph taken of Diabetic Retinopathy Study Field 2. Notice the relative magnification of the image compared with the 45° photograph.
One drop of 2.5% phenylephrine and one drop of 1% tropicamide were then instilled in the cul-de-sac of each eye to obtain dilation. Another photograph of the same retinal field was taken with the 45° camera in 103 of the 112 subjects. Shortly thereafter, stereoscopic color retinal photographs were taken of the macula with a standard 30° (Zeiss) fundus camera (Diabetic Retinopathy Study [DRS] Standard Field 2). All photographs for a given subject were taken with either color Kodachrome 25 or Ektachrome 100 slide film and were processed and returned as 2 X 2 inch color slides. The slides were mounted in clear plastic mounting sheets and graded at the University of Wisconsin by a single experienced masked grader. First, nonmydriatic camera photographs taken through a nonpharmacologically dilated pupil were graded, then (at least 1 week later) nonmydriatic photographs taken through a pharmacologically dilated pupil were graded, and finally (at least 1 week later) photographs taken with the standard 30° camera were graded. The grader, masked to subject information, used the National Health and Nutrition Examination Survey III grading protocol, described in detail elsewhere. 14 Briefly, after assessing the quality of the photographs (clarity, field definition, presence of artifacts, and the ability to grade lesions), the severity of specific lesions associated with diabetic retinopathy was determined using a modification of the Airlie House Classification. 7 ,18,19 Next, the presence of any drusen, soft drusen, retinal pigment epithelial degeneration, increased retinal pigment, geographic atrophy, and signs of exudative macular degeneration (subretinal hemorrhage, subretinal fibrous scar, retinal pigment epithelium detachment and/or serous detachment of the sensory retina) were determined using a modification of the Wisconsin Age-Related Maculopathy grading system. 20 ,21 The location oflesions within the temporal retinal vascular arcades in field 2 was assessed. To examine between-grader variation, a second experienced grader graded 44 sets of 45° and 30° photographs from the orig-
1687
Uphthalmology
VoLume 99, Number 11, November 1992 photographs: 45° photographs taken with the non mydriatic camera through a nonpharmacologically dilated pupil; and 30° photographs taken through a dilated pupil. Exact agreement was 75.0% (84 of 112). The unweighted kappa score was 0.60 (standard error [SE], 0.06). Gradings of 45 ° photographs and 30° stereoscopic photographs taken through pharmacologically dilated pupils were in agreement 89.3% (92 of 103) of the time. The unweighted kappa score was 0.82 (SE, 0.07). There was a greater likelihood of "underestimating" early age-related maculopathy when grading 45° photographs compared to when grading 30° stereoscopic photographs. Of the 45° photographs taken through a nonpharmacologically dilated pupil, 9.8% (11 of 112) could not be graded. Of the 45° photographs taken through the pharmacologically dilated pupil, 1.9% (2 of 103) could not be graded. Of the 112 30° photographs taken with the standard camera through a dilated pupil, drusen were found in the macular area in 87, soft drusen in 42, retinal pigment epithelial degeneration in 25, increased retinal pigment in 29, exudative macular degeneration in 11, and geographic atrophy in 7. Analyses were performed to compare gradings of photographs for specific age-related maculopathy lesions. The results are summarized in Table 2. Exact agreement between gradings of photographs taken with the standard 30° camera ranged from 72% to 89% compared with gradings of 45° photographs taken through nonpharmacologically dilated pupils and ranged from 79% to 98% compared with gradings of 45 ° photographs taken through pharmacologically dilated pupils. The kappa scores were generally in the moderate to substantial agreement cate-
inal set using the same grading procedure. To examine within-grader variation, the original grader regraded 20 sets of 45° and 30° photographs after an interval of 6 months.
Definitions Early age-related maculopathy was defined as either the presence of soft drusen or the presence of any drusen type with retinal pigment epithelial degeneration or increased retinal pigment in the macular area in the absence of late age-related maculopathy as defined below. 6 Late age-related maculopathy was defined as the presence of signs of exudative age-related macular degeneration or pure geographic atrophy.
