Downloaded from http://bjo.bmj.com/ on March 26, 2015 - Published by group.bmj.com
Clinical science
The Spaeth/Richman contrast sensitivity test (SPARCS): design, reproducibility and ability to identify patients with glaucoma Jesse Richman, Camila Zangalli, Lan Lu, Sheryl S Wizov, Eric Spaeth, George L Spaeth Glaucoma Service, Wills Eye Hospital, Jefferson Medical College, Philadelphia, Pennsylvania, USA Correspondence to Dr Jesse Richman, Glaucoma Service, Wills Eye Hospital, Jefferson Medical College, 840 Walnut Street, Suite 1110, Philadelphia, PA 19107, USA; [email protected] Received 12 March 2014 Revised 19 June 2014 Accepted 28 June 2014 Published Online First 22 July 2014
ABSTRACT Aims (1) To determine the ability of a novel, internetbased contrast sensitivity test titled the Spaeth/Richman Contrast Sensitivity Test (SPARCS) to identify patients with glaucoma. (2) To determine the test-retest reliability of SPARCS. Methods A prospective, cross-sectional study of patients with glaucoma and controls was performed. Subjects were assessed by SPARCS and the Pelli-Robson chart. Reliability of each test was assessed by the intraclass correlation coefficient and the coefficient of repeatability. Sensitivity and specificity for identifying glaucoma was also evaluated. Results The intraclass correlation coefficient for SPARCS was 0.97 and 0.98 for Pelli-Robson. The coefficient of repeatability for SPARCS was ±6.7% and ±6.4% for Pelli-Robson. SPARCS identified patients with glaucoma with 79% sensitivity and 93% specificity. Conclusions SPARCS has high test-retest reliability. It is easily accessible via the internet and identifies patients with glaucoma well. Trial Registration: NCT01300949.
INTRODUCTION
To cite: Richman J, Zangalli C, Lu L, et al. Br J Ophthalmol 2015;99: 16–20. 16
The ability to discern contrast strongly affects vision-related functional ability.1–5 Contrast sensitivity plays a role in visual acuity, visual field, dark adaptation, motion recognition and pattern recognition. Patients with glaucoma have diminished central contrast sensitivity.6–8 Prior studies have attempted to use central contrast sensitivity to diagnose glaucoma before the emergence of visual field defects. These studies did not demonstrate a high enough sensitivity and specificity for contrast sensitivity to be used as a screening test for diagnosing or detecting early glaucoma.8–12 The contrast sensitivity tests used in those studies often were inadequately evaluated and the tests had poor reliability, making it difficult to distinguish normal vision from glaucoma. It is also possible that, even with reliable and sensitive testing methods, central contrast sensitivity may not be an effective screening test for glaucoma. Classic glaucomatous visual field loss is usually detected first in the periphery of the field, rather than centrally. Attempts have been made to assess peripheral contrast sensitivity, which has been shown to be diminished in patients with glaucoma.13–17 However, studies by Tochel et al13 and Lundh have found that the areas of peripheral contrast loss did not correlate well with the location of white-on-white visual field loss, indicating that they
may be evaluating different aspects of visual function, or that areas of peripheral vision may be impaired before the loss can be detected using standard perimetry.16 We have designed a contrast sensitivity test titled the Spaeth/Richman Contrast Sensitivity Test (SPARCS), which evaluates patients’ ability to detect contrast, with their central vision and peripheral vision. SPARCS is an internet-based test that features multiple answer choices and a bracketing technique to determine contrast threshold. The test does not use words or letters and eliminates the problem of chart fading. Here we detail our testing design as well as scores in controls, glaucoma suspects and patients with glaucoma. This study also addresses the reliability of SPARCS and compares it with Pelli-Robson, which has been shown to be a highly reliable test of central contrast sensitivity in patients with glaucoma.18 19
