Prerandomization dropouts occurred because of visual field defects that were located outside the central 30° area (n = 7) and fixation problems (n = 2). Postrandomization dropouts occurred because of mild headache during training (n = 2) or lack of time to complete the schedule (n = 2). gVRT indicates vision restoration training for glaucoma.
The x-axis and the y-axis show degrees of visual angle. Upper panels: To assess the visual fields with high-resolution computer-based perimetry, suprathreshold stimuli were presented at random locations, from which simple detection charts were created. White indicates intact visual field sector; black, regions of absolute blindness; and gray, areas where response accuracy is inconsistent, which are known as areas of residual vision6 or relative defects. Gray regions are interpreted by us as representing partial damage in which only some cells remain connected with their target structure. The extent of deafferentiation has a functional correlate: with the greater the loss, the lower the detection accuracy, ranging from white through shades of gray to black. Lower panels: Standard perimetry results at lower resolution. For more details, see eTable 1 and eTable 2 in the Supplement.
Pre-post differences in the vision restoration training for glaucoma group (experimental) and placebo group (control) measured by high-resolution perimetry (HRP). W/W indicates white-on-white perimetry; B/Y, blue-on-yellow perimetry. Asterisks indicate statistically significant between-group differences. For specific P values, see the Results section.
Shown are the horizontal and vertical average mean gaze position and average mean gaze position variability before and after training in each patient of the vision restoration training for glaucoma group (E1-E15) and placebo group (C1-C15). For further explanation, see eTable 2 in the Supplement.
eTable 1. Patient Baseline Characteristics
eTable 2. Mean Gaze Position (in Degrees) and Mean Gaze Position Variability (SD in Degrees) on the Horizontal (X) and Vertical (Y) Axes Before and After Training in the gVRT and Placebo Groups
Sabel BA, Gudlin J. Vision restoration training for glaucoma: a randomized clinical trial. JAMA Ophthalmol. Published online February 6, 2014. doi:10.1001/jamaophthalmol.2013.7963.
Sabel BA, Gudlin J. Vision Restoration Training for GlaucomaA Randomized Clinical Trial. JAMA Ophthalmol. 2014;132(4):381-389. doi:10.1001/jamaophthalmol.2013.7963
Visual field loss due to retinal damage is considered irreversible, and methods are needed to achieve vision restoration. Behavioral vision training, shown to improve visual fields in hemianopia and optic nerve damage, might comprise such a method.
To determine if behavioral activation of areas of residual vision using daily 1-hour vision restoration training for glaucoma for 3 months improves detection accuracy compared with placebo.
Design and Setting
Prospective, double-blind, randomized, placebo-controlled clinical trial in an ambulatory care and home training setting.
Volunteer sample of patients with glaucoma (mean age, 61.7 years; age range, 39-79 years) with stable visual fields and well-controlled intraocular pressure. After randomization, 4 patients withdrew from the trial because of mild headaches (n = 2) or lack of time to complete the schedule (n = 2).
Computer-based vision restoration training for glaucoma (n = 15) or visual discrimination placebo training in the intact visual field (n = 15).
Main Outcomes and Measures
The primary end point was change in detection accuracy in high-resolution perimetry. Secondary end points were 30° white-on-white and 30° blue-on-yellow near-threshold perimetry, as well as reaction time, eye movements, and vision-related and health-related quality of life.
Vision restoration training for glaucoma led to significant detection accuracy gains in high-resolution perimetry (P = .007), which were not found with white-on-white or blue-on-yellow perimetry. Furthermore, the pre-post differences after vision restoration training for glaucoma were greater compared with placebo in all perimetry tests (P = .02 for high-resolution perimetry, P = .04 for white on white, and P = .04 for blue on yellow), and these results were independent of eye movements. Vision restoration training for glaucoma (but not placebo) also led to faster reaction time (P = .009). Vision-related quality of life was unaffected, but the health-related quality-of-life mental health domain increased in both groups.
Conclusions and Relevance
Visual field defects caused by glaucoma can be improved by repetitively activating residual vision through training the visual field borders and areas of residual vision, thereby increasing their detection sensitivity. Our randomized clinical trial revealed evidence that visual field loss is in part reversible by behavioral, computer-based, online controlled vision training, comprising a new rehabilitation treatment option in glaucoma. Neuroplasticity of the visual cortex or higher cortical areas is the proposed mechanism of action.
clinicaltrials.gov Identifier: NCT01799707.
