A Visual Field Index for Calculation of Glaucoma Rate of Progression

Transcription

1 A Visual Field Index for Calculation of Glaucoma Rate of Progression BOEL BENGTSSON AND ANDERS HEIJL PURPOSE: To present a new perimetric index forglaucoma IS A PROGRESSIVE DISEASE THAT calculating the rate of glaucomatous progression and to causes morphologic changes of the optic disk and compare its performance with the traditional mean deviation index (MDI). associated effects on the visual field. Detection and mon- the retinal nerve fiber layer, accompanied by DESIGN: Experimental study describing a device and itoring of glaucoma patients are based on identification of retrospective cohort study. structural and functional changes, but only functional METHODS: We developed a new visual field index, themeasurements can quantify the patient s visual status. glaucoma progression index (GPI), intended to be lesstherefore, visual field testing is a very important exami- factor in the management of glaucoma patients. affected by cataract than the MDI by calculating age-natiocorrected defect depth at test points identified as signifi-severacantly depressed in pattern deviation probability maps. progression have been developed 1 4 and used in clinical different methods for identification of visual field The valid operating range for pattern deviation analysistrials. 3 5 Most of these methods involve event-based anal- that are designed primarily to detect whether progres- was estimated. When exceeding this range, the totalyses deviation probability maps were used for identification ofsion has occurred, but because glaucoma is a progressive significantly depressed points. The GPI is expressed indisease, most glaucoma patients will show some progression if followed up long enough. From a long-term percentage, where 100% represents a normal visual field 6 and 0% represents a perimetrically blind field, and isperspective, it is important to determine a patient s veloc- of disease progression, the so-called rate of progression. plotted vs patient age. Rate of progression, presented asity yearly change in the GPI, is calculated by linear regression analysis. We conducted a pilot evaluation in threeby so-called trend analyses, which typically are performed by Information about the rate of progression usually is provided groups of patients: 1) eyes with developing cataract, 2) 7 regressing a global visual field index over time. The Hum- perimeter (Carl Zeiss Meditec, Dublin, California, eyes without cataract, and 3) eyes in which cataractphrey surgery was performed in the middle of the series. USA) Statpac mean deviation index (MDI) and the Octopus RESULTS: The cut-off for pattern deviation was, at perimeter (Haag Streit AG, Koeniz, Switzerland) mean defect index present the rate of progression of field loss in mean deviation, worse than 20 decibels (db) in fields in which the eighty-fifth percentile of the total deviation 2,8 decibels (db) per year. We know from retrospective studies value was significantly depressed. In the first group (n of patient records (Heijl A, Bengtsson B, unpublished data, 45), the measured rate of progression was greater with2006) and also from prospectively collected longitudinal the MDI than with the GPI P < (.0001). The mean loss data 9,10 that rates of progression vary widely among glaucoma per year was 3.6%/year for the MDI and 2.1%/year for patients. One investigation reporting risk factors for glaucoma the GPI. In the second group (n 42), the rate of progression showed that the rate of progression was the progression did not differ between the MDI and the GPI 11 strongest predictive factor for further progression. Thus, (P.52); the means were 2.7%/year and 2.6%/year, progression velocity may be an important and useful identifier respectively. In the third group (n 44), the confidence of patients at risk for developing sight-threatening glaucoma- visual field loss. limits for the rate of progression were significantlytous smaller with the GPI than with the MDI P (.04). Today, the MDI is the standard index for estimating the CONCLUSIONS: Glaucoma progression rates calculated rate of glaucomatous progression. The MDI is affected not using the GPI seem to be considerably less affected by only by increasing glaucoma, but also by cataract. Thus, cataract and cataract surgery than rates based on the increasing cataract can falsely suggest a high progression traditional MDI. (Am J Ophthalmol 2008;145: rate. Similarly, after cataract surgery, MDI values will by Elsevier Inc. All rights reserved.) improve, jeopardizing any evaluation of glaucoma progression using preoperative and postoperative MDI values We also realized that MDI is only very weakly center See accompanying Editorial on page 191. Accepted for publication Sep 27, From the Department of Clinical Sciences, Ophthalmology, Malmö University Hospital, Lund University, Malmö, Sweden. Inquiries to Boel Bengtsson, Department of Clinical Sciences, Ophthalmology, Malmö University Hospital, SE Malmö, Sweden; boel.bengtsson@med.lu.se weighted, 17 and therefore is not as well correlated to patient visual function as may be desired. For these reasons, we sought to develop a new agecorrected visual field index for assessment of progression /08/$ BY ELSEVIER INC. ALL RIGHTS RESERVED. 343 doi: /j.ajo