METHODS The Institutional Review Board of Wills Eye Hospital approved the study protocol, which was in accordance with the Declaration of Helsinki and the Health Insurance Portability and Accountability Act. Eligible patients signed an informed consent document before enrolling in the study. Subsequently, we tested 136 eyes of 91 patients with glaucoma or features suspicious for glaucoma from Wills Eye Hospital between January 2010 and August 2011. All medical records of patients scheduled for examinations were reviewed consecutively. Inclusion criteria included age older than 18 years; ability to provide fully informed consent; and being either a patient with glaucoma or glaucoma suspect. Glaucoma was considered present when the patient had characteristic optic nerve damage and visual field loss.20 No case was considered to be glaucoma if there was any other reasonable cause for the optic disc or visual field abnormality. Glaucoma suspects were defined as having either an intraocular pressure above 21 with a normal optic disc and visual field or a suspicious appearing optic disc without definitive glaucomatous field changes. Seventy healthy controls with no ocular disease were also recruited for the study. Specifically, 10 controls for each age group (20–30 years, 30–40 years, 40–50 years, 50– 60 years, 60–70 years, 70–80 years, over age 80) were recruited. Patients with glaucoma were excluded if they had incisional eye surgery within the past 3 months or laser therapy within the previous month. In order to avoid multiple aetiologies of decreased contrast sensitivity, patients with a cataract of grade 2 or higher
Richman J, et al. Br J Ophthalmol 2015;99:16–20. doi:10.1136/bjophthalmol-2014-305223
Downloaded from http://bjo.bmj.com/ on March 26, 2015 - Published by group.bmj.com
Clinical science according to the Lens Opacities Classification System III21 or any cause for visual reduction other than glaucoma were also excluded. For example, patients with other ocular or neurological conditions that may influence vision were not able to participate in the study. To ensure that patients with a full range of glaucomatous damage were included, selection was based partially on the amount of optic disc damage.22 Patients were also recruited with the goal of enrolling 15 patients per decade of age. SPARCS is a novel, internet-based test designed to determine the contrast threshold of patients’ central vision and peripheral vision. It is performed on a standard computer with internet access. It is designed to be used on a monitor set to 1024×768 resolution, 256 grey levels and a size of at least 22 cm width and 26.5 cm height. SPARCS can be accessed via https://www. sparcscontrastcenter.com where each patient gets a unique identification number. The website provides instructions on how to take the test. Patients are seated 50 cm from the computer monitor. At this testing distance, the test occupies 30° of vision horizontally and 23.5° of vision vertically. The central test area subtends 5° horizontally and 3.5° vertically. Patients are instructed to fixate on the central area and identify which of the areas appears different. When patients are ready, they click on the central area. Vertical square wave gratings with a spatial frequency of 0.4 cycles per degree (figure 1) appear for 0.3 s in one of the five tested areas. Patients then temporarily break fixation to select the area. Subsequently, patients fixate again on the central area and click it to activate the programme to show the next image. The area with the gratings appears at random. Correct and incorrect responses are recorded by SPARCS until the contrast threshold is determined in each area. The contrast threshold is determined using a staircase strategy with reversals. Initial correct responses advance four levels until an incorrect response is made. After the incorrect response, the contrast level presented is two levels easier. Thereafter, the algorithm advances or regresses one level at a time until two incorrect responses are made at a specific level, which establishes the threshold. If a patient stops trying to guess the correct area and simply clicks the same location again and again, the test terminates and explains to the patient to attempt to choose the location the image appeared. The range of contrast tested is from 100% to 0.45% (log contrast sensitivity 0.00 to 2.35) and decreases by approximately 0.15 log units between levels. The contrast value is calculated by Weber contrast. The central area and four peripheral areas