Glaucoma is a progressive optic neuropathy caused by a pathologic optic nerve head with slow degeneration of retinal ganglion cells.1 The resulting vision loss is considered irreversible, with no chance of recovery. Because of neuroplasticity,2- 6 or the ability of the brain to adapt to change, visual system structures are modifiable even in adulthood and old age by the following means: (1) in adult visual cortex, receptive fields can change in size and location after retinal injury7- 14; (2) training eccentric (peripheral) viewing in macular degeneration activates cortical regions normally representing central visual areas15,16; (3) visual training in hemianopia activates visual cortex at the lesion border17,18; (4) training-induced vision restoration is greater in cases in which the cortical regions are not directly injured, as in after optic nerve damage19; and (5) noninvasive transorbital alternating current stimulation can improve visual fields in optic neuropathy, with corresponding electrophysiological changes.6,20- 22
Evidence exists that visual system plasticity is behaviorally meaningful: regular practice in normal perceptual learning23,24 can improve visual acuity and contrast sensitivity.25 Furthermore, vision loss after visual system damage can spontaneously recover in animals26,27 and patients,28 and stimulating visual field borders in patients with cortical or optic nerve lesions improves detection accuracy and enlarges visual fields,19,29- 36 which are stable and improve activities of daily living.29,37,38
Evidence also indicates that areas of residual vision (ARVs) contain fibers spared by the lesion that can activate the plastic cortex6 as suggested by the following observations: (1) 10% to 20% retinal ganglion cell survival in animals with optic nerve damage is sufficient to achieve recovery of visually guided behavior up to 80% detection in animals,26,27,39 (2) retinal ganglion cells spared by the damage compensate by increasing cell soma size40 and becoming hyperactive,41 (3) training-induced vision improvement in patients following stroke is directly related to ARV size,42 and (4) vision restoration is a function of residual activity locally and in the immediate surrounding area.6,43- 45 A pilot trial also showed that training improved detection accuracy in 4 of 5 patients with glaucoma,46 but definitive evidence of vision restoration training efficacy in glaucoma is still lacking.
A prospective, double-blind, randomized, placebo-controlled clinical trial was approved by the local ethics committee and was carried out from March 2004 to August 2007. After informed consent, patients were randomly assigned by lot in numbered containers to a vision restoration training for glaucoma (gVRT) group (n = 15) or to a placebo group (n = 15). Participants and outcome assessors were masked with respect to group identities.
Patient recruitment identified 93 potential patients with glaucoma, 43 of whom were enrolled in the study. Nine patients dropped out during the baseline assessments, and 4 patients dropped out after randomization (Figure 1).
Inclusion criteria for study entry were (1) visual field defects due to glaucoma, (2) a stable visual field defect inside 30° eccentricity in at least 1 eye at 2 consecutive examinations during the past 12 months, (3) well-controlled intraocular pressure, and (4) age between 25 and 80 years. Exclusion criteria were a history of the following: (1) medical condition precluding scheduled study visits, (2) chronic degenerative or inflammatory disease affecting the visual field (eg, multiple sclerosis or tumor), (3) trauma or any nonglaucoma ocular disease (eg, diabetic retinopathy, age-related macular degeneration, macular detachment, or vascular occlusion), (4) severe cognitive or motor impairment, (5) insufficient fixation ability, (6) photosensitivity, (7) intraocular surgery or laser treatment during the past 12 months, or (8) scheduled intraocular surgery.
The study sample comprised 14 men and 16 women (mean [SD] age, 61.7 [10.1] years; age range, 39-79 years) with primary chronic open-angle glaucoma (n = 20), normal-tension glaucoma (n = 5), or other types of chronic glaucoma (1 angle-closure glaucoma and 4 secondary glaucoma). Groups did not statistically differ with respect to age, time of first glaucoma diagnosis, or intraocular pressure and visual acuity (patient characteristics are summarized in eTable 1 in the Supplement).
Both eyes were affected in 26 patients, and 1 eye was affected in 4 patients. In 10 patients each, vision loss classification of visual field defects47 was mild (ie, monocular loss of less than half of the visual field), moderate (ie, monocular loss of greater than half of the visual field or binocular loss of less than half of the visual field), or severe (binocular loss of greater than half of the visual field in either eye). In binocular loss, the field defect was always asymmetric. Herein, the more affected eye that had a significantly greater percentage of absolute defects in both conventional white-on-white (W/W) and blue-on-yellow (B/Y) perimetry received training. The mean (SE) values for W/W were 34.49 (19.36) for the trained eye and 23.06 (27.68) for the untrained eye (z = −2.32, P = .02). The mean (SE) values for B/Y were 31.42 (21.89) for the trained eye and 19.70 (25.87) for the untrained eye (z = −2.20, P = .02).