2 called our new index the glaucoma progression index (GPI). The GPI is expressed as a percent of a normal visual field, making it possible to calculate a rate of functional loss. We chose to present results relative to patient age to emphasize the role of life expectancy considerations in determining a therapeutic course. This article describes our new global visual field index and reports our results from evaluating its performance in series of glaucomatous visual fields. METHODS PATIENTS AND VISUAL FIELD DATA: All visual fields data analyzed in the current study were retrieved from the records of glaucoma patients followed up at the Department of Ophthalmology at Malmö University Hospital, Malmö, Sweden. We listed all glaucoma patients who had been followed up with Humphrey Swedish interactive threshold algorithm (SITA) standard visual fields for at least five years for whom at least five tests were available. We also reviewed all corresponding patient records and noted the dates of any cataract surgeries. All visual field tests having frequencies of false-positive answers of more than 15% were excluded. Tests with high frequencies of false-negative answers were not excluded, because a high number of false-negative answers is strongly associated with visual field status, rather than with patient reliability. 18 Neither were tests having high frequencies of fixation loss excluded, as measured by the blind spot method, as long as the blind spot was visible in the gray scale and the threshold value was less than 0 db at the location for the blind spot. FIGURE 1. A visual field printout of an eye with suspect glaucoma and cataract. (Top left) Measured threshold values are recalculated into deviations from age-normal threshold in (Middle left) the total deviation map, with negative values representing reduced sensitivity relative to age-normal values. In this example, all measured sensitivity values are lower than age-normal values. (Middle right) The pattern deviation map displays deviations from age-normal values, adjusted for general depression such as that produced by cataract. (Bottom left and right) The significances of the deviations are displayed in probability maps. In this test, cataract explains the difference in number of significantly depressed points in the two probability maps. GHT glaucoma hemifield test; MD mean deviation; PSD pattern standard deviation. that was largely independent of cataract and reflected more accurately the relative importance of the central and more peripheral visual fields to patient visual function. We GENERAL FEATURES OF THE NEW THE GLAUCOMA PROGRESSION INDEX: The new GPI was designed for the Humphrey 30-2 and 24-2 test point patterns, which are the most commonly used patterns in glaucoma management. To facilitate interpretation, visual field findings were adjusted for age and were converted into percent; in this way, the GPI of a perimetrically normal field was set to 100%, and the GPI of a perimetrically blind field was set to 0%. IDENTIFICATION OF DEPRESSED POINTS: Standard printouts of Humphrey visual field test results display two numerical test point maps in addition to the map presenting threshold sensitivity values. One of these maps, the total deviation map, displays decibel deviations from agecorrected normal threshold sensitivity (Figure 1). 17 The other map, the pattern deviation map, presents decibel deviations from the age-corrected normal threshold values, after adjustment for any overall elevation or depression of the field, such as that commonly caused by cataract. Each deviation map is accompanied by an associated probability map that displays the significance of each numerical 344 AMERICAN JOURNAL OF OPHTHALMOLOGY FEBRUARY 2008

3 FIGURE 2. A visual field with glaucomatous defects covering approximately half the field, but with the test points closest to fovea within normal limits. The unweighted glaucoma progression index (GPI) estimates remaining function to be 50.3%, while the weighted GPI estimate function at 61.0%. NEG negative; POS positive; SITA Swedish interactive threshold algorithm. VOL. 145, NO. 2 GLAUCOMA RATE OF PROGRESSION 345

