December 2005
Volume 46, Issue 12
Free
Glaucoma  |   December 2005
Visual Field Progression in Glaucoma: Total Versus Pattern Deviation Analyses
Author Affiliations
  • Paul H. Artes
    From the Department of Ophthalmology and Visual Sciences, Dalhousie University, Halifax, Nova Scotia, Canada; and the
    Faculty of Life Sciences, The University of Manchester, Manchester, United Kingdom.
  • Marcelo T. Nicolela
    From the Department of Ophthalmology and Visual Sciences, Dalhousie University, Halifax, Nova Scotia, Canada; and the
  • Raymond P. LeBlanc
    From the Department of Ophthalmology and Visual Sciences, Dalhousie University, Halifax, Nova Scotia, Canada; and the
  • Balwantray C. Chauhan
    From the Department of Ophthalmology and Visual Sciences, Dalhousie University, Halifax, Nova Scotia, Canada; and the
Investigative Ophthalmology & Visual Science December 2005, Vol.46, 4600-4606. doi:10.1167/iovs.05-0827
  • Views
  • PDF
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Paul H. Artes, Marcelo T. Nicolela, Raymond P. LeBlanc, Balwantray C. Chauhan; Visual Field Progression in Glaucoma: Total Versus Pattern Deviation Analyses. Invest. Ophthalmol. Vis. Sci. 2005;46(12):4600-4606. doi: 10.1167/iovs.05-0827.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

purpose. To compare visual field progression with total and pattern deviation analyses in a prospective longitudinal study of patients with glaucoma and healthy control subjects.

methods. A group of 101 patients with glaucoma (168 eyes) with early to moderately advanced visual field loss at baseline (average mean deviation [MD], −3.9 dB) and no clinical evidence of media opacity were selected from a prospective longitudinal study on visual field progression in glaucoma. Patients were examined with static automated perimetry at 6-month intervals for a median follow-up of 9 years. At each test location, change was established with event and trend analyses of total and pattern deviation. The event analyses compared each follow-up test to a baseline obtained from averaging the first two tests, and visual field progression was defined as deterioration beyond the 5th percentile of test–retest variability at three test locations, observed on three consecutive tests. The trend analyses were based on point-wise linear regression, and visual field progression was defined as statistically significant deterioration (P < 5%) worse than −1 dB/year at three locations, confirmed by independently omitting the last and the penultimate observation. The incidence and the time-to-progression were compared between total and pattern deviation analyses. To estimate the specificity of the progression analyses, identical criteria were applied to visual fields obtained in 102 healthy control subjects, and the rate of visual field improvement was established in the patients with glaucoma and the healthy control subjects.

results. With both event and trend methods, pattern deviation analyses classified approximately 15% fewer eyes as having progressed than did the total deviation analyses. In eyes classified as progressing by both the total and pattern deviation methods, total deviation analyses tended to detect progression earlier than the pattern deviation analyses. A comparison of the changes observed in MD and the visual fields’ general height (estimated by the 85th percentile of the total deviation values) confirmed that change in the glaucomatous eyes almost always comprised a diffuse component. Pattern deviation analyses of progression may therefore underestimate the true amount of glaucomatous visual field progression.

conclusions. Pattern deviation analyses of visual field progression may underestimate visual field progression in glaucoma, particularly when there is no clinical evidence of increasing media opacity. Clinicians should have access to both total and pattern deviation analyses to make informed decisions on visual field progression in glaucoma.