During baseline examinations and after training, we assessed different vision tests and used vision-related quality-of-life (25-Item National Eye Institute Visual Function Questionnaire [NEI-VFQ-25]) and health-related quality-of-life (36-Item Short Form Health Survey) questionnaires. The primary outcome criterion was detection accuracy in high-resolution perimetry (HRP) visual field tests. Visual field diagnostic procedures were performed on 2 consecutive days and under the supervision of 2 technicians who were masked with respect to group identities. To minimize the effect of fatigue, patients were given sufficient time to rest between visual field examinations. To reduce variance, the order of the tests was kept constant among all patients in baseline and posttraining assessments.
Visual fields were measured monocularly with automated perimetry (Twinfield model 56900; Oculus) and HRP. Automated perimetry was performed by an automatized testing strategy according to the manufacturer’s manual. For 30° W/W and 30° B/Y perimetry, Goldmann III (but not V) was selected, and detection values rather than sensitivity values were used as outcomes to be able to compare them directly with HRP detection accuracy results. Luminance class determinations were made to identify absolute defects (no response even to the highest luminance intensity), relative defects (elevated luminance intensity), or intact spots (initial luminance intensity).
To assess the visual field, computer-based HRP was used, which is a valid and reliable testing method with suprathreshold stimuli (up to ±27° horizontal and ±20° vertical eccentricity).48 Responses within 150 to 1000 milliseconds were considered correct detections (hits) and outside this time window were considered delayed responses. To determine a stable baseline before training, 5 HRP repeated visual field charts were obtained.48 Visual field tests were considered valid if they had less than 10% fixation errors. Visual field was measured without refractive correction because acuity was of minor concern given the large stimulus sizes we used and because wearing spectacles during perimetry interfered with the head holder and eye tracking.
Because patients with glaucoma typically experience some variability in their visual fields,49,50 we measured natural visual field variability for 2 days before training. Variability was then analyzed (1) as repeated measures and (2) as range variability (ie, the absolute difference between the lowest and highest detection rates51) by first determining the range variability for each patient and then calculating group differences.
Vision-related quality of life was assessed with a German version52 of the NEI-VFQ-25 survey.53 We also administered the 36-Item Short Form Health Survey54 before and after the training, which is a test that is not disease specific but can probe generic physical and mental health status (health-related quality of life) in patients with glaucoma.55,56 After all other tests were completed, satisfaction with the training was assessed using a Likert-type scale ranging from 1 to 10. Only then were patients informed of the group to which they belonged. Patients in the placebo group were subsequently offered gVRT.
One may argue that detection accuracy gains are an artifact of eye movements. Therefore, we measured eye movements monocularly during HRP to determine the stability of fixation before and after training using an eye tracker (Tobii 1750; Tobii Technology AB) with an eye gaze frame rate of 50 Hz, 0.5° accuracy, and 1° drift rate. Data were not considered for further analysis if the eye-tracking period was shorter than 75% of the HRP testing time. Therefore, only 26 patients were included in the pre-post eye tracker analysis. Eye movements and stability of fixation were quantified using (1) the number of horizontal saccades per minute, (2) the mean gaze position, and (3) gaze position variability (standard deviation).
Training was performed 6 days a week for 3 months; the duration was 30 minutes twice daily. After completing the baseline assessments, patients performed the training at home on a commercially available personal PC as previously described19 with adaptive parameter adjustments online. Seventeen patients trained with the right eye, and 13 patients trained with the left eye.
The gVRT group performed a variant of the classic vision restoration training19,31 developed in our laboratory whereby visual stimuli are presented in ARVs. The training consisted of luminance increment stimuli similar to perimetry, and the task was simple detection (pressing a key whenever a target stimulus was detected). The training parameters were adaptive using online adjustments of levels of difficulty on a monthly basis (a mean [SD] of 500  stimuli were delivered twice daily, 80% were presented in ARVs, and 20% were administered in the seeing visual field) (Figure 2).