4 deviation relative to the range of values found in normal subjects. To avoid effects of cataract on the new GPI, we used the pattern deviation probability maps 19 to identify test points with normal sensitivity and those demonstrating relative loss. Test points having threshold sensitivities within normal limits on the pattern deviation probability maps were considered normal and were scored to 100% sensitivity. Test points having absolute defects, defined as measured threshold sensitivities of less than 0 db, were scored to 0% sensitivity. Points with significantly depressed sensitivity, but not blind (relative loss), were identified as test points with sensitivities depressed below the P.05 significance limits in the pattern deviation map. The sensitivity at these points were scored in percent as: 100 [( total deviation age-corrected normal threshold) 100], where total deviation is the absolute value of the numerical total deviation value and age-corrected normal threshold is: intercept age coefficient patient age. WEIGHTING PROCEDURE: A weighting procedure was applied with the goal of recognizing the higher importance of vision at central and paracentral test point locations vs points more in the periphery. For this purpose, the test point pattern was divided into five concentric rings of increasing eccentricity. We adopted a published estimate of the spatial magnification present in the occipital cortex as the basis for this weighting. 20 Cortical magnification is assumed to reflect ganglion cell density and to describe the number of neurons in an area of the visual cortex responsible for processing a stimulus of a given size as a function of visual field location. The four test points located in the inner ring, not including measurement of foveal threshold, was allotted a weight of The weights decreased with increasing eccentricity from 1.28, 0.79, and 0.57 to 0.45 at the most outer ring. The GPI is the mean of all weighted scores in percent. The effects of this weighting procedure on the GPI are most pronounced in the parafoveal region (Figure 2). DETERMINATION OF A VALID OPERATING RANGE FOR PATTERN DEVIATION ANALYSIS: The GPI relies on the well-known ability of pattern deviation analysis to account for cataract effects; but like all methods, this ability has limits. We wished to apply the correction only where pattern deviation is known to work well and to seek other approaches when analyzing visual field test results falling outside this range. Pattern deviation values are adjusted for the general height of the field, and those adjustments are based on measurements at the eighty-fifth percentile most sensitive point relative to age-normal values. 19 This approach effectively corrects for most cataract effects, but the method FIGURE 3. Scatterplot showing the relationship between the number of significantly depressed points in visual field pattern deviation probability plots and the global mean deviation index (MDI) in 307 visual field tests of 30 eyes. The number of significantly depressed points increased with increasing visual field damage from the normal end of the y-axis (0 db) until approximately 20 db, and then this number decreased with further damage. The coefficient of determination, R 2, was breaks down in very advanced field loss, where even the eighty-fifth percentile most sensitive test point has been affected significantly by disease. In these situations, the pattern deviation analysis may suggest artifactual visual field improvements despite the reality of further field deterioration. Therefore, we sought to identify an MDI cut-off value, beyond which pattern deviation would no longer be applied to our analysis. We defined a cut-off by examining consecutive field series from eyes progressing from mild or moderate to very severe loss, with very severe loss defined as any field having an MDI worse than 24 db. The number of significantly depressed points in the pattern deviation probability maps was counted and plotted vs the MDI, and a second-order polynomial regression analysis was applied to identify the peak of the function. We also calculated individual differences in the number of significantly depressed test points in total deviation vs pattern deviation probability maps and adapted a polynomial curve of these differences as a function of the MDI to find the MDI level with the smallest difference. PILOT EVALUATION: Patients. From a database that included all glaucoma patients participating in our study, we identified those having at least five SITA standard field tests collected over at least five years who also had undergone cataract surgery at some point during the course of follow-up. Approximately 200 field series were printed. Only one eye per patient was included; if both eyes were eligible, one eye was selected randomly. Most patients had 30-2 test results only, but some had undergone one or two 24-2 tests. In series including 24-2 tests, the GPI was calculated using the 24-2 pattern for all tests, because AMERICAN JOURNAL OF OPHTHALMOLOGY FEBRUARY 2008

