Current Eye Research, Early Online, 1–8, 2015 ! Informa Healthcare USA, Inc. ISSN: 0271-3683 print / 1460-2202 online DOI: 10.3109/02713683.2015.1020169
ORIGINAL ARTICLE
Evaluation of Visual Field Progression in Glaucoma: Quasar Regression Program and Event Analysis Valentı´n T. Dı´az-Alema´n, Marta Gonza´lez-Herna´ndez, Daniel Perera-Sanz and Karintia Armas-Domı´nguez
Curr Eye Res Downloaded from informahealthcare.com by Yale Dermatologic Surgery on 07/10/15 For personal use only.
Department of Ophthalmology, Hospital Universitario de Canarias, Santa Cruz de Tenerife, Spain
ABSTRACT Purpose: To determine the sensitivity, specificity and agreement between the Quasar program, glaucoma progression analysis (GPA II) event analysis and expert opinion in the detection of glaucomatous progression. Materials and methods: The Quasar program is based on linear regression analysis of both mean defect (MD) and pattern standard deviation (PSD). Each series of visual fields was evaluated by three methods; Quasar, GPA II and four experts. The sensitivity, specificity and agreement (kappa) for each method was calculated, using expert opinion as the reference standard. Results: The study included 439 SITA Standard visual fields of 56 eyes of 42 patients, with a mean of 7.8 ± 0.8 visual fields per eye. When suspected cases of progression were considered stable, sensitivity and specificity of Quasar, GPA II and the experts were 86.6% and 70.7%, 26.6% and 95.1%, and 86.6% and 92.6% respectively. When suspected cases of progression were considered as progressing, sensitivity and specificity of Quasar, GPA II and the experts were 79.1% and 81.2%, 45.8% and 90.6%, and 85.4% and 90.6% respectively. The agreement between Quasar and GPA II when suspected cases were considered stable or progressing was 0.03 and 0.28 respectively. The degree of agreement between Quasar and the experts when suspected cases were considered stable or progressing was 0.472 and 0.507. The degree of agreement between GPA II and the experts when suspected cases were considered stable or progressing was 0.262 and 0.342. Conclusions: The combination of MD and PSD regression analysis in the Quasar program showed better agreement with the experts and higher sensitivity than GPA II. Keywords: Event, glaucoma, progression, regression, visual field
INTRODUCTION
learning effect, changes in lens transparency and patient age.1 Mean defect (MD) presents less fluctuation than individual thresholds. There is general agreement that it is a sensitive procedure, but produces results too late. An additional limitation of this index is that it does not differentiate between focal and diffuse progression. Patients tend to suffer glaucoma at the same age as the appearance of cataracts, so increased MD is not always attributable to glaucoma. The study of progression using loss variance (LV) or its
Glaucoma is a chronic disease that causes irreversible loss of visual acuity. Early detection of visual field progression is a clinical problem that remains to be satisfactorily resolved. In this regard, most authors have highlighted the absence of a generally accepted reference standard. The subjective interpretation of series of visual fields has long been used to detect progression. This method may lead to errors induced by threshold fluctuation, the fatigue effect, the
Received 10 June 2014; revised 6 December 2014; accepted 9 February 2015; published online 10 April 2015 Correspondence: Valentı´n T. Dı´az-Alema´n, Ph.D., Department of Ophthalmology, Hospital Universitario de Canarias, CP 38230, Santa Cruz de Tenerife, Spain. Tel: +34 922678766. E-mail: [email protected]
1
Curr Eye Res Downloaded from informahealthcare.com by Yale Dermatologic Surgery on 07/10/15 For personal use only.
2
V. T. Dı´az-Alema´n et al.
equivalent in Humphrey perimetry – pattern standard deviation (PSD), could provide information on the focal or diffuse nature of progression when MD is below 15–17 dB. The lack of linearity limits the use of this index to differentiate between cataract and glaucomatous progression. Some investigators have reported better results with LV or PSD than with MD,2 while others have found no better detection capability.3 Other programs based on linear regression analysis of individual threshold values have demonstrated their ability to detect progression in patients with glaucoma. These include the Progressor program, Bebie curve sector analysis and Threshold Noiseless Trend (TNT).4–6 These programs employ mathematical filtering techniques that reduce threshold variability and visual field fluctuation, thus improving the detection of glaucomatous progression. A second group of programs is based on what is known as event analysis. An initial or baseline visual field is obtained from the first two or three tests, to define the field in which subsequent test results may change, taking into account their expected fluctuation. This is usually done estimating 5% and 95% percentiles that should not be exceeded in stable cases. When this does occur, the point is marked as possibly progressed from baseline, and if this occurs again in subsequent tests the diagnosis of progression acquires greater certainty. These programs include ‘‘Glaucoma Change Probability (GCP)’’, and ‘‘‘Glaucoma Progression Analysis (GPA II)’’’ designed for use in Humphrey perimeters, which analyze total deviations from normality or those observed in the Pattern Deviation Map. Our working group has developed a program called Quasar to evaluate visual field progression in patients with glaucoma, using regression analysis of both MD and PSD, hypothesizing that the combination of these two parameters should improve the sensitivity of each method separately. The present study was designed to test our hypothesis. The main objective was to compare the sensitivity and specificity of the Quasar program with that of GPA II. A secondary objective was to assess the degree of agreement (kappa) between each method and expert opinion taken as the reference standard.
MATERIALS AND METHODS Selection of the Study Sample We selected a sample of visual fields from patients diagnosed with primary open-angle, closed-angle, normal-tension, pigmentary and exfoliative glaucoma. Diagnosis was based on the following criteria: glaucomatous excavation of the optic nerve (including a thinning or ‘‘notch’’ located in the neuro-retinal
rim, cup-to-disc ratio 40.8 or verticalization of the cup), or asymmetry between visual fields and optic nerves of the two eyes (a difference greater than 0.2 in cup-to-disc ratio or 2 dB in MD between the two eyes). We excluded all patients with concomitant eye disease different from glaucoma, disease or systemic treatment that could affect the visual field, visual acuity less than 20/40, refractive error more than 5 diopters spherical equivalent or 3 diopters of astigmatism, rates of false positives, false negatives, and fixation errors equal to or greater than 25%, and those patients undergoing eye surgery during the follow-up period. Since progression estimated by regression procedures depends more on the number of tests performed than on monitoring time, a criterion for inclusion was a minimum of seven visual fields per eye and a minimum follow-up period of two years. Although all patients included in the study had visual field test experience, the results of the first two tests were excluded to avoid the learning effect. White-on-white perimetry was performed by two experienced clinicians using a Humphrey Field Analyzer II model 720i, with Goldmann size III stimuli and 24-2 SITA Standard strategy, with optical correction for observation distance from the perimeter dome.