each receive separate scores. Each log-based score is then scaled out of 20 by dividing by 2.35 and multiplying by 20. A total SPARCS score is summated from each of the five areas, making 100 the perfect summed score from all five areas. Patient testing was conducted at Wills Eye Hospital’s Glaucoma Research Center. Patients were tested with SPARCS and the Pelli-Robson contrast sensitivity chart. Monocular testing was performed on each eye twice, with different observers each time. The order of testing was randomised to SPARCS, Pelli-Robson, Pelli-Robson, SPARCS; or Pelli-Robson, SPARCS, SPARCS, Pelli-Robson. A study coordinator explained and administered each test. Fluorescent lighting was used in a room without daylight to minimise glare. The Pelli-Robson contrast sensitivity chart was tested at 1 m. The chart has large Sloan letters that occupy approximately one cycle per degree of vision. The letters are arranged in triplets, which decrease in contrast by 0.15 log units for each triplet. The contrast tested ranges from 100% to 0.56% (log contrast sensitivity 0.00–2.25). Patients were given credit for a contrast level if they answer two of the three letters in a triplet correctly.23 Visual field testing was performed on an Octopus visual field (Haag-Streit) with appropriate refractive correction. The G-Top program was used for monocular visual field testing. If patients had been tested with this method within the past 6 months, field testing was not repeated. The optic nerve was evaluated using the Disc Damage Likelihood Scale (staging from 1 to 10).22 Monocular visual acuity was measured using a rearilluminated Early Treatment Diabetic Retinopathy Study distance chart, second edition (Precision Vision, LaSalle, Illinois). Visual acuity was scored as the total number of letters identified correctly and converted to logMAR (log10 minimum angle resolvable). Test-retest repeatability of SPARCS and Pelli-Robson was assessed by the intraclass correlation coefficient (ICC) and the coefficient of repeatability (COR). The ICC is a measure of test-retest repeatability and is sensitive to differences between subjects. ICC values closer to 1 indicate better test-retest reliability. As described by Bland and Altman, the COR is calculated as 1.96 multiplied by the SD of the difference between test and retest scores.24 A COR value closer to 0 indicates better test-retest reliability. The distribution and relationships of all the variables in the study were analysed in a correlation matrix and by scatterplot. The Mann-Whitney test was used to determine differences among dichotomous variables, such as gender and laterality. The Kruskal-Wallis test was used to determine if there were differences owing to age, race or diagnosis. Measurements with p
Clinical science
The Spaeth/Richman contrast sensitivity test (SPARCS): design, reproducibility and ability to identify patients with glaucoma Jesse Richman, Camila Zangalli, Lan Lu, Sheryl S Wizov, Eric Spaeth, George L Spaeth Glaucoma Service, Wills Eye Hospital, Jefferson Medical College, Philadelphia, Pennsylvania, USA Correspondence to Dr Jesse Richman, Glaucoma Service, Wills Eye Hospital, Jefferson Medical College, 840 Walnut Street, Suite 1110, Philadelphia, PA 19107, USA; [email protected] Received 12 March 2014 Revised 19 June 2014 Accepted 28 June 2014 Published Online First 22 July 2014
ABSTRACT Aims (1) To determine the ability of a novel, internetbased contrast sensitivity test titled the Spaeth/Richman Contrast Sensitivity Test (SPARCS) to identify patients with glaucoma. (2) To determine the test-retest reliability of SPARCS. Methods A prospective, cross-sectional study of patients with glaucoma and controls was performed. Subjects were assessed by SPARCS and the Pelli-Robson chart. Reliability of each test was assessed by the intraclass correlation coefficient and the coefficient of repeatability. Sensitivity and specificity for identifying glaucoma was also evaluated. Results The intraclass correlation coefficient for SPARCS was 0.97 and 0.98 for Pelli-Robson. The coefficient of repeatability for SPARCS was ±6.7% and ±6.4% for Pelli-Robson. SPARCS identified patients with glaucoma with 79% sensitivity and 93% specificity. Conclusions SPARCS has high test-retest reliability. It is easily accessible via the internet and identifies patients with glaucoma well. Trial Registration: NCT01300949.