Determining visual field progression remains one of the most important but challenging aspects of glaucoma management. In addition to influencing the clinical care of patients, visual field progression is used as an end-point in many clinical trials in glaucoma and ocular hypertension. 1 2 3 4 5 6 Visual fields vary from one examination to the next, even if no real change has taken place. This variability can both mask and mimic glaucomatous change, and several statistical methods have been devised to distinguish it from true progression. 
One method of identifying visual field change is to derive statistical limits of test–retest variability in a group of patients with glaucoma tested frequently in close succession (e.g., over a few weeks) so that any differences between the test results can more likely be attributed to variability rather than to true change. 7 8 These limits are then applied to visual fields of patients who are observed over time. If, at any given test location, the difference between the current test and a previously established baseline is beyond the lower limit of test–retest variability, the location has likely worsened. This event analysis, incorporated into software (the older Statpac program of the Humphrey Field Analyzer [HFA]; Carl Zeiss Meditec, Dublin, CA) as glaucoma change probability (GCP) analysis, 9 is based on total deviation (i.e., the difference between the observed threshold and the age-corrected normal value). 
Because glaucoma is an age-related disease, glaucomatous visual field changes must be distinguished from those caused by other age-related diseases, principally cataract. The glaucoma progression analysis (GPA) was devised for this purpose. It is based on pattern deviation, defined as the deviation from age-corrected values adjusted for the general height (GH) of the visual field. The GH is estimated from the relative depression of the 15th percentile of the total deviation values. By adjusting for any diffuse changes in visual field sensitivity, pattern deviation analyses seek to separate glaucomatous changes from those due to cataract, on the assumption that cataract causes a diffuse reduction of visual field sensitivity, whereas glaucoma leads to focal visual field changes. 10 The hypothesis that glaucomatous visual field loss is exclusively focal, however, has often been challenged. Several studies have shown that focal visual field loss in glaucoma is usually associated with a diffuse component, 11 12 whereas others have demonstrated that a minority of patients with glaucoma may present with exclusively diffuse visual field losses. 13  
Trend analyses of visual field change (available in the Progressor 14 and the Peridata 15 software packages), are based on a statistical approach different from that of the event analyses. Using regression analysis, they establish the statistical significance of change over the entire series of examinations. Unlike event analyses of visual field change, trend analyses relate the magnitude of change to the variability observed within the individual data series and therefore do not rely on population-derived variability limits that can be inappropriate in individual patients. 
The objective of this study was to compare total and pattern deviation change analyses of visual field progression in patients with glaucoma with a long prospective follow-up and no clinical evidence of increasing media opacity. To estimate the specificity of the progression criteria, we applied the same analyses to a group of healthy control eyes followed up identically, as well as estimating the rate of visual field improvement in the patients with glaucoma. 
Methods
Study Design and Material
Analyses were performed on visual field data collected for a prospective longitudinal study of patients with glaucoma and healthy control subjects. Patients with glaucoma were recruited consecutively from the clinics at the QEII Health Sciences Centre in Halifax, Nova Scotia. Patients were included if they had a clinical diagnosis of open-angle glaucoma based on optic disc and/or visual field changes, a visual field mean deviation (MD) equal to or better than −10 dB, and no other ocular disease. Healthy control subjects were recruited from relatives, employees of a local telephone company, and church groups, and had to have normal findings in an eye examination at baseline. For both groups of subjects, the inclusion criteria also stipulated a refractive error within ±5 D equivalent sphere or 3 D of astigmatism and visual acuity (VA) equal to or better than 6/12 (+0.3 logMAR). In accordance with the Declaration of Helsinki, all subjects gave informed consent to participate in the study, and the protocol was approved by the Ethics Review Board of the QEII Health Sciences Centre. 
Patients and control subjects were examined in 6-month intervals with standard automated perimetry (Humphrey Field Analyzer 30-2 test, full-threshold strategy). For the present analyses, only the 52 test locations of the 24-2 test were used (excluding the two blind-spot locations). We intended to compare the differences between total and pattern deviation analyses in eyes without clinical evidence of increasing media opacity. Therefore, if participants had undergone cataract surgery during the course of the study, only those visual fields obtained later than 2 months after surgery were included in the analysis. Also, if during two or more examinations of the study a participant’s VA was reduced below 0.4 logMAR (20/50) or was reduced by >0.2 logMAR below baseline, they were excluded from the analysis. To reduce the influence of learning effects, we excluded the first visual field examination of each subject from the analyses. 
In the patients with glaucoma, both eyes were examined, whereas in healthy control subjects, only the randomly selected study eye was examined. We used both eyes of the patients with glaucoma in the descriptive analyses if they met the inclusion criteria. For statistical significance testing, however, data were analyzed from only one randomly selected eye of each patient. 
Analyses of Visual Field Progression
Event Analyses of Total and Pattern Deviation.
The limits of test–retest variability of total and pattern deviation were determined in 64 patients with glaucoma who had been examined five times within 4 weeks for an earlier study. 8 Treating the five tests as interchangeable, we examined all 30 permutations of two baseline tests and one follow-up test. For each level of baseline deviation (in steps of 1 dB), we derived the empiric 5th and 95th percentiles of the distribution of follow-up data. The 5% and 95% change probability limits were derived by smoothing the empiric percentiles using fourth-order polynomials. 
Event analyses of visual field progression and improvement were performed by a point-by-point comparison between the baseline deviation, calculated as the mean value at the first two examinations, with all subsequent examinations. Test locations at which the deviation was more negative than the lower test–retest limit of the baseline value were classified as having deteriorated and those with deviations more positive than the upper limit were classified as having improved. The criteria for visual field progression and improvement were defined in terms of locations with changes from baseline on three consecutive follow-up examinations (Fig. 1) , so that progression or improvement could be ascertained only after five or more examinations (two baseline and three follow-up tests). 