Quasar Program This is based on the linear regression analysis of MD and PSD. The latter does not present a linear distribution, so the values of PSD were recalculated to transform them into a linear distribution, using a modification of the following formula for Octopus perimeters:3 If MD 4 16:33, then sLV ¼ sLV þ ð½MD 16:33=0:84Þ This formula had to be modified for application to Humphrey visual fields, using a formula for conversion between Octopus and Humphrey threshold values and indices,7 and the resulting formula was as follows If MD 5 17:35, then PSD ¼ PSD þ ð½MD þ 17:35=0:84Þ The Quasar program indicates progression if it detects a significant change in MD or PSD. It also calculates MD change per year and predicts visual field loss (in years) if the tendency is not corrected (Figure 1). The Quasar program was developed by our research group, affiliated to the University of La Laguna, Tenerife and the University Hospital of the Canary Islands. It was programmed in Python 2.7.3 language, in Linux Fedora 17 environment. Current Eye Research
Regression Analysis and Event Analysis
3
Curr Eye Res Downloaded from informahealthcare.com by Yale Dermatologic Surgery on 07/10/15 For personal use only.
FIGURE 1 Quasar program panel with an example of progression.
Glaucoma Progression Analysis (GPA II) This software was developed by Humphrey for use in Humphrey II 700 series perimeters. GPA II uses event analysis to determine glaucomatous progression. This strategy accepts as change (p50.05) deviation from baseline values and variability in one or more locations of consecutive visual fields. The mean of the first two visual field tests is taken as the baseline value representing the initial state of the patient. To confirm progression or ‘‘likely progression’’, GPA II accepts as sufficient the observation of significant change (p50.05) in at least three locations on three consecutive tests. The program considers suspected progression or ‘‘possible progression’’, as significant change occurring in at least three locations on two consecutive tests. Professor Dr B. Bengtsson and collaborators of the Early Manifest Glaucoma Trial (EMGT) group validated these criteria to define glaucomatous progression and differentiate change caused by glaucoma from that caused by cataract.8,9
Reference Standard For the purposes of the present study, the subjective opinion of four experts in glaucoma was taken as the reference standard. Each expert independently evaluated visual field results for glaucomatous progression in our selected cases. The only three possible answers were ‘‘Yes’’, and ‘‘No’’, and ‘‘suspected progression’’. The results were recorded on a MicrosoftÕ Excel sheet for subsequent comparison with each other and with the Quasar and GPA II results. Progression was deemed to exist when at least three experts independently detected glaucomatous progression. When only one or two experts detected progression, the case was considered as suspected progression. No progression, or visual field stability, !
2015 Informa Healthcare USA, Inc.
was deemed to exist when none of the experts detected progression. The sensitivity and specificity of Quasar, GPA II and Experts were calculated using two different criteria; considering suspected progression as stable or as ‘‘progression’’. In addition, we assessed interobserver agreement as well as agreement between each expert and the two programs, using the kappa index. The study was performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki. Informed consent was obtained from all participants and the study was approved by the Research Ethics Committee of our hospital.
RESULTS The study included a total of 439 SITA Standard visual fields from 56 eyes of 42 patients, 23 (54.8%) men and 19 (45.2%) women, with a mean 7.8 ± 0.8 visual fields per eye. Mean age was 70.4 ± 12 years; mean followup was 5.48 ± 1.04 years and initial or baseline MD was 3.44 ± 4.0 dB.
Reference Standard Separate evaluation of the 56 eyes by the experts 1, 2, 3 and 4 respectively produced the following results, as shown in Table 1, no progression 27 (48.2%), 35 (62.5%) 34 (60.7%) and 34 (60.7%); progression 19 (33.9%), 16 (28.6%), 10 (17.8%) and 19 (33.9%); suspected progression 10 (17.8%), 5 (8.9%), 12 (21.4%) and 3 (5.3%). Taken together, the experts detected stability in 32 eyes (57.1%), suspected progression in nine eyes (16.0%) and progression in 15 eyes (26.7%). On considering the suspected cases as stable, mean sensitivity was 86.6%, and mean specificity was
4
V. T. Dı´az-Alema´n et al.
TABLE 1 Experts results. Expert 1 2 3 4
Progression [n (%)] 19 16 10 19
(33.9) (28.6) (17.8) (33.9)
TABLE 3 Baseline mean defect.
No progression [n (%)] 27 35 34 34
Suspected progression [n (%)]
(48.2) (62.5) (60.7) (60.7)
10 (17.8) 5 (8.9) 12 (21.4) 3 (5.3)
Gold standard a
4.45 (4.21 ) 3.41 (4.17)
Progressing Stable
GPA II
Quasar
2.06 (2.73) 3.75 (4.40)
3.5 (4.01) 3.74 (4.08)
a
Standard deviation.
TABLE 4 Rates of kappa concordance when suspected cases were considered to be progressing. TABLE 2 Sensitivity and specificity of the experts, GPA II and Quasar.
Curr Eye Res Downloaded from informahealthcare.com by Yale Dermatologic Surgery on 07/10/15 For personal use only.
Suspected cases = stable
Expert 1 Expert 2 Expert 3 Expert 4 GPA II Quasar
Suspected cases = progressing
Sensitivity (%)
Specificity (%)
Sensitivity (%)
Specificity (%)
93.3 93.3 60.0 100 26.6 86.6
87.8 95.1 97.5 90.2 95.1 70.7
95.8 79.1 87.5 79.1 45.8 79.1
81.2 93.7 96.8 90.6 90.6 81.2
92.6%. On considering suspected cases as progressing, mean sensitivity was 85.4%, and mean specificity was 90.6%. Table 2 shows the sensitivity and specificity of the reference standard, GPA II and Quasar. In cases evaluated by the experts as glaucomatous progression, baseline MD was 4.45 dB with a rate of progression of 1.22 dB per year. In cases evaluated as stable, baseline MD was 3.41 dB with a rate of progression of 0.08 dB per year.
Quasar Our program detected stability in 31 eyes (55.4%), and progression in 25 eyes (44.6%). On considering suspected cases as stable, the Quasar program showed a sensitivity of 86.6%, and specificity of 70.7%. On considering suspected cases as progressing, sensitivity was 79.1%, and specificity was 81.2%. In cases in which Quasar detected glaucomatous progression, average baseline MD was 3.50 dB with a rate of progression of 0.79 dB per year. Significant differences were observed (p50.05, t-test) between Quasar and the reference standard in the rate of progression of these cases. In cases evaluated as stable, average baseline MD was 3.74 dB with a rate of progression of 0.03 dB per year.
Program GPA II This strategy detected stability in 42 cases (75%), suspected progression in six cases (10.8%) and progression in eight (14.2%) cases.
