메인 Ophthalmology Agreement among Optometrists, Ophthalmologists, and Residents in Evaluating the Optic Disc for...
문제 보고This book has a different problem? Report it to us
"네" 선택하시는 조건: "네" 선택하시는 조건: "네" 선택하시는 조건: "네" 선택하시는 조건:
파일 열기 성공했습니다
파을 내용은 책 (또는 만화)입니다
책 내용이 적당합니다
파일의 제목, 작성자와 언어가 책 설명과 일치합니다. 다른 필드는 보조이므로 무시하셔도 좋습니다.
"아니요" 선택하시는 조건: "아니요" 선택하시는 조건: "아니요" 선택하시는 조건: "아니요" 선택하시는 조건:
- 잘못된 파일입니다
- 이 파일이 DRM으로 보호돼 있습니다
- 파일은 책이 아닙니다 (예: xls, html, xml)
- 파일은 기사입니다
- 파일은 책에 일부입니다
- 파일은 잡지입니다
- 파일은 시험지 또는 테스트입니다
- 파일은 스팸입니다
책의 내용이 적당하지 않으며 차단되어야 한다고 생각합니다
파일의 제목, 작성자와 언어가 책 설명과 일치하지 않습니다. 다른 필드는 무시하셔도 좋습니다.
Change your answer
Agreement among Optometrists, Ophthalmologists, and Residents in Evaluating the Optic Disc for Glaucoma Lisa S. Abrams, MD,' Ingrid U. Scott, MD,' George L. Spaeth, MD, 2 Harry A. Quigley, MD,' Rohit Varma, MD, MPH',3 Purpose: To determine the agreement among optometrists, ophthalmologists, and ophthalmology residents in assessing glaucomatous optic nerve damage. The authors also determined the sensitivity of each group of observers for identifying glaucomatous optic nerve damage. Methods: Six optometrists, six general ophthalmologists, and six third-year ophthalmology residents evaluated 75 stereoscopic optic disc photographs. Observers estimated the vertical cup:disc ratio (C:D) and assessed the presence of glaucomatous damage. Agreement among and within observers was estimated by the kappa statistic (K w , k). The sensitivity and specificity for the identification of glaucomatous optic nerve damage were determined for each group of participants. Results: Intraobserver agreement (K w 0.69-0.79) was greater than interobserver agreement (K w 0.56-0.68) in assessing the C:D ratio and glaucomatous optic nerve damage for optometrists, ophthalmologists, and residents. Interobserver agreement for ophthalmologists (Kw 0.68) was substantial and significantly higher than for optometrists (Kw 0.56) and residents (K w 0.56) when estimating the C:D ratio. Ophthalmologists and residents had higher sensitivity (78%) in identifying glaucomatous optic nerve damage than did optometrists (56%). The specificity for all three groups was relatively poor (range, 47%-60%). Conclusion: The moderate interobserver agreement across all three groups of observers suggests the need to develop standardized criteria for assessing glaucomatous optic disc damage. Ophthalmologists in this study have a higher interobserver agreement in estimating the C:D ratio and are more sensitive than optometrists in assessing glaucomatous optic nerve damage. Ophthalmology 1994;101:1662-1667 Assessment of the optic disc is an essential part of glauOrigin; ally received: September 9, 1993. Revision accepted: February 16, 1994. I The Wilmer Ophthalmological Institute, Johns Hopkins University School of Medicine, Baltimore. 2 Glaucoma Service, The Wills Eye Hospital, Philadelphia. 3 Glaucoma Service, Doheny Eye Institute, University of Southern California, Los Angeles. Presented in part at the Association for Research in Vision and Ophthalmology Annual Meeting, Sarasota, May 1993. Reprint requests to Rohit Varma, MD, MPH, Glaucoma Service, Doheny Eye Institute, 1450 San Pablo St, Los Angeles, CA 90033. 1662 coma screening and management. Important characteristics of the optic nerve head include the horizontal and vertical cup:disc ratio (CD), 1 the neuroretinal rim area.i' optic disc pallor," and the nerve fiber layer appearance.' However, consistency in evaluating the CD ratio, even among glaucoma specialists, has been variable.v '? Lichter? studied the observer variability of 16 experts in evaluating 20 stereoscopic and monoscopic disc photographs from ten patients and found high intraobserver agreement for grading the CD ratio but lower agreement among observers both for grading the CD ratio and for differentiating between normal and glaucomatous eyes. Intraobserver and interobserver agreement for determining the Abrams et al . Evaluating the Optic Disc for Glaucoma C:D ratio was more reliable for stereoscopic as opposed to monoscopic photographs. Tielsch and co-workers" evaluated two subspecialists in assessing the optic disc both clinically and from photographs. They found that intraobserver agreement was greater than interobserver agreement for estimating the C:D ratio. Varma et