Trend Analyses of Total and Pattern Deviation.
Trend analyses were performed by point-wise linear regression of the deviation values with follow-up time, using the two-omitting method to increase specificity. 16 This method establishes two regression lines: one while omitting the last test, and the other while omitting the penultimate test. Locations were flagged as progressing if the slopes of the deviation value were more negative than −1 dB/y and different from zero at a significance level of P < 5% on both occasions (Fig. 2) . Improvements were flagged using the same approach, with slopes greater than +1 dB/y. Similar to the event analyses described for total and patter deviation, trend analyses were performed when at least five visual field examinations were available for analysis. 
Comparison between Analyses
Progression rates were established by Kaplan-Meier analysis to account for censoring that occurs when subjects are lost to follow-up. 17 18 For a comparison between the total and pattern deviation analyses, we established which eyes had been classified as having progressed after 10 years of follow-up. This information was used to plot area-proportional Venn diagrams 19 to illustrate the agreement between the methods. Differences in time-to-progression were investigated only in those eyes that had been classified as having progressed with both total and pattern deviation analyses. To explain the differences between the total and pattern deviation analyses of visual field progression, the slope of GH over time was plotted against the slope of MD over time. In eyes with purely focal progression, the slope of the MD index would be negative, whereas that of GH would be zero. In eyes with entirely diffuse change, both MD and GH would show similar, negative slopes. For this analysis, eyes were stratified into three equal groups, according to baseline MD. 
To judge the likely specificity of the criteria for visual field progression, we derived the rate of progression in the healthy control eyes, as well as the rate of visual field improvement in the glaucomatous eyes, using the same criteria as had been used for detecting progression. 
Results
Subject Groups
Data from 101 patients with open-angle glaucoma and 102 healthy control subjects were included in the analysis. Details for both groups of subjects are shown in Table 1
Comparison between Total and Pattern Deviation Analyses
Both event and trend analyses of total deviation detected progression in a larger number of glaucomatous eyes than the corresponding pattern deviation analyses. With the total deviation event analysis, 50 (30%) glaucomatous eyes were classified as having progressed after 10 years of follow-up, but only 35 (21%) met the criteria with the pattern deviation analysis (Figs. 3a 3b 3c) . Similar findings were obtained with the trend analyses; after a 10-year follow-up, the total deviation trend analysis had classified progression in 61 (36%) eyes of which only 41 (24%) were classified with pattern deviation analyses (Figs. 3e 3f 3g) . The differences between the progression rates obtained with total and pattern deviation analyses were statistically significant in both event and trend analyses (P < 0.01; McNemar test). In eyes that were classified as progressing, with both the total and pattern deviation analyses, the time-to-progression differences between both analyses followed a skewed distribution that suggested earlier detection with the total deviation analyses (Figs. 3d 3h)
The systematic differences in time-to-progression were confirmed in a point-by-point analysis (Fig. 4) . Although most test locations were classified as having progressed at the same time, the distributions of the differences showed a long tail (earlier detection with the total deviation analyses). 
In most, but not all, eyes, the total deviation analyses classified more test locations as having progressed than did the pattern deviation analyses. These findings applied equally to both event and trend analyses (Fig. 5)
To interpret the observed differences between the total and pattern deviation analyses of progression, we compared the change in GH to the change in the MD index in both groups of subjects (Fig. 6) . Eyes with purely focal change would be associated with data points that lie on a horizontal line (slope of GH ∼ 0), whereas those with purely diffuse changes would lie along the dotted diagonal line (slope of GH ∼ slope of MD). Most glaucomatous eyes showed negative GH slopes as well as negative MD slopes, indicating that purely focal changes occurred rarely. In the healthy control eyes, almost all changes were diffuse. 
To obtain an indication of how specific (i.e., robust to random variability) the evaluated progression criteria were likely to be, we derived the progression rate in the healthy control eyes, as well as the improvement rate in the glaucoma eyes. Visual field progression in the healthy control eyes was rare with both event and trend criteria. After 10 years of follow-up, the cumulative progression rates in the healthy control eyes were between 2% and 6% in all analyses (Fig. 3 , dotted curves), and differences between the analyses were not statistically significant (P ≥ 0.18, paired comparison of proportions). The rate of visual field improvement in the glaucomatous eyes was also low (Table 2) . After 10 years of follow-up, the improvement rates were between 1% and 7% in all four analyses. Together, these findings imply that the analyses and criteria evaluated in this study possess high specificity. 
Case Examples
The following examples illustrate some of the findings of our study. The patient in the first example (Fig. 7) , showed widespread deterioration in both the inferior and superior hemifields, and changes appeared larger with the total deviation than with the pattern deviation analyses. In the second example (Fig. 8) , the grayscale plots revealed a large degree of progression that was underestimated by all four analyses, owing to large variability within the visual field series. In this example, the total-deviation analyses identified more test locations as having changed than the pattern deviation analyses. 
Discussion
Visual field progression is often the only clinical sign that a change in the patient’s vision has taken place. It is therefore usually regarded as a strong indication that the treatment regimen should be intensified. Many recent clinical trials in glaucoma have yielded important scientific information and jointly have the potential of dictating strong practice guidelines throughout the spectrum of the disease, from suspected to advanced glaucoma. However, one impediment of translating the results of these trials into clinical practice is the large difference in visual field endpoints that had been specified. This lack of standardization is understandable, given the extent of knowledge about the longitudinal behavior of visual fields and their variability when the trials were designed and executed. Newer and potentially better methods of analyzing visual field progression may, however, allow standardization of progression criteria, such that information from different studies can be compared and translated to clinical practice. 