The placebo group trained with a discrimination paradigm using a line segment (bar), which was always presented within the central ±5° visual field in 1 of 4 possible random orientations (horizontal, vertical, oblique to the right, or oblique to the left). If the patient had visual field defects in this central area, 80% of the stimuli were presented in the intact part of the training region. The task was to identify line orientations by pressing 1 of 4 assigned keyboard buttons as fast as possible. The number of stimulus presentations was a mean (SD) of 350 (25) sessions twice daily (700 per day), so that training time was the same as in the gVRT group, and this training was also adjusted monthly.
The primary outcome measure was detection accuracy (ie, the number of detected stimuli in the HRP visual field tests). Secondary outcome measures were quality-of-life scores, reaction time, eye movements, and reliability parameters in the perimetric measurements (fixation rate and delayed responses). For comparison of test results before and after training, Wilcoxon z test for paired samples was used. Comparisons of training-induced changes between gVRT and placebo were performed with the Mann-Whitney test. To determine the factors that were associated with the changed stimulus detection rate after training, Spearman rank correlation coefficients were calculated. Because the number of delayed responses significantly correlated with the detection performance changes, analysis of covariance was performed. Spearman rank correlation was used to examine the association between the size of binocular visual field loss and the NEI-VFQ-25 subscale scores. The binocular visual field was simulated using the best location model for the summation of the monocular visual fields.57,58 Herein, the higher sensitivity from each of the 2 corresponding visual field locations in the perimetry is determined to give an estimate of the sensitivity of overall vision at that visual field position as if viewing binocularly. The multiple visual field test comparisons were assessed by Friedman test. Data are expressed as means (SDs), and P < .05 was considered significant using statistical software (SPSS, version 13.0; SPSS Inc).
We conducted a prospective, double-blind, randomized, placebo-controlled clinical trial. We show that computer-controlled vision training significantly improves visual detection accuracy and temporal processing in glaucoma.
At baseline, the stimulus detections across the 5 repeated HRP examinations fluctuated significantly in the gVRT group (χ2 = 10.98, P = .03), whereas in the placebo group they did not. Expectedly, the detection rate improved during the course of the 5 baseline sessions in both groups. Using the range-based calculation, the mean variability was 8.55% (5.39%) in the gVRT group and 9.54% (4.54%) in the placebo group. The range-based variability was not significantly different between the groups.
The Table summarizes the mean detection performance of the trained eye before vs after training for both groups as measured by HRP and W/W and B/Y perimetry. Detection performance was calculated by computing the percentage detected stimuli in absolute terms. The mean HRP detection rate in the gVRT group was 37.45 (21.85) at baseline and increased significantly to 44.17 (24.39) after gVRT (z = −2.69, P = .007). The extent of improvement varied considerably. Whereas 5 patients (33.3%) did not show visual field improvement (detection rate, <3%), 4 patients (26.7%) showed moderate improvement (detection rate, 3%-10%), and 6 patients (40.0%) showed large improvement (detection rate, >10%). The mean placebo group values were 38.69 (27.03) detections at baseline, which is comparable to the gVRT group, but at final outcome the value was almost unchanged at 39.84 (29.15) (z = −0.71, P = .48). In contrast, detection accuracy analyses of the trained eye in conventional W/W and B/Y perimetry showed a nonsignificant increase in the gVRT group and a nonsignificant loss in the placebo group. As a control condition, the detection changes in the nontrained eye were measured by conventional perimetry, but no significant changes were found in either group or among the different perimetry methods. In the gVRT group, the mean values were 76.13 (23.18) (pre) and 78.65 (26.95) (post) for W/W (P = .63) and 80.03 (25.77) (pre) and 80.70 (25.93) (post) for B/Y (P = .71). In the placebo group, the mean values were 83.73 (24.41) (pre) and 84.76 (23.95) (post) for W/W (P = .58) and 85.98 (24.41) (pre) and 86.77 (22.11) (post) for B/Y (P = .59).
Because visual field changes may be considered real only if they exceed the natural variability,59 we analyzed visual field variability and found that 7 of 15 patients in the gVRT group achieved a detection performance improvement above the mean variability of 8.55% of the group’s range-based mean variability. Only 2 of 15 patients in the placebo group showed increased detection ability above their group variability (>9.54%).
Detection accuracy improvement in the trained eye was located mainly in ARVs (Figure 3). Unlike our experience in hemianopia, the HRP detection accuracy gains were not significantly greater in patients with larger ARVs (r = 0.17, P = .49). However, the detection accuracy gains were also not greater in cases with greater visual field variability at baseline (r = 0.49, P = .32).