al? in 1989 evaluated the agreement among two experienced observers and an automated image analyzer in analyzing 35 simultaneous stereoscopic optic disc photographs. They found excellent intraobserver and interobserver agreement in estimating the C:D ratio. In 1992, Varma et al,'" using the same photograph set as used in this report, evaluated the intraobserver and interobserver agreement among six glaucoma experts in estimating the CD ratio and assessing glaucomatous optic disc damage. Intraobserver agreement was excellent for both parameters. Interobserver agreement for estimating the CD ratio was substantial, but interobserver agreement for assessing glaucomatous damage was moderate. Thus, the current literature suggests that substantial agreement among experts is difficult to achieve. In addition to experts, other vision care providers examine patients for glaucoma. To our knowledge, the ability of these observers to evaluate the optic nerve for glaucomatous damage has not been estimated. In this study, we compared the ability of optometrists, ophthalmologists, and ophthalmology residents to estimate the C:D ratio and to assess glaucomatous optic nerve damage from stereoscopic optic disc photographs. We also determined the sensitivity and specificity of each group of participants to identify glaucomatous optic nerve damage. Methods Seventy-five optic disc photographs from 75 patients were chosen from the files of the Wills Eye Hospital Glaucoma Service. The photographs were selected because they were well focused and represented patients who did and did not have glaucomatous visual field loss. This was defined as reliable, typical, glaucomatous visual field loss in either automated static or Goldmann kinetic perimetric testing.!' The distribution of the C:D ratio among these photographs (Fig 1) represented the complete range of the ratio and was weighted toward a larger C:D ratio to amplify differences that may exist between various groups of observers. All photographs were obtained on a Topcon simultaneous stereoscopic fundus camera (Model TRCSS, Topcon Instrument Corp of America, Paramus, NJ) with the stereo base set at 3.0 mm, using Kodak Ektachrome 100 film. Photographs were viewed through a hand-held viewer (Asahi Pentax Stereo Viewer II, Asahi Optical Co, Tokyo, Japan). Six optometrists and six general ophthalmologists, each with at least 10 years of clinical experience, were recruited for this study. All six optometrists had completed 4 years at an accredited institution of optometric training. The ophthalmologists had all completed residency training in ophthalmology, were in private practices, and were not subspecialty trained nor were their practices limited to 40 35 .. e .2 ~ 30 25 ill ~ 20 0 CI ~ 15 . 0.. 10 5 0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 C/O Figure 1. Frequency distribution of the mean vertical cup.disc ratio of 75 optic discs as estimated by six expert observers. any particular subspecialty. Six third-year residents in ophthalmology were chosen from six different institutions; all residents had a medical degree and at least 3 years of postgraduate training. Each observer was asked to record the CD ratio and to assess the presence or absence of glaucomatous damage based solely on the appearance of the optic disc. Observers were masked to all clinical data as well as to the responses of the other observers. They were asked to define the disc margin as the inner margin of the scleral ring, and the cup margin as the point where the disc surface first begins to deform backward into the cup. With visual estimation, the longest vertical cup diameter was divided by the longest vertical disc diameter to obtain the C:D ratio. Observers were asked to use their own training and clinical experience to assess whether the optic discs had glaucomatous damage, and were not required to specify the criterion they used in reaching their conclusion. Observers examined all optic discs randomly, one at a time, stereoscopically. As a measure of intraobserver reliability, 25 of the optic disc photographs were duplicated and included for review. Thus, 100 optic disc photographs were viewed by each observer. Agreement among and within observers for the estimation of the C:D ratio was determined using the weighted kappa statistic (K w ). The