One of these new tools is the glaucoma progression analysis (GPA) which is based on pattern deviation values which compensate for changes in the GH of the visual field, thought to be caused primarily by cataract. 20 Important glaucoma-related clinical trials, such as the Early Manifest Glaucoma Trial, 6 have used similar analyses for defining endpoints. To date, however, the performance of the GPA has been evaluated in surprisingly few studies 21 and a comparison with the GCP analysis, based on total deviation, has been reported in only one published study. 22 The specificity of the GPA has not been published. Therefore, our objectives in this study were to investigate the differences between total and pattern deviation analyses of visual field progression in subjects without clinical evidence of increasing media opacity, as well as to estimate the specificity of the progression analyses over time. 
Our results indicate that between 10% and 15% of glaucomatous eyes show evidence of progression with the total deviation but not the pattern deviation analyses. In individual eyes, on average, total deviation analyses also classified more test locations as having progressed. Both these findings were similar with event and trend analyses. They agree with those of Katz 22 which showed a lower incidence of progression with the pattern deviation event analysis (GPA) compared with total deviation (GCP). One possible explanation for these findings is the confounding factor of concurrently progressing cataract, which would have been reduced with pattern deviation analyses. However, we feel that progressing cataract is unlikely to be the sole explanation for the differences between the total and pattern deviation analyses. We excluded patients whose visual acuity had deteriorated by more than 2 lines over the course of the study, as well as those who had undergone cataract surgery. Moreover, recent evidence established in the same population suggests that cataract surgery, despite a mean improvement of 2 lines of visual acuity, results in only a very small and statistically insignificant improvement in total deviation. 23 Almost all eyes which changed in either MD or GH changed with both indices, indicating that glaucomatous visual field progression comprises both focal and diffuse components. Moreover, in patients with moderately advanced damage, even purely focal changes (had they existed) would have affected GH 24 and therefore would have led to a systematic underestimation of change with the pattern deviation analyses. In patients with advanced damage, the pattern deviation calculations often produce obviously misleading results. 25  
Although, in most glaucomatous eyes, the total deviation analyses revealed a larger number of locations as progressing than did the pattern deviation analyses (Fig. 4) , the opposite was true in a small number of eyes. This finding can be explained by the fact that pattern deviation analyses may compensate for overall improvements of the visual field, which may partly mask concurrent progression in the total deviation analyses. Overall improvements in the visual field may be due to learning 26 27 28 but may also occur as an artifact of inappropriate age adjustment if the true age changes in an individual are different from those observed in a population. For example, visual fields that change relatively less would show increasing (i.e., more positive) total deviation values over time. Even though we had excluded the first visual field test of each subject from our analyses, statistically significant positive slopes of MD and GH with time occurred in a significant number of patients as well as healthy control subjects (Fig. 5) . In these subjects, pattern deviation analyses may have advantages over those based on total deviation. 
Although computer simulations have been used to investigate and compare the performance of different criteria to determine progression, 16 29 such simulations depend largely on comprehensive models for glaucomatous visual field change that are difficult to validate. We were interested primarily in the specificity of the progression analyses (i.e., the likelihood that a given visual field was classified as having progressed although no real change had taken place). To obtain estimates of specificity from real clinical data, we derived two independent proxy measures of specificity. First, visual field progression was analyzed in a group of healthy control subjects. The rationale for using the rate of progression in the normal control eyes as a proxy index of specificity in patients with glaucoma is based on the assumptions that: (1) normal ageing effects are accounted for by both total and pattern deviation analyses, (2) visual field progression is specific to glaucoma, and (3) both analyses compensate for the different levels of variability observed in patients with glaucoma and control subjects. Any deterioration observed in the healthy control subjects should therefore be due only to random variability. With any of the analyses performed, the progression rate in healthy eyes was low; after 10 years of follow-up, only between 2% and 6% of control eyes had met the criteria for progression. Second, the rate of visual field improvement was determined in the glaucomatous eyes, based on the rationale that genuine visual field improvement in glaucoma is rare. Although learning effects, which manifest as visual field improvement, occur to varying degrees and over various time periods, 26 27 28 30 31 they would lead us to underestimate, rather than overestimate, the specificity of the progression analyses. With both event and trend analyses, the improvement rates were below 10% after 10 years of follow-up, suggesting that the true specificity of the progression analyses was high. Although the differences between the proxy indices of specificity (rates of progression in control eyes, rates of improvement in both groups of subjects) with the event and trend analyses were not statistically significant, we cannot be confident that the specificities of the criteria are truly similar. To obtain sufficient statistical power to detect a real difference between two low rates, very large sample sizes are necessary. For example, a sample size of approximately 500 would be necessary to obtain 80% power to confirm the 5% difference between specificities of 90% and 95%. Because these rather small absolute differences in specificity can be clinically important, the lack of statistical power for comparisons of diagnostic tests is an important, though not easily addressed, problem. 
In summary, the pattern deviation analyses generally yielded a smaller number of progressing eyes than did the total deviation analyses. Although the average differences in time-to-progression were small (earlier detection with total deviation by approximately 6 months, or one examination), the spread of differences in individual patients was sufficiently wide to be of clinical importance. If the clinical evidence is not suggestive of increasing media opacity, pattern deviation analyses may significantly underestimate visual field progression in glaucoma. Because, in individual eyes, either total or pattern deviation analyses can have distinct advantages, we argue that clinicians should have access to both methods for clinical decision-making in glaucoma. 
 