5 TABLE. Baseline Patient Characteristics, Follow-up Duration, Intraocular Pressure, and Number of Interventions for Three Groups of Glaucoma Patients Eyes with Increasing Cataract (n 45) Eyes without Cataract (n 42) Eyes Undergoing Cataract Surgery (n 44) Mean age* (range), yrs 73 (62 to 83) 79 (63 to 90) 76 (48 to 86) Visual field status*: MD (db)/psd (db) 3.39/ / /4.43 Mean duration of follow-up (range), yrs 5.1 (2.4 to 8.4) 4.4 (2.0 to 9.0) 4.0 (2.1 to 7.5) Mean no. of fields per eye (range) 6.6 (5 to 11) 6.4 (5 to 14) 5 (5 to 5) Mean IOP (mm Hg) No. of interventions (range) 0to7 1to11 4to7 IOP intraocular pressure; MD mean deviation; PSD pattern standard deviation; db decibels. *At the first included visual field test. All series were chosen to have five tests contiguous to surgery. Change in intraocular pressure-lowering medications, argon laser trabeculoplasty, and surgery. is a subset of 30-2 test points. This procedure is identical to that currently applied in the Humphrey perimeter when calculating rate of progression using the MDI. Patients were divided into three groups. The first group included patients with progressive cataract. Only eyes having field series with at least five visual field test results before cataract surgery were eligible; no postsurgical test results were included in the analysis. The second group was comprised of patients who already had undergone cataract surgery with implantation of an intraocular lens. To be eligible, each series had to include at least five visual field test results after cataract surgery. In the third group, we wanted to study effects of cataract surgery on our index, and therefore included patients having visual field tests both before and after surgery. We used the five test results closest to surgery, randomly selecting either two or three test results before surgery and the remaining test results after. Analyses. Rate of progression was assessed by linear regression of global visual field sensitivity as expressed by the MDI in percent and by the GPI over time. The MDI in percent was calculated as: (1 (MDI meas MDI min )) 100, where MDI meas was the measured MDI, and MDI min was the minimum MDI for a subject of a certain age. The 95% confidence interval for the regression slope also was calculated. We compared rate of progression and confidence intervals between the new GPI and the traditional MDI, recalculated into percent. Because of the lack of gold standard, we considered the MDI as the reference standard for the rate of progression. However, knowing that the MDI-based rate of progression is affected by increasing cataract, we anticipated a nonspecified faster rate of progression with the MDI compared with the GPI in the group including patients with increasing cataract and a similar rate of progression in the group including pseudophakic eyes only. Differences in rates of progression and confidence intervals were calculated and tested by onesample t tests, because distributions of differences seemed to be distributed normally. RESULTS PATTERN DEVIATION OPERATING RANGE: In an attempt to identify a suitable cut-off for the pattern deviation calculation, we included 30 eyes of 29 patients, all with long series of fields. A total of 307 field tests were included; the mean number of tests per eye was 10.2, ranging from six to 16. The time span for the series ranged from 5.1 to 10.2 years, with a median of 8.8 years. At the time of the first SITA field test, patient age averaged 73.2 years, ranging from 54 to 87 years. The mean MDI of the first tests in each series was 16.1 db, and the mean MDI of the last tests was 26.6 db. A second-order polynomial regression model including all 30 eyes indicated that the number of significantly depressed points in pattern deviation probability maps peaked at an MDI value near 20 db (Figure 3). Thus, the number of significantly depressed test points in pattern deviation probability maps could be underestimated in fields having MDI values worse than 20 db. In 98% of all fields having MDI values worse than 20, the eighty-fifth percentile of the numerical total deviation value was significantly depressed. Thus, when calculating the GPI, the MDI cut-off level for using pattern deviation probabilities as indicators of points with normal sensitivity or relative loss was set at 20 db, as long as significant depression of less than the age-corrected normal sensitivity (P.05) of the eightyfifth percentile of the total deviation values also was present. For fields having MDIs worse than 20 db, total deviation probability maps were used to indicate test point locations having normal sensitivity or relative loss. The minimum difference between numbers of significantly depressed points in total and pattern deviation probability maps also was near an MDI of 20 db. VOL. 145, NO. 2 GLAUCOMA RATE OF PROGRESSION 347