Expert 1 Expert 2 Expert 3 Expert 4 Quasar GPA Expert 1 Expert 2 Expert 3 Expert 4 Quasar GPA
x 0.58 0.61 0.61 0.50 0.26
0.58 x 0.58 0.66 0.41 0.31
0.61 0.58 x 0.63 0.52 0.36
0.61 0.66 0.63 x 0.60 0.44
0.50 0.41 0.52 0.60 x 0.28
0.26 0.31 0.36 0.44 0.28 x
On considering suspected cases as stable, GPA II showed a sensitivity of 26.6%, and specificity of 95.1%. On considering suspected cases as progressing, sensitivity was 45.8%, and specificity was 90.6%. In cases in which GPA II detected progression, average baseline MD was 2.06 dB with a rate of progression of 1.05 dB per year. In cases in which stability was detected, average baseline MD was 3.75 dB with a rate of progression of 0.20 dB per year. Baseline MD of cases with progression and those evaluated as stable for the reference standard, Quasar and GPA II are shown in Table 3. There were no significant differences in baseline MD. Finally, of the 25 eyes in which Quasar detected progression, significant change in MD was detected in 22, and significant change in PSD was detected in only eight eyes. Of these eight cases, five also showed significant change in MD. In only three cases, the detection of progression was based on significant change in PSD. The rate of agreement (kappa) between the four experts on considering suspected cases as stable or progressing was 0.615 and 0.611, respectively. In general, Quasar obtained better agreement with our reference standard than GPA II: Quasar showed better agreement when suspected cases were considered as progressing (0.507) than when considered as stable (0.472). GPA II showed worse agreement with the experts when suspected cases were considered as stable (0.262) than when considered as progressing (0.342). Agreement between the two programs was low: 0.28 when suspected cases were considered to be stable and 0.03 when they were considered to be progressing. Tables 4 and 5 show the rates of concordance with the experts for the two programs when suspected Current Eye Research
Regression Analysis and Event Analysis TABLE 5 Rates of kappa concordance when suspected cases were considered to be stable. Expert 1 Expert 2 Expert 3 Expert 4 Quasar GPA Expert 1 Expert 2 Expert 3 Expert 4 Quasar GPA
x 0.71 0.50 0.76 0.56 0.19
0.71 x 0.51 0.71 0.51 0.25
0.50 0.51 x 0.50 0.19 0.42
0.76 0.71 0.50 x 0.63 0.19
0.56 0.51 0.19 0.63 x 0.03
0.19 0.25 0.42 0.19 0.03 x
Curr Eye Res Downloaded from informahealthcare.com by Yale Dermatologic Surgery on 07/10/15 For personal use only.
cases were considered to be progressing and when considered to be stable.
DISCUSSION The lack of a generally accepted reference standard that defines glaucomatous progression is a major challenge for this type of study. The opinion of experts with experience in glaucoma is a valid option but is not risk-free. In the present study the experts obtained moderate rates of agreement10 (0.615 and 0.611) when eyes with suspected progression were considered stable or progressing, superior in both cases to that obtained by the software programs. Variability in the interpretation of visual fields in patients with glaucoma has been reported by various authors. The degree of agreement depends on the number of experts, follow-up time and the number of visual fields in each series. Tanna et al.11 and Viswanathan et al.,12 found inter-expert agreement rates of 0.45 and 0.32, respectively. The study by Viswanathan et al. included 27 cases with 16 visual fields per case, evaluated by five experts; the authors concluded that a high number of visual fields per patient and long follow-up periods can hinder and complicate the detection of progression. In our study, inter-expert agreement was greater, possibly due to the lower number of visual fields per case. In cases where the experts and programs detected progression, we measured the rate of progression in decibels (dB) per year. For our experts, the rate of progression was 1.22 dB per year, higher than the rates obtained with GPA II and Quasar (1.09 and 0.79 dB per year, respectively). As mentioned, the difference in progression rate between the experts and the Quasar program was significant. This difference suggests that our experts required greater change in MD than Quasar to detect progression. The lack of sensitivity affected the specificity of the programs used. On the one hand, it penalized the specificity of the most sensitive program (Quasar), and on the other it favored the specificity of the less sensitive program (GPA II). The experts detected progression in cases with higher initial MD than the programs. Although the !
2015 Informa Healthcare USA, Inc.
5
difference was not significant, this suggests that the experts may have confused visual field severity with progression. If this were the case, it would compromise the specificity of the most sensitive program. When suspected cases were considered as stable or progressing in the present study, GPA II showed low sensitivity (26.6% and 45.8%, respectively) and high specificity (95.1% and 90.6% respectively). Several authors have studied the sensitivity and specificity of GPA II event analysis, with contradictory results. Ang et al.,13 using the opinion of four experts as their reference standard, found rates of sensitivity and specificity (41% and 90%, respectively) that were very similar to our values. However other authors1,14 have obtained higher rates of sensitivity for GPA II when the reference standard comprised experts using objective systems to detect progression such as those of Hodapp et al.15 or Boden et al.16 based on pattern deviation plot. Pattern deviation analysis may underestimate the number of cases with progression since it does not detect diffuse changes of glaucomatous progression17–19 or changes in very advanced cases where thresholds are very low or out of range;14 it may also detect progression in cases of perimetric fluctuation. The inability to detect atypical or initial diffuse glaucomatous progression5 and changes in advanced glaucoma, coupled with the use of very strict criteria to detect progression explains why GPA II in our work presented low sensitivity when compared to a reference standard using subjective criteria of progression. GPA II event analysis showed high values for specificity; 95.1% when suspected cases were considered stable, and 90.6% when considered as progressing. These values are explained by the low sensitivity of the program which results in stable cases being considered as progressing (Figure 2). The Quasar program obtained higher sensitivity (86.6% and 79.1%) than GPA II event analysis and better kappa indices of agreement, according to our reference standard. Several studies have analyzed the evolution of MD in patients with glaucoma. The general consensus is that this is a sensitive procedure, but produces results too late.3,20 Being an averaged value, MD presents less fluctuation than individual threshold values and provides information on visual field changes or the learning effect. The drawback is that it fails to inform about the type of progression (focal or diffuse) in patients at an age when glaucoma and cataract often coexist. To compensate for this limitation, PSD analysis has been proposed since, in theory, this method provides information about the focal or diffuse nature of progression. For some authors PSD analysis yields favorable results,2 while others have not found it useful to detect progression.3 In our work, PSD analysis was less effective than MD analysis, only detecting progression in eight cases. However, the combination of the two parameters
V. T. Dı´az-Alema´n et al.
Curr Eye Res Downloaded from informahealthcare.com by Yale Dermatologic Surgery on 07/10/15 For personal use only.
6
FIGURE 2 In this image you may see an example with the first and last visual fields of a series of seven cases in 3.3 years. (A) Reference standard: the four examiners detected progression. (B) Quasar detects significant change. (C) GPA does not detect significant change.
resulted in higher overall sensitivity than for each one separately and for GPA II event analysis. The combination of criteria to detect progression has already been used by various investigators.
The TNT program employs three criteria to indicate progression based on linear regression analysis of MD, number of visual field points showing significant change, and significant change in eight sectors of the Current Eye Research
Curr Eye Res Downloaded from informahealthcare.com by Yale Dermatologic Surgery on 07/10/15 For personal use only.