Figure 1.
 
Illustration of event analysis at one test location. Solid horizontal line: the mean of the two baseline test results. If the deviation at a subsequent follow-up examination is more negative than the lower 5% limit of test–retest variability (dotted line), deterioration is flagged with a filled triangle. In this example, the progression criterion has been met after the last observation (3 years).
Figure 1.
 
Illustration of event analysis at one test location. Solid horizontal line: the mean of the two baseline test results. If the deviation at a subsequent follow-up examination is more negative than the lower 5% limit of test–retest variability (dotted line), deterioration is flagged with a filled triangle. In this example, the progression criterion has been met after the last observation (3 years).
Figure 2.
 
Illustration of trend analysis with point-wise linear regression using the two-omitting method. A test location was classified as having progressed if the slope of the linear regression line was more negative than −1 dB/y and significantly different from zero at P < 5%. These criteria must be met when the last examination is excluded (a) and also when the penultimate examination is excluded (b). Solid lines: regression obtained by omitting the last or penultimate observation (open circles). Dotted line: regression including all observations. In this example, the location was not classified as having progressed because the slope was not significant when the last observation was excluded (a).
Figure 2.
 
Illustration of trend analysis with point-wise linear regression using the two-omitting method. A test location was classified as having progressed if the slope of the linear regression line was more negative than −1 dB/y and significantly different from zero at P < 5%. These criteria must be met when the last examination is excluded (a) and also when the penultimate examination is excluded (b). Solid lines: regression obtained by omitting the last or penultimate observation (open circles). Dotted line: regression including all observations. In this example, the location was not classified as having progressed because the slope was not significant when the last observation was excluded (a).
Table 1.
 
Demographic Details of Patients with Glaucoma and Healthy Control Subjects
Table 1.
 
Demographic Details of Patients with Glaucoma and Healthy Control Subjects
Glaucoma (n = 101) Controls (n = 102)
Age at baseline (y) 61 (43, 81) 47 (34–72)
Eyes (n) 168 102
Follow-up time (y) 10 (3–13) 10 (3–13)
Examinations (n) 21 (6–26) 11 (7–26)
Baseline MD (dB) −3.3 (−9.1 to −0.4) −0.9 (−3.9 to +1.3)
Final MD (dB) −4.8 (−14.6 to 0.0) −1.0 (−4.5 to +1.2)
Baseline VA (logMAR) 0.00 (0.00 to +0.15) 0.00 (0.00 to +0.15)
Final VA (logMAR) 0.00 (0.00 to +0.15) 0.00 (0.00 to +0.15)
Figure 3.
 
Comparison of visual field progression with total and pattern deviation analyses. (ad) Data from the event analyses; (eh) those of the trend analyses. Solid lines in the Kaplan-Meier plots (a, b, e, f) give the progression rate in the glaucomatous eyes, and the dotted lines give the progression rate in the healthy control eyes. The percentages above the curves refer to the Kaplan-Meier estimate of the cumulative progression rate after 10 years of follow-up. Area-proportional Venn diagrams (c, g) show the agreement between total and pattern deviation analyses after 10 years of follow-up. Histograms (d, h) of the differences in time-to-progression show a skew toward earlier detection by the total deviation analyses.
Figure 3.
 
Comparison of visual field progression with total and pattern deviation analyses. (ad) Data from the event analyses; (eh) those of the trend analyses. Solid lines in the Kaplan-Meier plots (a, b, e, f) give the progression rate in the glaucomatous eyes, and the dotted lines give the progression rate in the healthy control eyes. The percentages above the curves refer to the Kaplan-Meier estimate of the cumulative progression rate after 10 years of follow-up. Area-proportional Venn diagrams (c, g) show the agreement between total and pattern deviation analyses after 10 years of follow-up. Histograms (d, h) of the differences in time-to-progression show a skew toward earlier detection by the total deviation analyses.
Figure 4.
 
Point-by-point differences in time-to-progression with event (a) and trend (b) analyses of total and pattern deviation. On average, the total deviation–based analyses detected progression approximately 6 months (1 examination) earlier than the pattern deviation analyses. Numbers on top of each histogram bar give percentages.
Figure 4.
 
Point-by-point differences in time-to-progression with event (a) and trend (b) analyses of total and pattern deviation. On average, the total deviation–based analyses detected progression approximately 6 months (1 examination) earlier than the pattern deviation analyses. Numbers on top of each histogram bar give percentages.
Figure 5.
 
Number of test locations classified as having progressed with total versus pattern deviation in event (a) and trend analyses (b). For this comparison, one visual field was selected at random from the entire available series of each eye. Log axes avoid an overemphasis on fields with many progressing locations. Open circles: single observations; numbers: multiple observations.
Figure 5.
 
Number of test locations classified as having progressed with total versus pattern deviation in event (a) and trend analyses (b). For this comparison, one visual field was selected at random from the entire available series of each eye. Log axes avoid an overemphasis on fields with many progressing locations. Open circles: single observations; numbers: multiple observations.
Figure 6.
 
Slope of GH plotted against slope of MD, in patients with glaucoma (a) and healthy control subjects (b). To avoid clutter, data points are shown only if either the MD or the GH slope was significantly different from zero (P < 5%). Visual fields with purely focal changes would lie in on the horizontal line, those with purely diffuse changes on the diagonal dotted line of equality. Eyes were stratified according to baseline MD. Three data points with MD slopes < −1.5 dB/y were omitted in (a).
Figure 6.
 
Slope of GH plotted against slope of MD, in patients with glaucoma (a) and healthy control subjects (b). To avoid clutter, data points are shown only if either the MD or the GH slope was significantly different from zero (P < 5%). Visual fields with purely focal changes would lie in on the horizontal line, those with purely diffuse changes on the diagonal dotted line of equality. Eyes were stratified according to baseline MD. Three data points with MD slopes < −1.5 dB/y were omitted in (a).
Table 2.
 
Number of Improvement Events in Eyes of Healthy Control Subjects and Patients with Glaucoma with Event and Trend Analyses of Total and Pattern Deviation
Table 2.
 