The HRP reaction time was significantly faster after training in the gVRT group than in the placebo group. In the gVRT group, the mean values were 579.73 (72.97) milliseconds (pre) and 541.67 (74.78) milliseconds (post) (z = −2.61, P = .009). In the placebo group, the mean values were 558.53 (71.28) milliseconds (pre) and 559.53 (61.1) milliseconds (post) (z = 0.00, P > 99).
Nieuwenhuis et al60 emphasize that statistical evidence of between-group comparison is superior to within-group comparison. Our between-group comparison of the pre-post detection changes showed significantly greater improvement in all 3 visual field test procedures in the gVRT group compared with the placebo group: HRP (z = −2.03, P = .02), W/W (z = −1.68, P = .04), and B/Y (z = −1.74, P = .04) (Figure 4). Furthermore, the gVRT had significantly faster reaction time in HRP (z = −2.32, P = .01).
Neither the mean gaze position nor the mean gaze position variability (standard deviation) changed significantly after training compared with baseline in either group. In fact, the groups did not differ in fixation test results, and no statistical differences were observed in the blind spot positions before vs after treatment (eTable 2 in the Supplement)
In the gVRT group, the mean rate of delayed responses increased in all 3 visual field test methods after training, but this improvement was significant only in B/Y perimetry: 99.07 (2.46) (pre) and 97.33 (3.51) (post) (z = −2.03, P = .04). In the placebo group, the mean percentage of delayed responses decreased slightly in HRP. Nevertheless, no significant differences were observed between the 2 groups regarding the delayed response rate changes in any of the visual field tests. In the gVRT group, we found significant correlations between the increased number of delayed responses and HRP detection accuracy (r = 0.68, P = .005) and W/W perimetry (r = −0.53, P = .04) but not B/Y perimetry (r = −0.20, P = .11). In the placebo group, no such correlations were found. To rule out that delayed responses explain the main effect, we used them as covariates. In this analysis of covariance, a significant main effect for training was seen (F1,13 = 14.01, P = .002) as well as a significant interaction between the detection accuracy improvement and delayed responses (F1,13 = 7.06, P = .02). Therefore, the training effect was greater than the interaction effect, showing that the detection accuracy improvement cannot be explained by delayed responses alone.
The baseline NEI-VFQ-25 scores were moderate or high (>65) on all dimensions except general health. Therefore, our patients had few everyday vision deficits at baseline, and none of the subscales showed training effects. The baseline 36-Item Short Form Health Survey showed deficits only in the physical functioning domain, which correlated with severity of binocular visual field defect (r = 0.41, P = .02); significant mean improvement was only seen in the gVRT group in the mental health subscale: 69.33 (18.98) (pre) and 77.60 (12.98) (post) (z = −2.30, P = .02). Both groups reported a high mean level of satisfaction with the training: 6.93 (2.40) for the gVRT group and 7.53 (2.87) for the placebo group (z = −0.95, P = .44). Therefore, our masking procedure was validated. Only the gVRT group reported noticing mean subjective visual improvement: 4.47 (1.95) for the gVRT group and 2.80 (1.93) for the placebo group (z = −2.35, P = .02). However, this finding did not correlate with improved detection accuracy (r = 0.12, P = .37).
Daily training with gVRT significantly improved vision-related performance in patients with glaucoma without affecting eye movements. Compared with placebo patients, gVRT-treated patients had detection accuracy improvement and reaction time gains, which is evidence of sensitivity improvement in residual vision. Specifically, between-group comparisons of pre-post changes showed that gVRT compared with placebo led to significantly improved detection in HRP and W/W and B/Y perimetry, which was confirmed by the patients’ subjective reports that training improved their vision. As in a previous pilot study46 by our group and in other studies19,22,32,35,36,38,46,61 of vision restoration training, approximately one-third of the patients were nonresponders, one-third achieved moderate improvement, and one-third achieved substantial improvement.
The gVRT group also had significantly faster reaction time. This observation confirms the conclusion of improved temporal processing as noted in earlier studies38,62 and may result from increased synchronization of neuronal firing.38 That genuine sensitivity can improve is supported by the observation that reaction time and suprathreshold contrast are associated.63
It is well known that visual fields often fluctuate in glaucoma,49,50 which could mask or mimic true visual field changes. However, the changes we observed cannot be explained by natural fluctuations: if one counts only the detection accuracy improvement greater than baseline fluctuations as valid, then 7 patients in the gVRT group and 2 patients in the placebo group showed such improvement. Furthermore, visual field improvement did not correlate with visual field variability at baseline. Therefore, visual field fluctuation cannot explain the detection accuracy improvement after gVRT.