difference between the estimates of the C:D ratio within and among observers was determined and averaged for each group of participants. Agreement among and within observers for the assessment of glaucomatous damage was determined using the unweighted kappa statistic (k). An overall Ke, and k were calculated'? to determine the overall intraobserver and interobserver agreement for each group of participants. The kappa statistic is used with categorical data and expresses the agreement between two measures corrected for chance agreement. The value of kappa ranges from -1 (complete disagreement) to 0 (chance agreement) to + 1 (perfect agreement). The unweighted kappa statistic treats all disagreement, large or small, identically. The weighted kappa statistic assigns different weights to various 1663 Ophthalmology Volume 101, Number 10, October 1994 Table 1. Determination of Weights for Calculating K, Difference between Estimates of Vertical C:D (~) (DD) 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Weight (1 - ~) 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 CoD = cup-disc; DD = disc diameter. degrees ofdisagreement. In this study, the weights assigned to each observation for estimating the C:D ratio were equal to one minus the difference between the two estimates (Table I). Thus, smaller disagreements were weighted more heavily than larger disagreements, as is the convention when using the weighted kappa statistic. A kappa or weighted kappa value of -1.0 to 0.0 was considered poor agreement; from 0.01 to 0.2 was considered slight agreement; from 0.21 to 0.40 was considered fair agreement; from 0.41 to 0.60 was considered moderate agreement; from 0.61 to 0.80 was considered substantial agreement; from 0.81 to 0.99 was considered almost perfect agreement; and + 1.0 was considered perfect agreement. 13 The 95% confidence intervals for each K w and k also were calculated for each group of observers. Differences were considered to be statistically significant when the 95% confidence intervals of the K w or k of two groups of observers did not overlap. For the purposes of comparison, we re-analyzed the assessments by six glaucoma experts on the same optic disc photographs. 10 We reviewed the optic disc photographs to identify only those optic discs for which all six glaucoma experts unanimously agreed on the presence or absence of glaucomatous damage. All six experts identified 18 optic discs as having glaucomatous damage and 15 optic discs as being normal. The expert assessment for these 33 optic disc photographs was used as the diagnostic "gold standard." The responses of the optometrists, ophthalmologists, and residents for these 33 discs were then pooled to determine the sensitivity and specificity for the identification of glaucomatous optic nerve damage for each group of observers. Results Demographic and clinical data (including visual fields) were known for 60 of the 75 patients (80% of the study 1664 group). The mean age was 51 years (range, 19-78 years). There were 35 men and 25 women, 85% of whom were white, 12% were black, and 3% were ofother racial origins. Twenty-nine patients (49%) had glaucomatous visual loss (defined as reliable, typical glaucomatous visual field defects on automated static or Goldmann kinetic perimetric testing). Twenty-six patients had primary open-angle glaucoma, two had secondary glaucoma, and one had congenital glaucoma. The remaining 31 individuals had no evidence of glaucomatous visual loss by visual field examination. Intraobserver agreement in estimating the CD ratio for all three groups of observers was substantial (K w , 0.690.79) (Table 2). There were no statistically significant differences in the intraobserver agreement among optometrists, ophthalmologists, and residents (Table 2). Intraobserver estimates of the CD ratio differed by more than 0.2 disc diameters in fewer than 5% of observations for all three groups (Fig 1). Interobserver agreement in estimating the C:D ratio was significantly higher for ophthalmologists (K w , 0.68) than for optometrists (K w , 0.56) or residents (K w , 0.56) (Table 2). Interobserver estimates of the C:D ratio differed by more than 0.2 disc diameters in approximately 5% of observations for ophthalmologists and in 12% of observations for optometrists and residents (Fig 