Statistics Wisconsin Storage and Retrieval, a data processing software system, was used for processing all subject files. 22 Percentages of exact agreement were calculated, and unweighted kappa scores were computed. 23 For interpreting kappa values, we used guidelines set forth by Landis and Koch,23 which describe the relative strength of agreement associated with different values of kappa: fair, 0.21 to 0.40; moderate, 0.4 I to 0.61; substantial, 0.61 to 0.80; and almost perfect, 0.81 to 1.00.
Results Table 1 compares severity of age-related maculopathy (none, early, or late) determined by grading two types of
Table 1. Comparison of Severity of Age-related Maculopathy by Grading Photographs Taken with Both a Nonmydriatic 45 0 Camera and a Standard 30 0 Camera Standard 30° Camera (dilated pupil) Nonmydriatic 45° Camera Nonpharmacologically Dilated Pupil No age-related maculopathy Early age-related maculopathy Late age-related maculopathy Cannot grade Total Pharmacologically Dilated Pupil No age-related maculopathy Early age-related maculopathy Late age-related maculopathy Cannot grade Pictures not taken Total
1688
No Age-related Maculapathy
Early Age-related Maculapathy
Late Age-related Maculapathy
47
12
0
3 0
22
2 15
0
7 57
4
0
38
17
52 2
7 24
0 0
0
0
16
1 2 57
1 6
0
38
1 17
Total
59 27 15 11 112 59 26 16 2 9 112
Klein et al . Detection of Drusen and ARM Table 2. Percent of Eyes with Agreement between Gradings of Specific Age-related Maculopathy Lesions 45° Photographs Taken Through Nonpharmacologically Dilated Pupil and 30° Photographs Taken Through Pharmacologically Dilated Pupil
Characteristic Any drusen Soft drusen Retinal pigment epithelial degeneration Increased retinal pigment Exudative macular degeneration Geographic atrophy SE
=
45° Photographs and 30° Photographs Taken Through Pharmacologically Dilated Pupil
Exact Agreement (%)
Kappa (SE of Kappa)
Exact Agreement (%)
(SE of Kappa)
74.9 72.3
0.42 (0.08) 0.50 (0.07)
78.6 79.6
0.44 (0.09) 0.62 (0.08)
72.3 74.1 89.3 84.8
0.39 (0.07) 0.42 (0.07) 0.60 (0.07) 0.33 (0.06)
85.5 82.5 98.0 97.1
0.54 0.51 0.91 0.79
Kappa
(0.08) (0.08) (0.09) (0.09)
standard error.
gories. There was a greater likelihood of "underestimating" the presence of soft drusen (31.3% versus 37.5%), retinal pigment epithelial degeneration (13.4% versus 22.3%), and increased retinal pigment (15.2% versus 25.9%) when grading 45° photographs taken through a nonpharmacologically dilated pupil compared with grading 30° stereoscopic photographs. There were no photographs taken with the 45 0 camera in which the grader misclassified drusen and hard exudate or soft exudate as determined by grading 30° photographs. Between- and within-grader agreement for detection of specific age-related maculopathy lesions from photographs taken with the 45 ° camera (through a nonpharmacologically dilated and a pharmacologically dilated pupil) and/or the 30° camera are presented in Tables 3 and 4. The kappa scores were generally in the moderate to substantial agreement categories. The quality of the photographs was judged to be fair to good in 81 (72.3%), borderline in 26 (23.2%), and poor in 5 (4.5%) of the 45° photographs taken through a nonpharmacologically dilated pupil. All 112 30° stereoscopic photographs were considered to be offair to good quality. Focus problems were present in 37 (33.0%) and illumination problems were present in 20 (17.9%) of the 112 45° photographs taken through a non pharmacologically dilated pupil. Agreement with 30° stereoscopic photographs was significantly higher (P < 0.05) for 45° photographs taken through a nonpharmacologically dilated pupil that were of fair to good quality than for those of borderline to poor quality for the following lesions: presence of drusen (82.7% versus 58.0%), soft drusen (77.7% versus 58.0%), retinal pigment epithelial degeneration (80.2% versus 51.6%), increased retinal pigment (81.4%
versus 54.8%), exudative macular degeneration (98.7% versus 64.5%), and geographic atrophy (85.1 % versus 54.8%).