6 FIGURE 4. Bar graphs showing the distributions of rates of progression from visual field series in eyes with developing cataract. The calculated rate of progression was significantly faster with the (Right) (MDI; 3.6%/year) than with the (Left) (GPI; 2.1%/year). FIGURE 5. Visual field rates of progression with (Top) the GPI and (Bottom) the MDI from an eye with progressing glaucoma and cataract. The GPI indicates 1.14% (95% confidence interval, 0.80%) loss per year and the MDI indicates 3.04% (95% confidence interval, 2.69%) loss per year. 348 AMERICAN JOURNAL OF OPHTHALMOLOGY FEBRUARY 2008

7 FIGURE 6. Bar graphs showing the distributions of rates of progression from visual field series in pseudophakic eyes. There was no significant difference between (Left) the GPI and (Right) the MDI in velocity of progression. The mean loss was 2.6%/year with the GPI and 2.7%/year with the MDI. FIGURE 7. Visual field rate of progression with (Top) the GPI and (Bottom) the MDI from a pseudophakic eye with progressing glaucoma. The GPI indicates 5.94% (95% confidence interval, 1.67%) loss per year and the MDI indicates 5.46% (95% confidence interval, 2.66%) loss per year. VOL. 145, NO. 2 GLAUCOMA RATE OF PROGRESSION 349

8 FIGURE 8. Bar graphs showing the distributions of rates of progression of visual field series of eyes undergoing cataract surgery. The rates of progression were significantly (P.015) smaller with (Right) the MDI, in which the median was 0.8% per year, than with (Left) the GPI, in which the median was 2.1% per year. FIGURE 9. Visual field rates of progression with (Top) the GPI and (Bottom) the MDI of an eye with glaucoma operated on for cataract after the first three field tests. The scatter across the regression line, estimated by mean square error, is 72% 2 using the GPI and % 2 with the MDI. 350 AMERICAN JOURNAL OF OPHTHALMOLOGY FEBRUARY 2008

9 PILOT EVALUATION: One hundred and thirty-one patients were included in our pilot evaluation. Patient characteristics are presented in the Table. Forty-five eyes having five or more visual field tests before cataract surgery were identified (group 1). As anticipated, eyes with developing cataract showed a significantly faster rate of MDI progression compared with the GPI (P.0001; Figure 4). The mean loss was 3.6%/year using the MDI and 2.1%/year using the GPI. A field series of an eye included in the group with increasing cataract is shown in Figure 5. The 95% confidence limits for the rate of progression were narrower with the GPI than with the MDI (P.02). The median confidence limit was 1.5% for the GPI and 2.2% for the MDI. Only pseudophakic eyes were included in the cataractfree second group. We identified 42 eyes having five fields or more after surgery. In this group, there was no difference in rate of progression (P.52) or confidence limits (P.54) between the GPI and MDI (Figure 6). The mean loss with the GPI was 2.6%/year, vs 2.7%/year with the MDI, and the median confidence limits were 2.3% and 2.6% on average. A field series of an eye included in the group including pseudophakic eye only is shown in Figure 7. In the group undergoing cataract surgery during the study (group 3), the rate of progression was smaller with the MDI than with the GPI (Figure 8). In the first tests in the series, the MDI indicated more field loss than the GPI, with a difference of 8.8%, whereas the difference decreased to 4% at the last test in the series. The MDI showed significantly more scatter across the regression line than the GPI (P.04). The median confidence limits were 4.4% with the MDI and 2.9% with the GPI. Thus, confidence in progression rate estimates was significantly better with the GPI compared with the MDI. A field series of an eye included in the group undergoing cataract surgery is shown in Figure 9. DISCUSSION THE GPI IS A NEW SUMMARY VISUAL FIELD INDEX FOR glaucoma management. In this study, GPI gave results very similar to the present standard, the MDI, in eyes with no cataract (pseudophakic eyes), whereas in eyes with increasing cataract, the rate of progression was smaller and the confidence limits were narrower with the GPI. In