Regression Analysis and Event Analysis Bebie curve. Progression is considered to exist when two or three criteria are positive, or if the one positive criterion is repeated on two successive examinations. The TNT program has demonstrated that the combination of these three criteria is superior to regression analysis of isolated variables such as MD, neuroretinal rim area, Glaucoma Probability Score and mean nerve fiber layer thickness (NFLT) for the early detection of progression in cases of incipient glaucoma.5 TNT has also shown greater efficacy than GPA for detecting glaucomatous progression.21 It could be interesting to compare Quasar and TNT in the future. Finally, Glaucoma Staging System (GSS) is a method of classifying visual field impairment, proposed by Brusini22 in 1995. It uses MD and PSD/LV values to classify severity (Stage 0, 1, 2, 3, 4, 5) and type of damage (generalized, mixed and localized defects). This method has proved useful in routine clinical practice and in the field of research but, as they authors have pointed out, it should not be used to determine significant progression. GSS provides useful information about MD trends over time, but the change from one stage to another cannot be automatically considered as significant progression.23 The present study has certain limitations. The first, as in many other similar studies, is the lack of a reference standard that defines progression. The second is the absence of morphological evidence to assess progression. We believe such evidence is important, although visual field tests are currently the most widely used method to determine glaucomatous progression.5 In conclusion, the present work shows that the combined use of the two parameters (MD and PSD) in the Quasar program improves the individual sensitivity of each parameter alone and that of GPA II event analysis. The specificity of the Quasar program was adversely affected by the choice of a reference standard with low sensitivity, which favored the specificity of the GPA II program. Finally, Quasar showed better agreement10 with the experts than GPA II. Further studies are needed to confirm these data.
DECLARATION OF INTEREST The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.
REFERENCES 1. Arnalich-Montiel F, Casas-Llera P, Mun˜oz-Negrete FJ, Rebolleda G. Performance of glaucoma progression analysis software in a glaucoma population. Graefes Arch Clin Exp Ophthalmol 2009;247:391–397. !
2015 Informa Healthcare USA, Inc.
7
2. Chauhan BC, Drance SM, Douglas GR. The use of visual field indices in detecting changes in the visual field in glaucoma. Invest Ophthalmol Vis Sci 1990;31: 512–520. 3. Gonza´lez De La Rosa M, Gonza´lez-Hernandez M, Dı´azAleman T. Linear regression analysis of the cumulative defect curve by sectors and other criteria of glaucomatous visual field progression. Eur J Ophthalmol 2009;19: 416–424. 4. Viswanathan AC, Fitzke FW, Hitchings RA. Early detection of visual field progression in glaucoma: a comparison of PROGRESSOR and STATPAC 2. Br J Ophthalmol 1997;81: 1037–1042. 5. Gonza´lez de la Rosa M, Gonza´lez-Hernandez M, Sa´nchez-Ma´ndez M, Medina-Mesa E, Rodriguez de la Vega R. Detection of morphological and functional progression in initial glaucoma. Br J Ophthalmol 2010; 94:414–418. 6. Gonza´lez de la Rosa M, Dı´az-Alema´n T, Gonza´lezHerna´ndez M, Jerez-Fidalgo M. Specificity of the program threshold noiseless trend for perimetric progression analysis. Curr Eye Res 2010;35:302–307. 7. Zeyen T, Roche M, Brigatti L, Caprioli J. Formulas for conversion between Octopus and Humphrey threshold values and indices. Graefes Arch Clin Exp Ophthalmol 1995;233:627–634. 8. Bengtsson B, Lindgren A, Heijl A, Lindgren G, Asman P, Patella M. Perimetric probability maps to separate change caused by glaucoma from that caused by cataract. Acta Ophthalmol Scand 1997;75:184–188. 9. Leske MC, Heijl A, Hussein M, Bengtsson B, Hyman L, Komaroff E. Factors for glaucoma progression and the effect of treatment. The Early Manifest Glaucoma Trial. Arch Ophthalmol 2003;121:48–56. 10. Lo´pez de Ullibarri, Galparsoro I, Pita Ferna´ndez S. Medidas de concordancia: el ´ındice de Kappa. Cad Aten Primaria 1999;6:169–171. 11. Tanna AP, Bandi JR, Budenz DL, Feuer WJ, Feldman RM, Herndon LW, et al. Interobserver agreement and intraobserver reproducibility of the subjective determination of glaucomatous visual field progression. Ophthalmology 2011;118:60–65. 12. Viswanathan AC, Crabb DP, McNaught AI, Westcott MC, Kamal D, Garway-Heath DF, et al. Interobserver agreement on visual field progression in glaucoma: a comparison of methods. Br J Ophthalmol 2003;87: 726–730. 13. Ang GS, Mustafa MS, Scott N, Diaz-Aleman VT, AzuaraBlanco A. Perimetric progression in open angle glaucoma and the Visual Field Index (VFI). J Glaucoma 2011;20: 223–227. 14. Anto´n A, Pazos M, Martı´n B, Navero JM, Ayala ME, Castany M, et al. Glaucoma progression detection: agreement, sensitivity, and specificity of expert visual field evaluation, event analysis, and trend analysis. Eur J Ophthalmol 2013;23:187–195. 15. Hodapp E, Parrish RK, Anderson DR. Clinical decisions in glaucoma. St Louis: CV Mosby; 1993. 16. Boden C, Blumenthal EZ, Pascual J, McEwan G, Weinreb RN, Medeiros F, et al. Patterns of glaucomatous visual field progression identified by three progression criteria. Am J Ophthalmol 2004;138:1029–1036. 17. Bengtsson B, Lindgren A, Heijl A, Lindgren G, Asman P, Patella M. Perimetric probability maps to separate change caused by glaucoma from that caused by cataract. Acta Ophthalmol Scand 1997;75:184–188. 18. Katz J. A comparison of the pattern and total deviation based Glaucoma Change Probability programs. Invest Ophthalmol Vis Sci 2000;41:1012–1016.
8
V. T. Dı´az-Alema´n et al. visual field deterioration by Glaucoma Progression Analysis and Threshold Noiseless Trend Programs. Br J Ophthalmol 2009;93:322–328. 22. Brusini P. Clinical use of a new method for visual field damage classification in glaucoma. Eur J Ophthalmol 1996; 6:402–407. 23. Brusini P, Filacorda S. Enhanced Glaucoma Staging System (GSS 2) for classifying functional damage in glaucoma. J Glaucoma 2006;15:40–46.
Curr Eye Res Downloaded from informahealthcare.com by Yale Dermatologic Surgery on 07/10/15 For personal use only.