Discussion Our data suggest that photographs taken through a nonpharmacologically dilated pupil with a nonmydriatic camera may be of value in the detection of age-related maculopathy in some epidemiologic studies. However, if the objective of the study is to use age-related maculopathy as an outcome variable for a controlled clinical trial or to reliably document the appearance of specific age-related maculopathy lesions, grading of stereoscopic 30 0 fundus photographs would be the first choice. Photographs taken with either camera would probably be oflimited value in the routine screening of asymptomatic healthy older patients in areas where ophthalmologic care is limited because of the relatively low prevalence of eyes with exudative macular degeneration that would benefit from laser ablation of extrafoveal subretinal new vessels. 6 ,24,25 Our data suggest that while agreement between gradings of fundus photographs taken with different cameras is very good for detection of drusen and late age-related macular degeneration, the 45 ° nonstereoscopic fundus photographs are less sensitive as a means of detecting early signs of age-related maculopathy, such as the presence of pigment epithelial abnormalities. This is not surprising, as the latter may be subtle, requiring excellent focus, magnification, and stereopsis to detect. In addition, in some cases increased retinal pigment may become more ap-
1689
Ophthalmology
Volume 99, Number 11, November 1992
Table 3. Percent of Agreement for Age-related Maculopathy Lesions and Age-related Maculopathy Severity between Two Graders for 44 Eyes 45° Photographs Taken Through Nonpharmacologically Dilated Pupil
Exact Agreement Characteristic
(%)
Kappa (SE of Kappa)
Any drusen Soft drusen Retinal pigment epithelial degeneration Increased retinal pigment Exudative macular degeneration Geographic atrophy Overall age-related maculopathy severity"
84.1 79.5 72.7 86.3 88.6 79.5 90.9
45° Photographs Taken Through Pharmacologically Dilated Pupil
Exact Agreement (%)
Kappa (SE of Kappa)
0.68 (0.12) 0.68 (0.11) 0.54 (0.10) 0.74 (0.12) 0.77 (0.11) 0.54 (0.10)
81.8 75.0 77.2 77.2 95.4 86.3
0.87 (0.09)
84.1
30° Photographs Taken Through Pharmacologically Dilated Pupil
Exact Agreement (%)
Kappa (SE of Kappa)
0.58 (0.13) 0.59 (0.11) 0.56 (0.12) 0.49 (0.13) 0.89 (0.13) 0.55 (0.13)
88.6 72.7 77.2 88.6 100.0 95.4
0.71 (0.15) 0.54 (0.11) 0.59 (0.12) 0.79 (0.13) 1.00 (0.15) 0.83 (0.15)
0.77 (0.10)
93.2
0.90 (0.11)
SE = standard error . • None, early, or late.
parent in one member of the stereoscopic 30° photograph pair because of variations in illumination. There are few data available regarding the ability to detect signs of age-related maculopathy using ophthalmoscopy. In the pilot phase of the Visual Acuity Impairment Survey, Sperduto et aJ26 reported exact agreement of84.2% (112 of 133; kappa, 0.44; SE, 0.09) for detection of signs of macular degeneration between ophthalmologists using ophthalmoscopy and masked ophthalmologists grading stereoscopic fundus photographs. In the Framingham Eye Study, retinal drusen and pigmentary abnormalities were significantly (P < 0.05) less likely to be detected by the examiners (third year ophthalmology residents or staff ophthalmologists) compared with detection by grading of stereoscopic fundus photographs. However, agreement between examiners and graders was similar for detection of exudative macular degeneration. 25 In the Table
Beaver Dam Eye Study, the exact agreement for detection of soft drusen by trained nonophthalmologist examiners using direct and indirect ophthalmoscopy and graders examining stereoscopic 30° fundus photographs taken with a standard fundus camera was 85.0% (kappa, 0.34; SE, 0.02) (Klein R, unpublished data). These data suggest that grading of 45° fundus photographs, taken with the nonmydriatic camera, may be of similar sensitivity for detecting signs of age-related maculopathy as ophthalmoscopy done by non ophthalmologists or ophthalmologists. However, fundus photography has several advantages. First, changes over time can be more readily recognized. In addition, the reliability of these assessments can be evaluated by replicate gradings. The quality of the photographs and gradings can be monitored, and remedial action can be taken as needed. Graders can be masked with respect to subject characteristics. Photographs also
4. Percent of Agreement for Age-related Maculopathy Lesions within Grader for 20 Eyes 45° Photographs Taken Through Nonpharmacologically Dilated Pupil
45° Photographs Taken Through Pharmacologically Dilated Pupil
Characteristic
Exact Agreement (%)
Kappa (SE of Kappa)
Exact Agreement (%)
Any drusen Soft drusen Retinal pigment epithelial degeneration Increased retinal pigment
90.0 85.0 85.0 70.0
0.80 (0.17) 0.74 (0.17) 0.71 (0.14) 0.40 (0.13)
90.0 85.0 95.0 95.0
SE
=
standard error.