INTRODUCTION
To cite: Richman J, Zangalli C, Lu L, et al. Br J Ophthalmol 2015;99: 16–20. 16
The ability to discern contrast strongly affects vision-related functional ability.1–5 Contrast sensitivity plays a role in visual acuity, visual field, dark adaptation, motion recognition and pattern recognition. Patients with glaucoma have diminished central contrast sensitivity.6–8 Prior studies have attempted to use central contrast sensitivity to diagnose glaucoma before the emergence of visual field defects. These studies did not demonstrate a high enough sensitivity and specificity for contrast sensitivity to be used as a screening test for diagnosing or detecting early glaucoma.8–12 The contrast sensitivity tests used in those studies often were inadequately evaluated and the tests had poor reliability, making it difficult to distinguish normal vision from glaucoma. It is also possible that, even with reliable and sensitive testing methods, central contrast sensitivity may not be an effective screening test for glaucoma. Classic glaucomatous visual field loss is usually detected first in the periphery of the field, rather than centrally. Attempts have been made to assess peripheral contrast sensitivity, which has been shown to be diminished in patients with glaucoma.13–17 However, studies by Tochel et al13 and Lundh have found that the areas of peripheral contrast loss did not correlate well with the location of white-on-white visual field loss, indicating that they
may be evaluating different aspects of visual function, or that areas of peripheral vision may be impaired before the loss can be detected using standard perimetry.16 We have designed a contrast sensitivity test titled the Spaeth/Richman Contrast Sensitivity Test (SPARCS), which evaluates patients’ ability to detect contrast, with their central vision and peripheral vision. SPARCS is an internet-based test that features multiple answer choices and a bracketing technique to determine contrast threshold. The test does not use words or letters and eliminates the problem of chart fading. Here we detail our testing design as well as scores in controls, glaucoma suspects and patients with glaucoma. This study also addresses the reliability of SPARCS and compares it with Pelli-Robson, which has been shown to be a highly reliable test of central contrast sensitivity in patients with glaucoma.18 19
METHODS The Institutional Review Board of Wills Eye Hospital approved the study protocol, which was in accordance with the Declaration of Helsinki and the Health Insurance Portability and Accountability Act. Eligible patients signed an informed consent document before enrolling in the study. Subsequently, we tested 136 eyes of 91 patients with glaucoma or features suspicious for glaucoma from Wills Eye Hospital between January 2010 and August 2011. All medical records of patients scheduled for examinations were reviewed consecutively. Inclusion criteria included age older than 18 years; ability to provide fully informed consent; and being either a patient with glaucoma or glaucoma suspect. Glaucoma was considered present when the patient had characteristic optic nerve damage and visual field loss.20 No case was considered to be glaucoma if there was any other reasonable cause for the optic disc or visual field abnormality. Glaucoma suspects were defined as having either an intraocular pressure above 21 with a normal optic disc and visual field or a suspicious appearing optic disc without definitive glaucomatous field changes. Seventy healthy controls with no ocular disease were also recruited for the study. Specifically, 10 controls for each age group (20–30 years, 30–40 years, 40–50 years, 50– 60 years, 60–70 years, 70–80 years, over age 80) were recruited. Patients with glaucoma were excluded if they had incisional eye surgery within the past 3 months or laser therapy within the previous month. In order to avoid multiple aetiologies of decreased contrast sensitivity, patients with a cataract of grade 2 or higher
Richman J, et al. Br J Ophthalmol 2015;99:16–20. doi:10.1136/bjophthalmol-2014-305223
Downloaded from http://bjo.bmj.com/ on March 26, 2015 - Published by group.bmj.com