19. Artes PH, Nicolela MT, LeBlanc RP, Chauhan BC. Visual field progression in glaucoma: total versus pattern deviation analyses. Invest Ophthalmol Vis Sci 2005;46: 4600–4606. 20. Nouri-Mahdavi K, Brigatti L, Weitzman M, Caprioli J. Comparison of methods to detect visual field progression in glaucoma. Ophthalmology 1997;104:1228–1236. 21. Diaz-Aleman VT, Anton A, Gonzalez de la Rosa M, Johnson ZK, McLeod S, Azuara-Blanco A. Detection of
Current Eye Research
ORIGINAL ARTICLE
Evaluation of Visual Field Progression in Glaucoma: Quasar Regression Program and Event Analysis Valentı´n T. Dı´az-Alema´n, Marta Gonza´lez-Herna´ndez, Daniel Perera-Sanz and Karintia Armas-Domı´nguez
Curr Eye Res Downloaded from informahealthcare.com by Yale Dermatologic Surgery on 07/10/15 For personal use only.
Department of Ophthalmology, Hospital Universitario de Canarias, Santa Cruz de Tenerife, Spain
ABSTRACT Purpose: To determine the sensitivity, specificity and agreement between the Quasar program, glaucoma progression analysis (GPA II) event analysis and expert opinion in the detection of glaucomatous progression. Materials and methods: The Quasar program is based on linear regression analysis of both mean defect (MD) and pattern standard deviation (PSD). Each series of visual fields was evaluated by three methods; Quasar, GPA II and four experts. The sensitivity, specificity and agreement (kappa) for each method was calculated, using expert opinion as the reference standard. Results: The study included 439 SITA Standard visual fields of 56 eyes of 42 patients, with a mean of 7.8 ± 0.8 visual fields per eye. When suspected cases of progression were considered stable, sensitivity and specificity of Quasar, GPA II and the experts were 86.6% and 70.7%, 26.6% and 95.1%, and 86.6% and 92.6% respectively. When suspected cases of progression were considered as progressing, sensitivity and specificity of Quasar, GPA II and the experts were 79.1% and 81.2%, 45.8% and 90.6%, and 85.4% and 90.6% respectively. The agreement between Quasar and GPA II when suspected cases were considered stable or progressing was 0.03 and 0.28 respectively. The degree of agreement between Quasar and the experts when suspected cases were considered stable or progressing was 0.472 and 0.507. The degree of agreement between GPA II and the experts when suspected cases were considered stable or progressing was 0.262 and 0.342. Conclusions: The combination of MD and PSD regression analysis in the Quasar program showed better agreement with the experts and higher sensitivity than GPA II. Keywords: Event, glaucoma, progression, regression, visual field
INTRODUCTION
learning effect, changes in lens transparency and patient age.1 Mean defect (MD) presents less fluctuation than individual thresholds. There is general agreement that it is a sensitive procedure, but produces results too late. An additional limitation of this index is that it does not differentiate between focal and diffuse progression. Patients tend to suffer glaucoma at the same age as the appearance of cataracts, so increased MD is not always attributable to glaucoma. The study of progression using loss variance (LV) or its
Glaucoma is a chronic disease that causes irreversible loss of visual acuity. Early detection of visual field progression is a clinical problem that remains to be satisfactorily resolved. In this regard, most authors have highlighted the absence of a generally accepted reference standard. The subjective interpretation of series of visual fields has long been used to detect progression. This method may lead to errors induced by threshold fluctuation, the fatigue effect, the
Received 10 June 2014; revised 6 December 2014; accepted 9 February 2015; published online 10 April 2015 Correspondence: Valentı´n T. Dı´az-Alema´n, Ph.D., Department of Ophthalmology, Hospital Universitario de Canarias, CP 38230, Santa Cruz de Tenerife, Spain. Tel: +34 922678766. E-mail: [email protected]
1
Curr Eye Res Downloaded from informahealthcare.com by Yale Dermatologic Surgery on 07/10/15 For personal use only.
2
V. T. Dı´az-Alema´n et al.
equivalent in Humphrey perimetry – pattern standard deviation (PSD), could provide information on the focal or diffuse nature of progression when MD is below 15–17 dB. The lack of linearity limits the use of this index to differentiate between cataract and glaucomatous progression. Some investigators have reported better results with LV or PSD than with MD,2 while others have found no better detection capability.3 Other programs based on linear regression analysis of individual threshold values have demonstrated their ability to detect progression in patients with glaucoma. These include the Progressor program, Bebie curve sector analysis and Threshold Noiseless Trend (TNT).4–6 These programs employ mathematical filtering techniques that reduce threshold variability and visual field fluctuation, thus improving the detection of glaucomatous progression. A second group of programs is based on what is known as event analysis. An initial or baseline visual field is obtained from the first two or three tests, to define the field in which subsequent test results may change, taking into account their expected fluctuation. This is usually done estimating 5% and 95% percentiles that should not be exceeded in stable cases. When this does occur, the point is marked as possibly progressed from baseline, and if this occurs again in subsequent tests the diagnosis of progression acquires greater certainty. These programs include ‘‘Glaucoma Change Probability (GCP)’’, and ‘‘‘Glaucoma Progression Analysis (GPA II)’’’ designed for use in Humphrey perimeters, which analyze total deviations from normality or those observed in the Pattern Deviation Map. Our working group has developed a program called Quasar to evaluate visual field progression in patients with glaucoma, using regression analysis of both MD and PSD, hypothesizing that the combination of these two parameters should improve the sensitivity of each method separately. The present study was designed to test our hypothesis. The main objective was to compare the sensitivity and specificity of the Quasar program with that of GPA II. A secondary objective was to assess the degree of agreement (kappa) between each method and expert opinion taken as the reference standard.
MATERIALS AND METHODS Selection of the Study Sample We selected a sample of visual fields from patients diagnosed with primary open-angle, closed-angle, normal-tension, pigmentary and exfoliative glaucoma. Diagnosis was based on the following criteria: glaucomatous excavation of the optic nerve (including a thinning or ‘‘notch’’ located in the neuro-retinal
rim, cup-to-disc ratio 40.8 or verticalization of the cup), or asymmetry between visual fields and optic nerves of the two eyes (a difference greater than 0.2 in cup-to-disc ratio or 2 dB in MD between the two eyes). We excluded all patients with concomitant eye disease different from glaucoma, disease or systemic treatment that could affect the visual field, visual acuity less than 20/40, refractive error more than 5 diopters spherical equivalent or 3 diopters of astigmatism, rates of false positives, false negatives, and fixation errors equal to or greater than 25%, and those patients undergoing eye surgery during the follow-up period. Since progression estimated by regression procedures depends more on the number of tests performed than on monitoring time, a criterion for inclusion was a minimum of seven visual fields per eye and a minimum follow-up period of two years. Although all patients included in the study had visual field test experience, the results of the first two tests were excluded to avoid the learning effect. White-on-white perimetry was performed by two experienced clinicians using a Humphrey Field Analyzer II model 720i, with Goldmann size III stimuli and 24-2 SITA Standard strategy, with optical correction for observation distance from the perimeter dome.