2). Intraobserver agreement for assessing glaucomatous damage was substantial and similar for all groups (k, 0.700.81) (Table 3). Interobserver agreement was significantly lower than intraobserver agreement for all groups. Interobserver agreement among residents (k, 0.50) was significantly better than for optometrists (k, 0.40). The interobserver agreement for ophthalmologists was not significantly different from that of residents or optometrists. The sensitivity for identifying glaucomatous disc damage was higher for ophthalmologists (78%) and residents (78%) than for optometrists (56%) (Table 4). The specificity was poor for all three groups of observers (range, 47%-60%). Discussion The current study has, for the first time, compared the agreement among optometrists, general ophthalmologists, Table 2. Observer Agreement in Estimating Vertical Cup:Disc Ratio Optometrists (n = 6) Residents (n = 6) Ophthalmologists (n = 6) Glaucoma experts (n = 6) CI = confidence interval. Intraobserver Interobserver Overall x, (95% en Overall x, (95% el) 0.69 0.78 0.79 0.88 (0.63-0.75) (0.73-0.82) (0.75-0.84) (0.84-0.92) 0.56 0.56 0.68 0.65 (0.53-0.58) (0.54-0.59) (0.66-0.71) (0.63-0.67) Abrams et al . Evaluating the Optic Disc for Glaucoma 100 Table 3. Observer Agreement in Assessing Glaucomatous Damage 90 80 70 60 50 Optometrists (n = 6) Residents (n = 6) 40 Ophthalmologists (n Experts (n = 6) 30 20 CI 10 1 a OPT iii] RES IIlI OPHTH and third-year ophthalmology residents in assessing the optic disc for glaucoma. As has been reported in similar studies evaluating glaucoma experts, we also found intraobserver agreement to be better than interobserver agreement for these three groups ofobservers in estimating the CD ratio and assessing glaucomatous damage. However, we found significant differences among these groups in their ability to identify glaucomatous optic nerve damage. D sing the same set of photographs as those used in this study, Varma et al'? previously have shown almost perfect 100 90 80 70 iii em 60 ~ .Q ~ 0 E e .0 50 .e, 40 a. 30 Overall k Overall k = = 6) 0.81 0.70 0.81 0.74 (0.71-0.91) (0.57-0.82) (0.71-0.90) (0.63-0.84) (95% CI) 0.40 0.50 0.47 0.51 (0.35-0.45) (0.47-0.53) (0.42-0.52) (0.46-0.56) confidence intervaL 1 Figure 2. Frequency distribution of intraobserver differences in an estimated vertical cup.disc ratio for optometrists, ophthalmology residents, and general ophthalmologists. c Interobserver (95% CI) 54 • Intraobserver 20 intraobserver agreement (Kw , 0.88) among glaucoma experts in estimating the C:D ratio to the nearest one-tenth unit. The intraobserver agreement among experts was significantly higher than for the optometrists, ophthalmologists, and ophthalmology residents in this study. Interobserver agreement among experts in the previous study and for ophthalmologists in the current study for estimating the C:D ratio was substantial (K w , 0.65 and 0.68, respectively) and greater than that of optometrists and residents (K w , 0.56). Also, intraobserver agreement among experts for evaluating glaucomatous damage was substantial, but not significantly better than for the three groups of observers in this study. Interobserver agreement among experts for assessing glaucomatous damage was moderate and similar to that of ophthalmologists and residents, but significantly better than for optometrists. Because intraobserver agreement for both estimating the C:D ratio and assessing glaucomatous damage was substantial or almost perfect for all groups of observers, serial observations of optic disc photographs by a single observer are likely to be highly reproducible. Yet, interobserver agreement was only fair to moderate and significantly lower than intraobserver agreement for estimating the C:D ratio. This suggests that observers use different anatomic clues when determining the CD ratio. In our study, optometrists and residents differed in their estimates of the C:D ratio by more than 0.2 disc diameters in 12% of cases, whereas ophthalmologists differed by more than 0.2 disc diameters in 5% of cases. This could potentially lead to erroneous assessment of the patient with glaucoma when such a patient is examined by more than one ophthalmic care provider. The advent of automated, quantitative, highly reproducible analysis of optic disc to- 10 0 0.0 DD 