Visual improvement also cannot be explained by eye movement artifacts because eye tracker recordings showed neither signs of shifted gaze positions nor more frequent saccades. Fixation actually improved with vision training64 (indicating more stable eye position), and when fixation inaccuracies occur, they do not correlate with detection improvement, confirming prior observations.6,32,64,65
Nevertheless, we found that significant visual field improvement correlated with increases in delayed responses as noted earlier.32,37 While such delayed reactions could comprise random responses, we believe they are true hits but that patients respond to the stimulus presentations too slowly. In fact, reaction time impairment is seen in patients with optic nerve damage not only near or inside the scotoma but also in the presumably intact sectors of the visual field,66 a phenomenon termed sightblindness.67 Therefore, more delayed reactions may be explained by recovering vision that is still too slow to reach the minimum reaction time criterion (<1000 milliseconds). However, it is also conceivable that patients responded to pseudohallucinations, which are perceptions of light flashes that start emerging during or after spontaneous or training-induced visual field recovery68,69 without stimulus presentations. In any event, increased numbers of delayed reactions are interpreted by us as signs of vision recovery, which is why they correlate with detection improvement.
A further goal of the present study was to assess the subjective outcome using vision-related and health-related quality-of-life questionnaires, but training did not lead to robust changes. Only mental health was found to have improved in both training groups, which may be caused by nonspecific training effects such as attention, alertness, or expectation. The general lack of subjective changes as probed by the NEI-VFQ-25 may be explained by a ceiling effect: at baseline, patients already had few difficulties in everyday life because most had asymmetrical field loss, which often goes unnoticed during binocular viewing, thus leaving little room for improvement. Nevertheless, gVRT patients in the posttraining interview reported significantly greater subjective improvement, and placebo patients did not.
Possible mechanisms of action for gVRT are as follows: (1) plasticity in the retina6; (2) changes at higher processing levels in the geniculate70 or the primary visual cortex, such as receptive field changes12- 14; (3) improved visual attention33; (4) normal perceptual learning6; or (5) a combination thereof. The fact that detection accuracy gains in glaucoma were not predominantly found in ARVs (as was seen in hemianopia6) suggests that surviving retinal ganglion cells deep in the damaged zones might contribute to vision restoration or changes in intact visual field sectors.66,67 To clearly delineate the mechanisms of vision restoration requires further study using an electroencephalogram or magnetic resonance imaging recordings.
In summary, our study confirms that visual system plasticity is maintained into adulthood and even old age despite widespread visual system degeneration.6,71,72 Our findings are compatible with the theory of activating residual vision6 following retina or brain lesions. Vision loss in glaucoma may not be permanent: it is partially reversible through brain plasticity and cortical reorganization,6,15,16 and this finding is clinically useful. Whatever the mechanism, we can now be more optimistic that the course of vision loss in glaucoma is not simply pointed downhill but has considerable uphill potential. This new understanding justifies additional research and provides evidence to support routine use of gVRT in the clinical setting. Our results show that vision restoration and rehabilitation are possible by means of activating residual vision through training-induced brain plasticity.
Submitted for Publication: March 14, 2013; final revision received October 10, 2013; accepted November 4, 2013.
Corresponding Author: Bernhard A. Sabel, PhD, Institute of Medical Psychology, Medical Faculty, Otto-von-Guericke University of Magdeburg, Leipziger Strasse 44, 39120 Magdeburg, Germany (firstname.lastname@example.org).
Published Online: February 6, 2014. doi:10.1001/jamaophthalmol.2013.7963.
Author Contributions: Drs Sabel and Gudlin had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: All authors.
Acquisition of data: Gudlin.
Analysis and interpretation of data: All authors.
Drafting of the manuscript: All authors.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Gudlin.
Obtained funding: Sabel.
Administrative, technical, and material support: Sabel.
Study supervision: Sabel.
Conflict of Interest Disclosures: None reported.
Funding/Support: This study was funded by NovaVision Inc (Dr Sabel), by Braingames Online GmbH (Dr Sabel), and by a predoctoral fellowship under the LOM program of the University of Magdeburg (Dr Gudlin).
Role of the Sponsor: The funding organizations had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Additional Contributions: Nicole Mäter and Sandra Heinrich assisted in examining the patients.