Quasar Program This is based on the linear regression analysis of MD and PSD. The latter does not present a linear distribution, so the values of PSD were recalculated to transform them into a linear distribution, using a modification of the following formula for Octopus perimeters:3 If MD 4 16:33, then sLV ¼ sLV þ ð½MD 16:33=0:84Þ This formula had to be modified for application to Humphrey visual fields, using a formula for conversion between Octopus and Humphrey threshold values and indices,7 and the resulting formula was as follows If MD 5 17:35, then PSD ¼ PSD þ ð½MD þ 17:35=0:84Þ The Quasar program indicates progression if it detects a significant change in MD or PSD. It also calculates MD change per year and predicts visual field loss (in years) if the tendency is not corrected (Figure 1). The Quasar program was developed by our research group, affiliated to the University of La Laguna, Tenerife and the University Hospital of the Canary Islands. It was programmed in Python 2.7.3 language, in Linux Fedora 17 environment. Current Eye Research
Regression Analysis and Event Analysis
3
Curr Eye Res Downloaded from informahealthcare.com by Yale Dermatologic Surgery on 07/10/15 For personal use only.
FIGURE 1 Quasar program panel with an example of progression.
Glaucoma Progression Analysis (GPA II) This software was developed by Humphrey for use in Humphrey II 700 series perimeters. GPA II uses event analysis to determine glaucomatous progression. This strategy accepts as change (p50.05) deviation from baseline values and variability in one or more locations of consecutive visual fields. The mean of the first two visual field tests is taken as the baseline value representing the initial state of the patient. To confirm progression or ‘‘likely progression’’, GPA II accepts as sufficient the observation of significant change (p50.05) in at least three locations on three consecutive tests. The program considers suspected progression or ‘‘possible progression’’, as significant change occurring in at least three locations on two consecutive tests. Professor Dr B. Bengtsson and collaborators of the Early Manifest Glaucoma Trial (EMGT) group validated these criteria to define glaucomatous progression and differentiate change caused by glaucoma from that caused by cataract.8,9
Reference Standard For the purposes of the present study, the subjective opinion of four experts in glaucoma was taken as the reference standard. Each expert independently evaluated visual field results for glaucomatous progression in our selected cases. The only three possible answers were ‘‘Yes’’, and ‘‘No’’, and ‘‘suspected progression’’. The results were recorded on a MicrosoftÕ Excel sheet for subsequent comparison with each other and with the Quasar and GPA II results. Progression was deemed to exist when at least three experts independently detected glaucomatous progression. When only one or two experts detected progression, the case was considered as suspected progression. No progression, or visual field stability, !
2015 Informa Healthcare USA, Inc.
was deemed to exist when none of the experts detected progression. The sensitivity and specificity of Quasar, GPA II and Experts were calculated using two different criteria; considering suspected progression as stable or as ‘‘progression’’. In addition, we assessed interobserver agreement as well as agreement between each expert and the two programs, using the kappa index. The study was performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki. Informed consent was obtained from all participants and the study was approved by the Research Ethics Committee of our hospital.
RESULTS The study included a total of 439 SITA Standard visual fields from 56 eyes of 42 patients, 23 (54.8%) men and 19 (45.2%) women, with a mean 7.8 ± 0.8 visual fields per eye. Mean age was 70.4 ± 12 years; mean followup was 5.48 ± 1.04 years and initial or baseline MD was 3.44 ± 4.0 dB.
Reference Standard Separate evaluation of the 56 eyes by the experts 1, 2, 3 and 4 respectively produced the following results, as shown in Table 1, no progression 27 (48.2%), 35 (62.5%) 34 (60.7%) and 34 (60.7%); progression 19 (33.9%), 16 (28.6%), 10 (17.8%) and 19 (33.9%); suspected progression 10 (17.8%), 5 (8.9%), 12 (21.4%) and 3 (5.3%). Taken together, the experts detected stability in 32 eyes (57.1%), suspected progression in nine eyes (16.0%) and progression in 15 eyes (26.7%). On considering the suspected cases as stable, mean sensitivity was 86.6%, and mean specificity was
4
V. T. Dı´az-Alema´n et al.
TABLE 1 Experts results. Expert 1 2 3 4
Progression [n (%)] 19 16 10 19
(33.9) (28.6) (17.8) (33.9)
TABLE 3 Baseline mean defect.
No progression [n (%)] 27 35 34 34
Suspected progression [n (%)]
(48.2) (62.5) (60.7) (60.7)
10 (17.8) 5 (8.9) 12 (21.4) 3 (5.3)
Gold standard a
4.45 (4.21 ) 3.41 (4.17)
Progressing Stable
GPA II
Quasar
2.06 (2.73) 3.75 (4.40)
3.5 (4.01) 3.74 (4.08)
a
Standard deviation.
TABLE 4 Rates of kappa concordance when suspected cases were considered to be progressing. TABLE 2 Sensitivity and specificity of the experts, GPA II and Quasar.
Curr Eye Res Downloaded from informahealthcare.com by Yale Dermatologic Surgery on 07/10/15 For personal use only.
Suspected cases = stable
Expert 1 Expert 2 Expert 3 Expert 4 GPA II Quasar
Suspected cases = progressing
Sensitivity (%)
Specificity (%)
Sensitivity (%)
Specificity (%)
93.3 93.3 60.0 100 26.6 86.6
87.8 95.1 97.5 90.2 95.1 70.7
95.8 79.1 87.5 79.1 45.8 79.1
81.2 93.7 96.8 90.6 90.6 81.2
92.6%. On considering suspected cases as progressing, mean sensitivity was 85.4%, and mean specificity was 90.6%. Table 2 shows the sensitivity and specificity of the reference standard, GPA II and Quasar. In cases evaluated by the experts as glaucomatous progression, baseline MD was 4.45 dB with a rate of progression of 1.22 dB per year. In cases evaluated as stable, baseline MD was 3.41 dB with a rate of progression of 0.08 dB per year.
Quasar Our program detected stability in 31 eyes (55.4%), and progression in 25 eyes (44.6%). On considering suspected cases as stable, the Quasar program showed a sensitivity of 86.6%, and specificity of 70.7%. On considering suspected cases as progressing, sensitivity was 79.1%, and specificity was 81.2%. In cases in which Quasar detected glaucomatous progression, average baseline MD was 3.50 dB with a rate of progression of 0.79 dB per year. Significant differences were observed (p50.05, t-test) between Quasar and the reference standard in the rate of progression of these cases. In cases evaluated as stable, average baseline MD was 3.74 dB with a rate of progression of 0.03 dB per year.
Program GPA II This strategy detected stability in 42 cases (75%), suspected progression in six cases (10.8%) and progression in eight (14.2%) cases.