1690
Kappa (SE of Kappa)
0.73 0.73 0.83 0.81
(0.22) (0.20) (0.19) (0.17)
30° Photographs Taken Through Pharmacologically Dilated Pupil
Exact Agreement (%)
Kappa (SE of Kappa;
90.0 90.0 85.0 85.0
0.69 (0.22) 0.80 (0.19) 0.70 (0.18) 0.73 (0.19)
Klein et al . Detection of Drusen and ARM permit an opportunity to add new items to the grading scheme. One concern when grading 45° nonstereoscopic fundus photographs is the misclassification oflesions such as retinal drusen and hard exudate. Such misclassification may result in erroneous conclusions regarding relationships of prevalence of hard exudate or in decisions to refer such patients for evaluation of possible macular edema. Our data suggest that this may not be a problem, as we found no such misclassification. This may have resulted from recognition of differences of the lesions in the 45° photographs by the experienced grader. The sharper margins, the pattern, and the juxtaposition to retinal microaneurysms are suggestive of the appearance of hard exudate material in contrast to individual round flat spots, pale yellowish to white in color, which are typical of small drusen. Similarly, there was no misclassification of soft drusen as soft exudates. There were fewer ungradable 45° fundus photographs (9.8% of those taken through a nonpharmacologically dilated pupil and 1.9% through a pharmacologically dilated pupil) compared with an earlier study (12.7% and 6.8%, respectively), which suggests an improvement in the technology and/or the photographers. IS In our study, better photographic quality was associated with increased ability to detect the lesions associated with age-related maculopathy. These data suggest that the optimal method of detecting the presence and severity of age-related maculopathy is grading of 30° stereoscopic fundus photographs. In older people, a higher frequency of poorer quality photographs taken through a nonpharmacologically dilated pupil is expected because of the high prevalence of media opacities such as cataract and a decrease in the ability of the pupil to dilate. In the Beaver Dam Eye Study, people 75 years of age or older had a higher frequency of late cataract (52.2% versus 0.5%) and smaller pupils after pharmacologic dilation (less than 5 mm in diameter) (9.6% versus 1.4%) than people 43 to 54 years of age.27 For these reasons, pharmacologic dilation of the pupil is important in studies of age-related maculopathy in people 75 years of age or older. In summary, the use of the nonmydriatic camera for detection of age-related maculopathy should only be considered in situations where the pupils cannot be pharmacologically dilated and retinal specialists are not available to examine the fundus. In these situations, it may offer a more sensitive and objective method of detecting maculopathy than direct ophthalmoscopy. Acknowledgment. The authors thank Jacqueline A.
Pugh, MD, for permitting them to use fundus photographs taken in her study, which was funded by the V.A. Health Services Research and Development Office.
References 1. Javitt JC, Canner JK, Sommer A. Cost effectiveness of current approaches to the control of retinopathy in type I diabetics. Ophthalmology 1989;96:255-64.