0.1 DD 0.2 DD > 0.2 DD . Magnitude of Intraobserver Disagreement In Estimating Vertical C:D Ratio (DD= disc diameters) • OPT m RES II Table 4. Sensitivity and Specificity for the Identification of Glaucomatous Discs (n = 33) OPH TH Figure 3. Frequency distribution of interobserver differences in an estimated vertical cup.disc ratio for optometrists, ophthalmology residents, and general ophthalmologists. Optometrists Residents General ophthalmologists Sensitivity (%) Specificity (%) 56 78 78 53 47 60 1665 Ophthalmology Volume 101, Number 10, October 1994 pography using image analyzers ma y improve our ability to evaluate the optic disc in glaucoma in the future. Interobserver agreement for assessing glaucomatous damage was significantly poorer than intraobserver agreement for all groups of observers, again indicating that indi vidual observers are internally consistent, but that the consensus among observers is limited. Such fair to moderate interobserver agreement raises questions about the knowledge among ophthalmic care providers of optic disc changes in glaucoma, and questions the use of this evaluation when screening for and treating patients with glaucoma. Clearl y, there is a need to identify optic disc characteristics that are associated with glaucomatous optic nerve damage. Such well-defined criteria would form the basis for providing better education of ophthalmic care providers. Ophthalmologists and ophthalmology residents had higher sensitivity than did optometrists in identifying glaucomatous optic discs. The sensitivity and specificity reported here represent the performance of these three groups of observers relative to a gold standard (i.e., the unanimous agreement among experts) for this specific set of optic disc photographs. The poor specificity for all three groups of observers ma y be due to (I) a lack of knowledge of the characteristics of the glaucomatous optic disc, (2) an inability to identify them, or (3) a lack of true differentiating features of glaucomatous discs from normal discs. On independent review , we found that observers had a tendency to use a larger C:O ratio as the major determinant of glaucomatous damage. While a larger C: D ratio is more likely to occur in glaucomatous than in normal eyes, large optic discs have large physiologic optic CUpSl4 and may be mistaken for discs with glaucomatous cup enlargement. The specificity of a large CO ratio is poor, and if this were the onl y sign used to determine glaucomatous optic nerve damage, the specificity for identifying such damage also would be poor. Although a larger C:O ratio often correlates with fewer nerve fibers, 15 other characteristics of the optic disc may be associated with glaucomatous optic nerve damage, including localized notching," optic disc hemorrhage,'? peripapillary atrophy," and large laminar pores." The strengths of this study lie in the relatively large number of optic disc photographs studied and the number of observers. We compared three groups of observers whose performance had never been studied previously. The optic disc photographs were all of good quality and represented a wide range of C:O ratios. Additionally, we obtained a sample of ophthalmic care providers from different institutions who had diverse training backgrounds to obtain a cross-section of community practitioners. A limitation of this study ma y be our definition of glaucomatous optic nerve damage as agreement among six glaucoma experts on the appearance of the optic disc. Some of the discs labeled as normal by the experts ma y have had visual field defects and therefore could have been erroneously categorized as normal. However, we chose the appearance of the optic nerve as our diagnostic standard for two reasons. First, glaucomatous changes in the optic nerve may precede visual field abnormalities' v P 1666 and thus, if we had used visual field loss as our onl y diagnostic criterion, we would have misclassified some optic discs as normal even though they had glaucomatous optic disc damage. Second, one of the aims of our study was to determine the sensitivity and specificity ofoptometrists, general ophthalmologists, and ophthalmology residents for identifying glaucoma optic disc characteristics. The only "gold standard" available for assessing glaucomatous optic disc changes is