Number of Improvement Events in Eyes of Healthy Control Subjects and Patients with Glaucoma with Event and Trend Analyses of Total and Pattern Deviation
Event Analyses Trend Analyses
Total Deviation Pattern Deviation Total Deviation Pattern Deviation
Control eyes (n = 102) 2 (2%) 0 (0%) 7 (7%) 2 (2%)
Glaucomatous eyes (n = 168) 3 (2%) 1 (1%) 12 (7%) 12 (7%)
Figure 7.
 
Case example 1. Grayscale plots of first (a) and last (b) visual field test results in a patient with extensive progression. (c, d) Locations with significant deterioration beyond the 5th percentile of test−retest variability on the last three consecutive examinations. (e, f) Test locations meeting the two-omitting criterion on the last examination.
Figure 7.
 
Case example 1. Grayscale plots of first (a) and last (b) visual field test results in a patient with extensive progression. (c, d) Locations with significant deterioration beyond the 5th percentile of test−retest variability on the last three consecutive examinations. (e, f) Test locations meeting the two-omitting criterion on the last examination.
Figure 8.
 
Case example 2. Grayscale plots of first (a) and last (b) visual field test results in a patient with extensive progression. (c, d) Locations with significant deterioration beyond the 5th percentile of test−retest variability on the last three consecutive examinations. (e, f) Test locations meeting the two-omitting criterion on the last examination.
Figure 8.
 
Case example 2. Grayscale plots of first (a) and last (b) visual field test results in a patient with extensive progression. (c, d) Locations with significant deterioration beyond the 5th percentile of test−retest variability on the last three consecutive examinations. (e, f) Test locations meeting the two-omitting criterion on the last examination.
The authors thank Chris Johnson and colleagues at Discoveries in Sight (Portland, OR) for providing a large part of the test–retest visual field data used to estimate the variability limits in the event analyses. 
Advanced Glaucoma Intervention Study. 2. Visual field test scoring and reliability. Ophthalmology. 1994;101:1445–1455. [CrossRef] [PubMed]
SchulzerM. Errors in the diagnosis of visual field progression in normal-tension glaucoma. Ophthalmology. 1994;101:1589–1594. [CrossRef] [PubMed]
DranceS, AndersonDR, SchulzerM. Risk factors for progression of visual field abnormalities in normal-tension glaucoma. Am J Ophthalmol. 2001;131:699–708. [CrossRef] [PubMed]
The European Glaucoma Prevention Study. Design and baseline description of the participants. Ophthalmology. 2002;109:1612–1621. [CrossRef] [PubMed]
KeltnerJL, JohnsonCA, CelloKE, et al. Classification of visual field abnormalities in the ocular hypertension treatment study. Arch Ophthalmol. 2003;121:643–650. [CrossRef] [PubMed]
HeijlA, LeskeMC, BengtssonB, BengtssonB, HusseinM. Measuring visual field progression in the Early Manifest Glaucoma Trial. Acta Ophthalmol Scand. 2003;81:286–293. [CrossRef] [PubMed]
HeijlA, LindgrenA, LindgrenG. Test-retest variability in glaucomatous visual fields. Am J Ophthalmol. 1989;108:130–135. [CrossRef] [PubMed]
ChauhanBC, JohnsonCA. Test-retest variability of frequency-doubling perimetry and conventional perimetry in glaucoma patients and normal subjects. Invest Ophthalmol Vis Sci. 1999;40:648–656. [PubMed]
HeijlA, LindgrenG, LindgrenA, et al. Extended empirical statistical package for evaluation of single and multiple fields in glaucoma: Statpac 2.MillsR HeijlA eds. Perimetry Update 1990/1. 1991;303–315.Kugler Publications Amsterdam.
HeijlA, LindgrenG, OlssonJ, AsmanP. Visual field interpretation with empiric probability maps. Arch Ophthalmol. 1989;107:204–208. [CrossRef] [PubMed]
HensonDB, ArtesPH, ChauhanBC. Diffuse loss of sensitivity in early glaucoma. Invest Ophthalmol Vis Sci. 1999;40:3147–3151. [PubMed]
DranceSM. Diffuse visual field loss in open-angle glaucoma. Ophthalmology. 1991;98:1533–1538. [CrossRef] [PubMed]
ChauhanBC, LeBlancRP, ShawAM, ChanAB, McCormickTA. Repeatable diffuse visual field loss in open-angle glaucoma. Ophthalmology. 1997;104:532–538. [CrossRef] [PubMed]
FitzkeFW, HitchingsRA, PoinoosawmyD, McNaughtAI, CrabbDP. Analysis of visual field progression in glaucoma. Br J Ophthalmol. 1996;80:40–48. [CrossRef] [PubMed]
Peridata website. 2005;http://www.peridata org. Accessed October 11, 2005.
GardinerSK, CrabbDP. Examination of different pointwise linear regression methods for determining visual field progression. Invest Ophthalmol Vis Sci. 2002;43:1400–1407. [PubMed]
AltmanDG, BlandJM. Time to event (survival) data. BMJ. 1998;317:468–469. [PubMed]
BlandJM, AltmanDG. Survival probabilities (the Kaplan-Meier method). BMJ. 1998;317:1572. [CrossRef] [PubMed]
ChowS, RuskeyF. Drawing area-proportional Venn and Euler diagrams. . 2004.466–477.
BengtssonB, LindgrenA, HeijlA, et al. Perimetric probability maps to separate change caused by glaucoma from that caused by cataract. Acta Ophthalmol Scand. 1997;75:184–188. [PubMed]
HeijlA, LeskeMC, BengtssonB, et al. Reduction of intraocular pressure and glaucoma progression: results from the Early Manifest Glaucoma Trial. Arch Ophthalmol. 2002;120:1268–1279. [CrossRef] [PubMed]
KatzJ. A comparison of the pattern- and total deviation-based Glaucoma Change Probability programs. Invest Ophthalmol Vis Sci. 2000;41:1012–1016. [PubMed]
CarrilloMM, ArtesPH, NicolelaMT, LeBlancRP, ChauhanBC. Effect of cataract extraction on the visual field of patients with glaucoma. Arch Ophthalmol. 2005;123:929–932. [CrossRef] [PubMed]
AsmanP, WildJM, HeijlA. Appearance of the pattern deviation map as a function of change in area of localized field loss. Invest Ophthalmol Vis Sci. 2004;45:3099–3106. [CrossRef] [PubMed]
BlumenthalEZ, Sapir-PichhadzeR. Misleading statistical calculations in far-advanced glaucomatous visual field loss. Ophthalmology. 2003;110:196–200. [CrossRef] [PubMed]
WoodJM, WildJM, HusseyMK, CrewsSJ. Serial examination of the normal visual field using Octopus automated projection perimetry: evidence for a learning effect. Acta Ophthalmol (Copenh). 1987;65:326–333. [PubMed]
WildJM, Dengler-HarlesM, SearleAE, O’NeillEC, CrewsSJ. The influence of the learning effect on automated perimetry in patients with suspected glaucoma. Acta Ophthalmol (Copenh). 1989;67:537–545. [PubMed]
WernerEB, KrupinT, AdelsonA, FeitlME. Effect of patient experience on the results of automated perimetry in glaucoma suspect patients. Ophthalmology. 1990;97:44–48. [CrossRef] [PubMed]
VestiE, JohnsonCA, ChauhanBC. Comparison of different methods for detecting glaucomatous visual field progression. Invest Ophthalmol Vis Sci. 2003;44:3873–3879. [CrossRef] [PubMed]
MarraG, FlammerJ. The learning and fatigue effect in automated perimetry. Graefes Arch Clin Exp Ophthalmol. 1991;229:501–504. [CrossRef] [PubMed]
WildJM, SearleAE, Dengler-HarlesM, O’NeillEC. Long-term follow-up of baseline learning and fatigue effects in the automated perimetry of glaucoma and ocular hypertensive patients. Acta Ophthalmol (Copenh). 1991;69:210–216. [PubMed]
Figure 1.
 