eyes undergoing cataract surgery in the middle of the series, the rate of progression based on the MDI was smaller than that based on the GPI, which was expected because of cataract effect on the first fields, which also explains the wider confidence intervals using the MDI. Thus, cataract and cataract surgery was less disturbing for the rate of progression calculation using the GPI compared with calculation based in the MDI. Further, and as opposed to the MDI, the GPI is expressed in percent of normal age-corrected visual function, and it is intended for use in calculating patient rates of progression and also in staging glaucomatous functional damage on a scale from normal function to perimetric blindness. Plotting GPI over time vs patient age and performing a linear regression analysis of GPI over time can achieve both of these goals. We used the pattern deviation probability maps of the standard Statpac program of the Humphrey perimeter 17 to make the GPI as resistant as possible to the effects of media opacities, while clearly depicting localized loss. The pattern deviation concept has been widely accepted and applied to perimetric analysis and is used as way of reducing the effects of developing cataract. We found that the number of significantly depressed points in pattern deviation maps peaked at MDI values of approximately 20 db. We therefore decided that, with rare exceptions, we should limit the use of pattern deviation to those fields having MDI values better than 20 db and to use total deviation maps to define abnormality in fields having MDIs worse than 20 db. We believe that we have made best use of pattern deviation in our index by applying it only as a tool for identifying abnormal test points. The extent of the field loss at those locations, then, is based on total deviation values. The validity of this approach seems to be supported by the fact that GPI-based and MDI-based rates of progression did not differ in the group of cataract-free eyes studied here. However, shifting from pattern deviation probabilities to total deviation probabilities for identification of depressed points is likely to result in a slight stepwise worsening of GPI near MDI values of 20 db. Two publications have suggested other techniques for adjusting for general depression by alternative ways of calculating the general height of the visual field. 24,25 The technique described by Åsman and associates would extend just marginally the valid operating range for pattern deviation (Åsman P, Heijl A, unpublished data, 2006), 24 whereas the techniques of Vingrys and Zele have been validated in simulations only. 25 This topic remains to be addressed in a separate prospective longitudinal study, where diffuse loss caused by cataract can be separated from that caused by glaucoma. We performed a pilot evaluation of the GPI, comparing it with the MDI in visual field series from eyes with developing cataract, from eyes free of cataract, and eyes in which cataract surgery was performed in the middle of the field series. We believe that our evaluation suggests that the GPI has significant advantages compared with the MDI because of the GPI s relative immunity to cataract effects. Regression analysis of GPI represents an attempt to create a tool for estimating the global functional rate of progression in glaucoma that is relatively resistant to the influence of cataract and cataract surgery. Cataracts and cataract surgery are common in elderly age groups where glaucoma also is common, and separation of field defects caused by glaucoma from those caused by cataract should VOL. 145, NO. 2 GLAUCOMA RATE OF PROGRESSION 351

10 clarify and simplify clinical analysis of test results. We also believe that expressing the GPI and rate of progression in percent of age-corrected normal function represents a significant advantage, compared with simply using decibels, particularly because the perimetrically blind 0% functional level corresponds to different decibel deviation levels, depending both on patient age and the test program used. We believe that it is essential that new perimetric tools be as simple and intuitively comprehensible as possible, so as to foster broad adoption worldwide among practitioners not necessarily specializing in glaucoma care. Central weighting of the index, based on cortical magnification, also should be advantageous, making the GPI more closely reflect ganglion cell loss and also visual function. Further study is required to document clearly the extent of these advantages. The GPI also may have some disadvantages. One is that GPI is relatively complicated to calculate; age-corrected threshold values, probability values, eccentricity weighting