the complete, unanimous agreement by six glaucomatous experts. Although it would have been useful to have had another diagnostic standard, such as glaucomatous visual field loss, unfortunately we had quantitative visual field data on onl y 60 of the 75 eyes. In summary, as seen in previous studies, optometrists, ophthalmologists, and ophthalmology residents were all more likely to agree with themselves than with others of similar training in their assessment of optic disc photographs. Additionally, ophthalmologists and residents had higher sensitivity than optometrists for identifying glaucomatous optic nerve damage. Agreement among observers and the ability of all three groups to identify glaucomatous optic nerve damage ma y be improved by the development ofa standardized, valid definition ofthe features of the glaucomatous optic disc. Acknowledgment. The authors thank the ophthalmologists, optometrists, and ophthalmology residents who voluntarily contributed their time to participate in this study. References I. Armal y MF, Sayegh RE. The cup/disc ratio: the findings of tonometry and tonography in the normal eye. Arch Ophthalmol 1969;82: 191-6. 2. Balazsi AG, Drance SM, Schulzer M, Douglas GR. Neuroretinal rim area in suspected glaucoma and early chronic open-angle glaucoma: correlation with parameters of visual function. Arch Ophthalmol 1984;102:1011-14. 3. Airaksinen PJ, Drance SM, Schulzer M. Neuroretinal rim area in early glaucoma. Am J Ophthalmol 1985;99: 1-4. 4. Schwartz B. Cupping and pallor of the optic disc. Arch Ophthalmol 1973;89:272-7. 5. Sommer A, Katz J, Quigley HA, et al. Clinically detectable nerve fiber atrophy precedes the onset ofglaucomatous field loss. Arch Ophthalmol 1991;109:77-83. 6. Kahn HA, Leibowitz H, Ganley JP , et al. Standardizing diagnostic procedures. Am J Ophthalmol 1975;79:768-75. 7. Lichter PRo Variability ofexpert observers in evaluating the optic disc. Trans Am Ophthalmol Soc 1976;74:532-72. 8. T ielsch JM, Katz J, Quigley HA, et al. Intraobserver and interobserver agreement in measurement ofoptic disc characteristics. Ophthalmology 1988;95:350-6. 9. Varma R, Spaeth GL , Steinmann WC, Katz U. Agreement between clinicians and an image anal yzer in estimating cupto-disc ratios. Arch Ophthalmol 1989;107:526-9. 10. Varma R, Steinmann WC, Scott IV. Expert agreement in evaluating the optic disc for glaucoma. Ophthalmology 1992;99:215-21 . II. Gliklich RE, Steinmann WC, Spaeth GL. Visual field change in low-tension glaucoma over a five-year follow-up. Ophthalmology 1989;96:316-20. 12. Fleiss JL. Statistical Methods for Rates and Proportions. New York: Wiley, 1981;212-35. Abrams et al . Evaluating the Optic Disc for Glaucoma 13. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977;33:159-74. 14. Jonas JB , Gusek GC, Naumann GOH. Optic disc, cup and neuroretinal rim size , configuration and correlations in normal eyes. Invest Ophthalmol Vis Sci 1988;29: 1151- 8. 15. Varma R, Quigley HA, Pease ME . Changes in optic disk characteristics and the number of nerve fibers in experimental glaucoma. Am J Ophthalmol 1992;114 :554-9. 16. Pederson JE , Anderson DR . The mode of progressive disc cupping in ocular hypertension and glaucoma. Arch Ophthalmol 1980;98:490-5. 17. Gloster J. Incidence of optic disc haemorrhages in chronic simple glaucoma and ocular hypertension. Br J Ophthalmol 1981;65:452-6. 18. Wilensky JT, Kolker AE. Peripapillary changes in glaucoma. Am J OphthalmoI1976;81:341-5. 19. Miller KM , Quigley HA . The clinical appearance of the lamina cribrosa as a function of the extent of glaucomatous optic ner ve damage. Ophthalmology 1988;95:135-8. 20. Sommer A, Pollack I, Maumenee AE. Optic disc parameters and onset of glaucomatous field loss. I. Methods and progressive changes in disc morphology. Arch Ophthalmol 1979;97: 1444-8. 21. Read RM , Spaeth GL. Symposium: Glaucoma. The practical clinical appraisal of the optic disc in glaucoma: the natural history of cup progression and some specific discfield correlations. Trans Am Acad Ophthalmol Otolaryngol 1974;78:0P255-74. 22. Quigley HA, Addicks EM, Green WR. Optic nerve damage in human glaucoma. III. Quantitative correlation of nerve fiber loss and visual field defect in glaucoma, ischem ic neuropathy, papilledema, and toxic neuropathy. Arch Ophthalmol 1982;100:135-46. 1667