Illustration of event analysis at one test location. Solid horizontal line: the mean of the two baseline test results. If the deviation at a subsequent follow-up examination is more negative than the lower 5% limit of test–retest variability (dotted line), deterioration is flagged with a filled triangle. In this example, the progression criterion has been met after the last observation (3 years).
Figure 1.
 
Illustration of event analysis at one test location. Solid horizontal line: the mean of the two baseline test results. If the deviation at a subsequent follow-up examination is more negative than the lower 5% limit of test–retest variability (dotted line), deterioration is flagged with a filled triangle. In this example, the progression criterion has been met after the last observation (3 years).
Figure 2.
 
Illustration of trend analysis with point-wise linear regression using the two-omitting method. A test location was classified as having progressed if the slope of the linear regression line was more negative than −1 dB/y and significantly different from zero at P < 5%. These criteria must be met when the last examination is excluded (a) and also when the penultimate examination is excluded (b). Solid lines: regression obtained by omitting the last or penultimate observation (open circles). Dotted line: regression including all observations. In this example, the location was not classified as having progressed because the slope was not significant when the last observation was excluded (a).
Figure 2.
 
Illustration of trend analysis with point-wise linear regression using the two-omitting method. A test location was classified as having progressed if the slope of the linear regression line was more negative than −1 dB/y and significantly different from zero at P < 5%. These criteria must be met when the last examination is excluded (a) and also when the penultimate examination is excluded (b). Solid lines: regression obtained by omitting the last or penultimate observation (open circles). Dotted line: regression including all observations. In this example, the location was not classified as having progressed because the slope was not significant when the last observation was excluded (a).
Figure 3.
 
Comparison of visual field progression with total and pattern deviation analyses. (ad) Data from the event analyses; (eh) those of the trend analyses. Solid lines in the Kaplan-Meier plots (a, b, e, f) give the progression rate in the glaucomatous eyes, and the dotted lines give the progression rate in the healthy control eyes. The percentages above the curves refer to the Kaplan-Meier estimate of the cumulative progression rate after 10 years of follow-up. Area-proportional Venn diagrams (c, g) show the agreement between total and pattern deviation analyses after 10 years of follow-up. Histograms (d, h) of the differences in time-to-progression show a skew toward earlier detection by the total deviation analyses.
Figure 3.
 