Expert 1 Expert 2 Expert 3 Expert 4 Quasar GPA Expert 1 Expert 2 Expert 3 Expert 4 Quasar GPA
x 0.58 0.61 0.61 0.50 0.26
0.58 x 0.58 0.66 0.41 0.31
0.61 0.58 x 0.63 0.52 0.36
0.61 0.66 0.63 x 0.60 0.44
0.50 0.41 0.52 0.60 x 0.28
0.26 0.31 0.36 0.44 0.28 x
On considering suspected cases as stable, GPA II showed a sensitivity of 26.6%, and specificity of 95.1%. On considering suspected cases as progressing, sensitivity was 45.8%, and specificity was 90.6%. In cases in which GPA II detected progression, average baseline MD was 2.06 dB with a rate of progression of 1.05 dB per year. In cases in which stability was detected, average baseline MD was 3.75 dB with a rate of progression of 0.20 dB per year. Baseline MD of cases with progression and those evaluated as stable for the reference standard, Quasar and GPA II are shown in Table 3. There were no significant differences in baseline MD. Finally, of the 25 eyes in which Quasar detected progression, significant change in MD was detected in 22, and significant change in PSD was detected in only eight eyes. Of these eight cases, five also showed significant change in MD. In only three cases, the detection of progression was based on significant change in PSD. The rate of agreement (kappa) between the four experts on considering suspected cases as stable or progressing was 0.615 and 0.611, respectively. In general, Quasar obtained better agreement with our reference standard than GPA II: Quasar showed better agreement when suspected cases were considered as progressing (0.507) than when considered as stable (0.472). GPA II showed worse agreement with the experts when suspected cases were considered as stable (0.262) than when considered as progressing (0.342). Agreement between the two programs was low: 0.28 when suspected cases were considered to be stable and 0.03 when they were considered to be progressing. Tables 4 and 5 show the rates of concordance with the experts for the two programs when suspected Current Eye Research
Regression Analysis and Event Analysis TABLE 5 Rates of kappa concordance when suspected cases were considered to be stable. Expert 1 Expert 2 Expert 3 Expert 4 Quasar GPA Expert 1 Expert 2 Expert 3 Expert 4 Quasar GPA
x 0.71 0.50 0.76 0.56 0.19
0.71 x 0.51 0.71 0.51 0.25
0.50 0.51 x 0.50 0.19 0.42
0.76 0.71 0.50 x 0.63 0.19
0.56 0.51 0.19 0.63 x 0.03
0.19 0.25 0.42 0.19 0.03 x
Curr Eye Res Downloaded from informahealthcare.com by Yale Dermatologic Surgery on 07/10/15 For personal use only.
cases were considered to be progressing and when considered to be stable.
DISCUSSION The lack of a generally accepted reference standard that defines glaucomatous progression is a major challenge for this type of study. The opinion of experts with experience in glaucoma is a valid option but is not risk-free. In the present study the experts obtained moderate rates of agreement10 (0.615 and 0.611) when eyes with suspected progression were considered stable or progressing, superior in both cases to that obtained by the software programs. Variability in the interpretation of visual fields in patients with glaucoma has been reported by various authors. The degree of agreement depends on the number of experts, follow-up time and the number of visual fields in each series. Tanna et al.11 and Viswanathan et al.,12 found inter-expert agreement rates of 0.45 and 0.32, respectively. The study by Viswanathan et al. included 27 cases with 16 visual fields per case, evaluated by five experts; the authors concluded that a high number of visual fields per patient and long follow-up periods can hinder and complicate the detection of progression. In our study, inter-expert agreement was greater, possibly due to the lower number of visual fields per case. In cases where the experts and programs detected progression, we measured the rate of progression in decibels (dB) per year. For our experts, the rate of progression was 1.22 dB per year, higher than the rates obtained with GPA II and Quasar (1.09 and 0.79 dB per year, respectively). As mentioned, the difference in progression rate between the experts and the Quasar program was significant. This difference suggests that our experts required greater change in MD than Quasar to detect progression. The lack of sensitivity affected the specificity of the programs used. On the one hand, it penalized the specificity of the most sensitive program (Quasar), and on the other it favored the specificity of the less sensitive program (GPA II). The experts detected progression in cases with higher initial MD than the programs. Although the !
2015 Informa Healthcare USA, Inc.
5
difference was not significant, this suggests that the experts may have confused visual field severity with progression. If this were the case, it would compromise the specificity of the most sensitive program. When suspected cases were considered as stable or progressing in the present study, GPA II showed low sensitivity (26.6% and 45.8%, respectively) and high specificity (95.1% and 90.6% respectively). Several authors have studied the sensitivity and specificity of GPA II event analysis, with contradictory results. Ang et al.,13 using the opinion of four experts as their reference standard, found rates of sensitivity and specificity (41% and 90%, respectively) that were very similar to our values. However other authors1,14 have obtained higher rates of sensitivity for GPA II when the reference standard comprised experts using objective systems to detect progression such as those of Hodapp et al.15 or Boden et al.16 based on pattern deviation plot. Pattern deviation analysis may underestimate the number of cases with progression since it does not detect diffuse changes of glaucomatous progression17–19 or changes in very advanced cases where thresholds are very low or out of range;14 it may also detect progression in cases of perimetric fluctuation. The inability to detect atypical or initial diffuse glaucomatous progression5 and changes in advanced glaucoma, coupled with the use of very strict criteria to detect progression explains why GPA II in our work presented low sensitivity when compared to a reference standard using subjective criteria of progression. GPA II event analysis showed high values for specificity; 95.1% when suspected cases were considered stable, and 90.6% when considered as progressing. These values are explained by the low sensitivity of the program which results in stable cases being considered as progressing (Figure 2). The Quasar program obtained higher sensitivity (86.6% and 79.1%) than GPA II event analysis and better kappa indices of agreement, according to our reference standard. Several studies have analyzed the evolution of MD in patients with glaucoma. The general consensus is that this is a sensitive procedure, but produces results too late.3,20 Being an averaged value, MD presents less fluctuation than individual threshold values and provides information on visual field changes or the learning effect. The drawback is that it fails to inform about the type of progression (focal or diffuse) in patients at an age when glaucoma and cataract often coexist. To compensate for this limitation, PSD analysis has been proposed since, in theory, this method provides information about the focal or diffuse nature of progression. For some authors PSD analysis yields favorable results,2 while others have not found it useful to detect progression.3 In our work, PSD analysis was less effective than MD analysis, only detecting progression in eight cases. However, the combination of the two parameters
V. T. Dı´az-Alema´n et al.
Curr Eye Res Downloaded from informahealthcare.com by Yale Dermatologic Surgery on 07/10/15 For personal use only.
6
FIGURE 2 In this image you may see an example with the first and last visual fields of a series of seven cases in 3.3 years. (A) Reference standard: the four examiners detected progression. (B) Quasar detects significant change. (C) GPA does not detect significant change.
resulted in higher overall sensitivity than for each one separately and for GPA II event analysis. The combination of criteria to detect progression has already been used by various investigators.
The TNT program employs three criteria to indicate progression based on linear regression analysis of MD, number of visual field points showing significant change, and significant change in eight sectors of the Current Eye Research
Curr Eye Res Downloaded from informahealthcare.com by Yale Dermatologic Surgery on 07/10/15 For personal use only.