Clinical science according to the Lens Opacities Classification System III21 or any cause for visual reduction other than glaucoma were also excluded. For example, patients with other ocular or neurological conditions that may influence vision were not able to participate in the study. To ensure that patients with a full range of glaucomatous damage were included, selection was based partially on the amount of optic disc damage.22 Patients were also recruited with the goal of enrolling 15 patients per decade of age. SPARCS is a novel, internet-based test designed to determine the contrast threshold of patients’ central vision and peripheral vision. It is performed on a standard computer with internet access. It is designed to be used on a monitor set to 1024×768 resolution, 256 grey levels and a size of at least 22 cm width and 26.5 cm height. SPARCS can be accessed via https://www. sparcscontrastcenter.com where each patient gets a unique identification number. The website provides instructions on how to take the test. Patients are seated 50 cm from the computer monitor. At this testing distance, the test occupies 30° of vision horizontally and 23.5° of vision vertically. The central test area subtends 5° horizontally and 3.5° vertically. Patients are instructed to fixate on the central area and identify which of the areas appears different. When patients are ready, they click on the central area. Vertical square wave gratings with a spatial frequency of 0.4 cycles per degree (figure 1) appear for 0.3 s in one of the five tested areas. Patients then temporarily break fixation to select the area. Subsequently, patients fixate again on the central area and click it to activate the programme to show the next image. The area with the gratings appears at random. Correct and incorrect responses are recorded by SPARCS until the contrast threshold is determined in each area. The contrast threshold is determined using a staircase strategy with reversals. Initial correct responses advance four levels until an incorrect response is made. After the incorrect response, the contrast level presented is two levels easier. Thereafter, the algorithm advances or regresses one level at a time until two incorrect responses are made at a specific level, which establishes the threshold. If a patient stops trying to guess the correct area and simply clicks the same location again and again, the test terminates and explains to the patient to attempt to choose the location the image appeared. The range of contrast tested is from 100% to 0.45% (log contrast sensitivity 0.00 to 2.35) and decreases by approximately 0.15 log units between levels. The contrast value is calculated by Weber contrast. The central area and four peripheral areas
each receive separate scores. Each log-based score is then scaled out of 20 by dividing by 2.35 and multiplying by 20. A total SPARCS score is summated from each of the five areas, making 100 the perfect summed score from all five areas. Patient testing was conducted at Wills Eye Hospital’s Glaucoma Research Center. Patients were tested with SPARCS and the Pelli-Robson contrast sensitivity chart. Monocular testing was performed on each eye twice, with different observers each time. The order of testing was randomised to SPARCS, Pelli-Robson, Pelli-Robson, SPARCS; or Pelli-Robson, SPARCS, SPARCS, Pelli-Robson. A study coordinator explained and administered each test. Fluorescent lighting was used in a room without daylight to minimise glare. The Pelli-Robson contrast sensitivity chart was tested at 1 m. The chart has large Sloan letters that occupy approximately one cycle per degree of vision. The letters are arranged in triplets, which decrease in contrast by 0.15 log units for each triplet. The contrast tested ranges from 100% to 0.56% (log contrast sensitivity 0.00–2.25). Patients were given credit for a contrast level if they answer two of the three letters in a triplet correctly.23 Visual field testing was performed on an Octopus visual field (Haag-Streit) with appropriate refractive correction. The G-Top program was used for monocular visual field testing. If patients had been tested with this method within the past 6 months, field testing was not repeated. The optic nerve was evaluated using the Disc Damage Likelihood Scale (staging from 1 to 10).22 Monocular visual acuity was measured using a rearilluminated Early Treatment Diabetic Retinopathy Study distance chart, second edition (Precision Vision, LaSalle, Illinois). Visual acuity was scored as the total number of letters identified correctly and converted to logMAR (log10 minimum angle resolvable). Test-retest repeatability of SPARCS and Pelli-Robson was assessed by the intraclass correlation coefficient (ICC) and the coefficient of repeatability (COR). The ICC is a measure of test-retest repeatability and is sensitive to differences between subjects. ICC values closer to 1 indicate better test-retest reliability. As described by Bland and Altman, the COR is calculated as 1.96 multiplied by the SD of the difference between test and retest scores.24 A COR value closer to 0 indicates better test-retest reliability. The distribution and relationships of all the variables in the study were analysed in a correlation matrix and by scatterplot. The Mann-Whitney test was used to determine differences among dichotomous variables, such as gender and laterality. The Kruskal-Wallis test was used to determine if there were differences owing to age, race or diagnosis. Measurements with p