Comparison of visual field progression with total and pattern deviation analyses. (ad) Data from the event analyses; (eh) those of the trend analyses. Solid lines in the Kaplan-Meier plots (a, b, e, f) give the progression rate in the glaucomatous eyes, and the dotted lines give the progression rate in the healthy control eyes. The percentages above the curves refer to the Kaplan-Meier estimate of the cumulative progression rate after 10 years of follow-up. Area-proportional Venn diagrams (c, g) show the agreement between total and pattern deviation analyses after 10 years of follow-up. Histograms (d, h) of the differences in time-to-progression show a skew toward earlier detection by the total deviation analyses.
Figure 4.
 
Point-by-point differences in time-to-progression with event (a) and trend (b) analyses of total and pattern deviation. On average, the total deviation–based analyses detected progression approximately 6 months (1 examination) earlier than the pattern deviation analyses. Numbers on top of each histogram bar give percentages.
Figure 4.
 
Point-by-point differences in time-to-progression with event (a) and trend (b) analyses of total and pattern deviation. On average, the total deviation–based analyses detected progression approximately 6 months (1 examination) earlier than the pattern deviation analyses. Numbers on top of each histogram bar give percentages.
Figure 5.
 
Number of test locations classified as having progressed with total versus pattern deviation in event (a) and trend analyses (b). For this comparison, one visual field was selected at random from the entire available series of each eye. Log axes avoid an overemphasis on fields with many progressing locations. Open circles: single observations; numbers: multiple observations.
Figure 5.
 
Number of test locations classified as having progressed with total versus pattern deviation in event (a) and trend analyses (b). For this comparison, one visual field was selected at random from the entire available series of each eye. Log axes avoid an overemphasis on fields with many progressing locations. Open circles: single observations; numbers: multiple observations.
Figure 6.
 
Slope of GH plotted against slope of MD, in patients with glaucoma (a) and healthy control subjects (b). To avoid clutter, data points are shown only if either the MD or the GH slope was significantly different from zero (P < 5%). Visual fields with purely focal changes would lie in on the horizontal line, those with purely diffuse changes on the diagonal dotted line of equality. Eyes were stratified according to baseline MD. Three data points with MD slopes < −1.5 dB/y were omitted in (a).
Figure 6.
 
Slope of GH plotted against slope of MD, in patients with glaucoma (a) and healthy control subjects (b). To avoid clutter, data points are shown only if either the MD or the GH slope was significantly different from zero (P < 5%). Visual fields with purely focal changes would lie in on the horizontal line, those with purely diffuse changes on the diagonal dotted line of equality. Eyes were stratified according to baseline MD. Three data points with MD slopes < −1.5 dB/y were omitted in (a).
Figure 7.
 
Case example 1. Grayscale plots of first (a) and last (b) visual field test results in a patient with extensive progression. (c, d) Locations with significant deterioration beyond the 5th percentile of test−retest variability on the last three consecutive examinations. (e, f) Test locations meeting the two-omitting criterion on the last examination.
Figure 7.
 
Case example 1. Grayscale plots of first (a) and last (b) visual field test results in a patient with extensive progression. (c, d) Locations with significant deterioration beyond the 5th percentile of test−retest variability on the last three consecutive examinations. (e, f) Test locations meeting the two-omitting criterion on the last examination.
Figure 8.
 
Case example 2. Grayscale plots of first (a) and last (b) visual field test results in a patient with extensive progression. (c, d) Locations with significant deterioration beyond the 5th percentile of test−retest variability on the last three consecutive examinations. (e, f) Test locations meeting the two-omitting criterion on the last examination.
Figure 8.
 
Case example 2. Grayscale plots of first (a) and last (b) visual field test results in a patient with extensive progression. (c, d) Locations with significant deterioration beyond the 5th percentile of test−retest variability on the last three consecutive examinations. (e, f) Test locations meeting the two-omitting criterion on the last examination.
Table 1.
 
Demographic Details of Patients with Glaucoma and Healthy Control Subjects
Table 1.
 
Demographic Details of Patients with Glaucoma and Healthy Control Subjects
Glaucoma (n = 101) Controls (n = 102)
Age at baseline (y) 61 (43, 81) 47 (34–72)
Eyes (n) 168 102
Follow-up time (y) 10 (3–13) 10 (3–13)
Examinations (n) 21 (6–26) 11 (7–26)
Baseline MD (dB) −3.3 (−9.1 to −0.4) −0.9 (−3.9 to +1.3)
Final MD (dB) −4.8 (−14.6 to 0.0) −1.0 (−4.5 to +1.2)
Baseline VA (logMAR) 0.00 (0.00 to +0.15) 0.00 (0.00 to +0.15)
Final VA (logMAR) 0.00 (0.00 to +0.15) 0.00 (0.00 to +0.15)
Table 2.
 
Number of Improvement Events in Eyes of Healthy Control Subjects and Patients with Glaucoma with Event and Trend Analyses of Total and Pattern Deviation
Table 2.
 
Number of Improvement Events in Eyes of Healthy Control Subjects and Patients with Glaucoma with Event and Trend Analyses of Total and Pattern Deviation
Event Analyses Trend Analyses
Total Deviation Pattern Deviation Total Deviation Pattern Deviation
Control eyes (n = 102) 2 (2%) 0 (0%) 7 (7%) 2 (2%)
Glaucomatous eyes (n = 168) 3 (2%) 1 (1%) 12 (7%) 12 (7%)
×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×