Regression Analysis and Event Analysis Bebie curve. Progression is considered to exist when two or three criteria are positive, or if the one positive criterion is repeated on two successive examinations. The TNT program has demonstrated that the combination of these three criteria is superior to regression analysis of isolated variables such as MD, neuroretinal rim area, Glaucoma Probability Score and mean nerve fiber layer thickness (NFLT) for the early detection of progression in cases of incipient glaucoma.5 TNT has also shown greater efficacy than GPA for detecting glaucomatous progression.21 It could be interesting to compare Quasar and TNT in the future. Finally, Glaucoma Staging System (GSS) is a method of classifying visual field impairment, proposed by Brusini22 in 1995. It uses MD and PSD/LV values to classify severity (Stage 0, 1, 2, 3, 4, 5) and type of damage (generalized, mixed and localized defects). This method has proved useful in routine clinical practice and in the field of research but, as they authors have pointed out, it should not be used to determine significant progression. GSS provides useful information about MD trends over time, but the change from one stage to another cannot be automatically considered as significant progression.23 The present study has certain limitations. The first, as in many other similar studies, is the lack of a reference standard that defines progression. The second is the absence of morphological evidence to assess progression. We believe such evidence is important, although visual field tests are currently the most widely used method to determine glaucomatous progression.5 In conclusion, the present work shows that the combined use of the two parameters (MD and PSD) in the Quasar program improves the individual sensitivity of each parameter alone and that of GPA II event analysis. The specificity of the Quasar program was adversely affected by the choice of a reference standard with low sensitivity, which favored the specificity of the GPA II program. Finally, Quasar showed better agreement10 with the experts than GPA II. Further studies are needed to confirm these data.
DECLARATION OF INTEREST The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.
REFERENCES 1. Arnalich-Montiel F, Casas-Llera P, Mun˜oz-Negrete FJ, Rebolleda G. Performance of glaucoma progression analysis software in a glaucoma population. Graefes Arch Clin Exp Ophthalmol 2009;247:391–397. !
2015 Informa Healthcare USA, Inc.
7
2. Chauhan BC, Drance SM, Douglas GR. The use of visual field indices in detecting changes in the visual field in glaucoma. Invest Ophthalmol Vis Sci 1990;31: 512–520. 3. Gonza´lez De La Rosa M, Gonza´lez-Hernandez M, Dı´azAleman T. Linear regression analysis of the cumulative defect curve by sectors and other criteria of glaucomatous visual field progression. Eur J Ophthalmol 2009;19: 416–424. 4. Viswanathan AC, Fitzke FW, Hitchings RA. Early detection of visual field progression in glaucoma: a comparison of PROGRESSOR and STATPAC 2. Br J Ophthalmol 1997;81: 1037–1042. 5. Gonza´lez de la Rosa M, Gonza´lez-Hernandez M, Sa´nchez-Ma´ndez M, Medina-Mesa E, Rodriguez de la Vega R. Detection of morphological and functional progression in initial glaucoma. Br J Ophthalmol 2010; 94:414–418. 6. Gonza´lez de la Rosa M, Dı´az-Alema´n T, Gonza´lezHerna´ndez M, Jerez-Fidalgo M. Specificity of the program threshold noiseless trend for perimetric progression analysis. Curr Eye Res 2010;35:302–307. 7. Zeyen T, Roche M, Brigatti L, Caprioli J. Formulas for conversion between Octopus and Humphrey threshold values and indices. Graefes Arch Clin Exp Ophthalmol 1995;233:627–634. 8. Bengtsson B, Lindgren A, Heijl A, Lindgren G, Asman P, Patella M. Perimetric probability maps to separate change caused by glaucoma from that caused by cataract. Acta Ophthalmol Scand 1997;75:184–188. 9. Leske MC, Heijl A, Hussein M, Bengtsson B, Hyman L, Komaroff E. Factors for glaucoma progression and the effect of treatment. The Early Manifest Glaucoma Trial. Arch Ophthalmol 2003;121:48–56. 10. Lo´pez de Ullibarri, Galparsoro I, Pita Ferna´ndez S. Medidas de concordancia: el ´ındice de Kappa. Cad Aten Primaria 1999;6:169–171. 11. Tanna AP, Bandi JR, Budenz DL, Feuer WJ, Feldman RM, Herndon LW, et al. Interobserver agreement and intraobserver reproducibility of the subjective determination of glaucomatous visual field progression. Ophthalmology 2011;118:60–65. 12. Viswanathan AC, Crabb DP, McNaught AI, Westcott MC, Kamal D, Garway-Heath DF, et al. Interobserver agreement on visual field progression in glaucoma: a comparison of methods. Br J Ophthalmol 2003;87: 726–730. 13. Ang GS, Mustafa MS, Scott N, Diaz-Aleman VT, AzuaraBlanco A. Perimetric progression in open angle glaucoma and the Visual Field Index (VFI). J Glaucoma 2011;20: 223–227. 14. Anto´n A, Pazos M, Martı´n B, Navero JM, Ayala ME, Castany M, et al. Glaucoma progression detection: agreement, sensitivity, and specificity of expert visual field evaluation, event analysis, and trend analysis. Eur J Ophthalmol 2013;23:187–195. 15. Hodapp E, Parrish RK, Anderson DR. Clinical decisions in glaucoma. St Louis: CV Mosby; 1993. 16. Boden C, Blumenthal EZ, Pascual J, McEwan G, Weinreb RN, Medeiros F, et al. Patterns of glaucomatous visual field progression identified by three progression criteria. Am J Ophthalmol 2004;138:1029–1036. 17. Bengtsson B, Lindgren A, Heijl A, Lindgren G, Asman P, Patella M. Perimetric probability maps to separate change caused by glaucoma from that caused by cataract. Acta Ophthalmol Scand 1997;75:184–188. 18. Katz J. A comparison of the pattern and total deviation based Glaucoma Change Probability programs. Invest Ophthalmol Vis Sci 2000;41:1012–1016.
8
V. T. Dı´az-Alema´n et al. visual field deterioration by Glaucoma Progression Analysis and Threshold Noiseless Trend Programs. Br J Ophthalmol 2009;93:322–328. 22. Brusini P. Clinical use of a new method for visual field damage classification in glaucoma. Eur J Ophthalmol 1996; 6:402–407. 23. Brusini P, Filacorda S. Enhanced Glaucoma Staging System (GSS 2) for classifying functional damage in glaucoma. J Glaucoma 2006;15:40–46.
Curr Eye Res Downloaded from informahealthcare.com by Yale Dermatologic Surgery on 07/10/15 For personal use only.
19. Artes PH, Nicolela MT, LeBlanc RP, Chauhan BC. Visual field progression in glaucoma: total versus pattern deviation analyses. Invest Ophthalmol Vis Sci 2005;46: 4600–4606. 20. Nouri-Mahdavi K, Brigatti L, Weitzman M, Caprioli J. Comparison of methods to detect visual field progression in glaucoma. Ophthalmology 1997;104:1228–1236. 21. Diaz-Aleman VT, Anton A, Gonzalez de la Rosa M, Johnson ZK, McLeod S, Azuara-Blanco A. Detection of
Current Eye Research