factors, plus estimates of the amount of diffuse field loss all are needed. However, all these calculations and the regression analyses are tasks that are simple to automate. A second disadvantage is associated with the use of pattern deviation to identify abnormal locations, the same concept that gives the advantage of reducing the impact of cataract and cataract surgery on GPI. Glaucomatous field loss can be divided into two components, localized and diffuse loss. There is rather broad agreement that almost no glaucomatous fields show diffuse loss in the absence of localized loss, 26,27 but with increasing field loss usually comes an increasing component of glaucomatous diffuse loss. 28 Including only localized field loss in progression analysis necessarily results in some level of underestimation of progression. 24,28,29 One may question whether disease progression in glaucoma is linear, and if predictions of a patient s future disease course can or should be based on linear regression analysis of past observations. Several randomized clinical trials indicate that both increasing age and increasing damage are associated with increased risk of progression These findings may suggest that typical disease course actually may be accelerating over a very long perspective. Long-term studies agree, however, that a linear disease model seems to fit observed data at least as well as any other model. 35,36 Linear regression also has been shown to best predict future risk of progression, 10,37 and in our clinical data set, the GPI certainly was close to linear in most examined cases, and the confidence intervals around the linear slopes usually were narrow. We hope that an analysis showing the observed rate of progression plus estimated visual function will be comprehensible intuitively and will be of value for clinical management (Figure 8). Rate of progression varies considerably from patient to patient and from eye to eye and cannot be predicted presently with any confidence based on population risk factors. With data on visual function and observed rate of progression, physicians would have an additional tool to help determine whether a patient s disease is under adequate control or is progressing quickly enough to risk the patient s quality of life in his lifetime. In conclusion, we have developed a new visual field index that estimates visual function relative to age-corrected normal values and that is relatively resistant to cataract and cataract surgery. We hope that plotting this index over time and calculating rate of progression will provide doctors with an easily understandable and useful tool for use in everyday glaucoma management. THIS STUDY WAS SUPPORTED BY AN UNRESTRICTED GRANT FROM ALCON RESEARCH LTD, FORT WORTH, TEXAS; THE Swedish Research Council, Stockholm, Sweden; K X A. The Järnhardt Foundation, Malmö, Sweden; and The Foundation for Visually Impaired in former Malmöhus län, Sweden. Drs Bengtsson and Heijl has been consultants to Carl Zeiss Meditec and Dr Heijl consultant to Allergan, Alcon, and Pfizer and have received lecture fees from Pfizer. Involved in conception and design of study, writing, critical revision, final approval (B.B., A.H.); Analysis and interpretation, data collection, statistical expertise, and literature search (B.B.); and Provision of materials, obtaining funding, and administrative support (A.H.). The Committee for Research Ethics at Lund University approved the study protocol and the study was conducted according to the tenets of the Declaration of Helsinki. REFERENCES 1. Fitzke FW, Hitchings RA, Poinoosawmy D, McNaught AI, Crabb DP. Analysis of visual field progression in glaucoma. Br J Ophthalmol 1996;80: Heijl A, Lindgren G, Lindgren A, et al. Extended empirical statistical package for evaluation of single and multiple fields: Statpac 2. In Mills RP, Heijl A, editors. Perimetry update 1990/1. New York, New York: Kugler and Ghedini, 1991: The Advanced Glaucoma Intervention Study Investigators. Advanced Glaucoma Intervention Study. 2. Visual field test scoring and reliability. Ophthalmology 1994;101: Musch DC, Lichter PR, Guire KE,, Standardi CL. The Collaborative Initial Glaucoma Treatment Study: study design, methods, and baseline characteristics of enrolled patients. Ophthalmology 1999;106: Leske MC, Heijl A, Hyman L, Bengtsson B. Early Manifest Glaucoma Trial: design and baseline data. Ophthalmology 1999;106: Leske C, Heijl A, Hyman L, Bengtsson B, Dong L, Yang Z, EMGT Group. Predictors of long-term progression in the early manifest glaucoma trial. Ophthalmology 2007;114: Holmin C, Krakau CET. Regression analysis on the central visual field in chronic glaucoma cases. A follow-up study using automatic perimetry. Acta Ophthalmol 1982;60: AMERICAN JOURNAL OF OPHTHALMOLOGY FEBRUARY 2008