2. Dasbach EJ, Fryback DG, Newcomb PA, et al. Cost-effectiveness of strategies for detecting diabetic retinopathy. Med Care 1991;29:20-39. 3. American Academy of Ophthalmology Quality of Care Committee, Retina Panel. Diabetic Retinopathy. Preferred Practice Pattern. San Francisco: The Academy, 1989. 4. Klein R. Retinopathy and other ocular complications in diabetes. In: Olefsky JM, Sherwin R, eds. Diabetes Mellitus: Management and Complications. New York: Churchill Livingstone, 1985; 10 1-58. 5. Klein R, Klein BEK, Moss SE, et al. The Wisconsin Epidemiologic Study of Diabetic Retinopathy.II.Prevalence and risk of diabetic retinopathy when age at diagnosis is less than 30 years. Arch Ophthalmol 1984;102:520-6. 6. Klein R, Klein BEK, Linton KLP. The prevalence of agerelated maculopathy. The Beaver Dam Eye Study. Ophthalmology 1992;99:933-43. 7. Diabetic Retinopathy Study Research Group. Report 7. A modification of the Airlie House classification of diabetic retinopathy. Invest Ophthalmol Vis Sci 1981 ;21 :210-26. 8. 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. 9. Davis MD, Hubbard LD, Trautman J, Klein R. Studies of retinopathy. Methodology for assessment and classification with fundus photographs. Diabetes 1985;34(SuppI3):42-9. 10. The Diabetes Control and Complications Trial Research Group. Color photography vs fluorescein angiography in the detection of diabetic retinopathy in the Diabetes Control and Complications Trial. Arch OphthalmoI1987;105:134451. 11. Jones D, Dolben J, Owens DR, etal. Non-mydriatic Polaroid photography in screening for diabetic retinopathy: evaluation in a clinical setting. Br Med J 1988;296: 1029-30. 12. Rogers D, Bitner-Glindzicz M, Harris C, Yudkin JS. Nonmydriatic retinal photography as a screening service for general practitioners. Diabetic Med 1990;7: 165-7. 13. Klein R, Barrett-Connor EL, Blunt BA, Wingard DL. Visual impairment and retinopathy in people with normal glucose tolerance, impaired glucose tolerance and newly diagnosed NIDDM. Diabetes Care 1991;14:914-8. 14. National Center for Health Statistics. National Health and Nutrition Examination Survey III Data Collection Forms. Hyattsville, MD: U.S. Department of Health and Human Services, 1990. 15. Klein R, Klein BEK, Neider MW, et al. Diabetic retinopathy as detected using ophthalmoscopy, a nonmydriatic camera and a standard fundus camera. Ophthalmology 1985;92: 485-91. 16. Ryder REJ, Vora JP, Ateia JA, et al. Possible new method to improve detection of diabetic retinopathy: Polaroid nonmydriatic retinal photography. Br Med J 1985;291:1256-7. 17. National Society to Prevent Blindness. Vision Problems in the U.S.: Data Analysis. New York: The Society, 1980. Available from: National Society to Prevent Blindness, 500 E. Remington Rd., Schaumburg, IL 60173. 18. Klein BEK, Davis MD, Segal P, et al. Diabetic retinopathy: assessment of severity and progression. Ophthalmology 1984;91: 10-7. 19. Klein R, Klein BEK, Magli YL, et al. An alternative method of grading diabetic retinopathy. Ophthalmology 1986;93: 1183-7.
1691
Ophthalmology Volume 99, Number 11, November 1992 20. Klein R, Davis MD, Magli YL, et al. The Wisconsin agerelated maculopathy grading system. Ophthalmology 1991;98:1128-34. 21. Klein R, Davis MD, Magli YL, Klein BEK. Wisconsin AgeRelated Maculopathy Grading System. Madison: Department of Ophthalmology, Univ of Wisconsin School of Medicine 1991. Available from: National Technical Information Service (Accession #PB91-184267/AS). 22. Entine S, Holladay D, Olscheske T. WISAR: Wisconsin Storage and Retrieval System. Madison: University ofWisconsin Clinical Cancer Center, 1981 . 23. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977;33:159-74.
1692
24. Bressler NM, Bressler SB, West SK, et al. The grading and prevalence of macular degeneration in Chesapeake Bay watermen. Arch Ophthalmol 1989;107:847-52. 25. Leibowitz HM, Krueger DE, Maunder LR, et al. The Framingham Eye Study Monograph. Surv Ophthalmol 1980;24(Suppl): 1-610. 26. Sperduto RD, Hiller MR, Podgor MJ, et al. Comparability of ophthalmic diagnoses by clinical and Reading Center examiners in the Visual Acuity Impairment Survey Pilot Study. Am J Epidemiol 1986;124:994-1003. 27. Klein BEl(, Klein R, Linton KLP. Prevalence of age-related lens opacities in a population. The Beaver Dam Eye Study. Ophthalmology 1992;99:546-52.