11 8. Weijland A, Fankhauser F, Bebie H, Flammer J. Automated perimetry, visual field digest. 5th ed. Switzerland: Haag-Streit AG, 2004: Collaborative Normal-Tension Glaucoma Study Group. Natural history of normal-tension glaucoma. Ophthalmology 2001;108: Heijl A, Leske MC, Bengtsson B, Hussein M. Early Manifest Glaucoma Trial Group. Acta Ophthalmol Scand 2003;81: Nouri-Mahdavi K, Hoffman D, Gaasterland D, Caprioli J. Prediction of visual field progression in glaucoma. Invest Ophthalmol Vis Sci 2004;45: Klein BE, Klein R, Jensen SC. Visual sensitivity and agerelated eye disease: the Beaver Dam Eye Study. Ophthalmic Epidemiol 1996;3: Smith SD, Katz J, Quigley HA. Effect of cataract extraction on the results of automated perimetry in glaucoma. Arch Ophthalmol 1997;115: Chen PP, Budenz DL. The effects of cataract extraction on the visual field of eyes with chronic open-angle glaucoma. Am J Ophthalmol 1998;125: Kim YY, Kim JS, Shin DH, Kim C, Jung HR. Effect of cataract extraction on blue-on-yellow visual field. Am J Ophthalmol 2001;132: Koucheki B, Nouri-Mahdavi K, Patel G, Gaasterland D, Caprioli J. Visual field changes after cataract extraction: the AGIS experience. Am J Ophthalmol 2004;138: Heijl A, Lindgren G, Olsson J. A package for the statistical analysis of visual fields. In: Greve EL, Heijl A, editors. Seventh international visual field symposium, Amsterdam Doc Ophthalmol Proc Ser. Dordrecht, The Netherlands: Martinus Nijhof/Dr W. Junk Publishers, 1987;49: Bengtsson B, Heijl A. False negative responses in glaucoma perimetry: indicators of patient performance or test reliability? Invest Ophthalmol Vis Sci 2000;41: Heijl A, Lindgren G, Olsson J, Åsman P. Visual field interpretation with empirical probability maps. Arch Ophthalmol 1989;107: Levi DM, Klein SA, Aitsebaomo AP. Vernier acuity, crowding, and cortical magnification. Vision Res 1985;25: Weijland A, Fankhauser F, Bebie H, Flammer J. Automated Perimetry. Visual field digest, 5th ed. Switzerland, Haag- Streit AG, 2004: Soliman MA, de Jong LA, Ismaeil AA, van den Berg TJ, de Smet MD. Standard achromatic perimetry, short wavelength automated perimetry, and frequency doubling technology for detection of glaucoma damage. Ophthalmology 2002;109: Johnson CA, Wall M, Fingeret M, Lalle P. In A primer for frequency doubling technology. Skaneateles, New York, New York: Welch Allyn, 1998: Åsman P, Wild JM, Heijl A. Appearance of the pattern deviation map as a function of change in area of localized field loss. Invest Ophthalmol Vis Sci 2004;45: Vingrys AJ, Zele AJ. Robust indices of clinical data: meaningless means. Invest Ophthalmol Vis Sci 2005;46: Åsman P, Heijl A. Diffuse visual field loss and glaucoma. Acta Ophthalmol 1994;72: Chauhan BC, LeBlanc RP, Shaw AM, Chan AB, McCormick TA. Repeatable diffuse visual field loss in open-angle glaucoma. Ophthalmology 1997;104; Weber JT. Topographie der funktionellen Schädigung beim chronishen glaukom. Heidelberg, Germany: Kaden Verlag, 1992: Artes PH, Nicolela MT, LeBlanc RP, Chauhan BC. Visual field progression in glaucoma: total versus pattern deviation analyses. Invest Ophthalmol Vis Sci 2005;46: Manassakorn A, Nouri-Mahdavi K, Kouecheki B, Law SK, Caprioli J. Pointwise linear regression analysis for detection of visual field progression with absolute versus corrected threshold sensitivities. Invest Ophthalmol Vis Sci 2006;47: Stewart WC, Kolker AE, Sharpe ED, et al. Factors associated with long-term progression or stability in primary open-angle glaucoma. Am J Ophthalmol 2000;130: Leske MC, Heijl A, Hussein M, Bengtsson B, Hyman L, Komaroff E. Factors for progression and the effect of treatment: the early manifest glaucoma trial. Arch Ophthalmol 2003;121: Nouri-Mahdavi K, Hoffman D, Coleman AL, et al. Advanced Glaucoma Intervention Study. Predictive factors for glaucomatous visual field progression in the Advanced Glaucoma Intervention Study. Ophthalmology 2004;111: Jonas JB, Martus P, Horn FK, Junemann A, Korth M, Budde WM. Predictive factors of the optic nerve head for development or progression of glaucomatous visual field loss. Invest Ophthalmol Vis Sci 2004;45: Mikelberg FS, Sculzer M, Drance SM, Lau W. The rate of progression of scotomas in glaucoma. Am J Ophthalmol 1986;101: Kwon YH, Kim CS, Zimmerman MB, Alward WL, Hayreh SS. Rate of visual field loss and long-term visual outcome in primary open-angle glaucoma. Am J Ophthalmol 2001;132: McNaught AI, Crabb DP, Fitzke FW, Hitchings RA. Modelling series of visual fields to detect progression in normaltension glaucoma. Graefes Arch Clin Exp Ophthalmol 1995; 233: VOL. 145, NO. 2 GLAUCOMA RATE OF PROGRESSION 353