- Email: [email protected]
A new perimeter using the preferential looking response to assess peripheral visual fields in young and developmentally delayed children
A new perimeter using the preferential looking response to assess peripheral visual fields in young and developmentally delayed children Louise E. Allen, MBBS, MD, FRCOphth,a,b Michael E. Slater, MEng,c Ruth V. Proffitt, BSc(Hons), DBO(d),b Elizabeth Quarton, BSc(Hons), DipHE(Child Nursing),b and Adar Pelah, PhDa,c PURPOSE
To compare the sensitivity, specificity, and interpretability of a newly developed semiautomated static perimeter based on the preferential looking response to the results of confrontation visual field testing in a group of young and/or developmentally delayed children with and without visual field deficit.
METHODS
The preferential looking perimeter (PLP) uses observation of the child’s natural eye movement response to an appearing target to determine the peripheral visual field. We compared preferential looking perimetry to confrontation testing in 74 children 3-10 years of age (mean, 6.6 years; median, 7 years), including 32 controls and 42 children with neurological and ocular disorders that could cause significant visual field deficit.
RESULTS
Using confrontation testing as the gold standard, the PLP was 100% sensitive and 100% specific (95% CI, 90%-100%), with excellent interobserver agreement. An interpretable result could be achieved in 15 (71%) of the 21 children in whom confrontation testing was unhelpful.
CONCLUSIONS
PLP is a useful new technique for assessing significant visual field loss in young or developmentally delayed children, with many advantages over confrontation testing. ( J AAPOS 2012;16:261-265)
A
variety of different neurological and ocular disorders can result in major visual field deficits, such as hemianopia and field constriction in young children. Recognition of a visual field deficit leading to early treatment can sometimes prevent subsequent loss of central vision, aid in the management of unpredictable tumors, such as optic glioma, and allow educational support tailored to the child’s visual deficit. In practice it can be difficult to assess accurately and interpret confrontation visual fields, particularly in children most in need of testing, namely, those with neurofibromatosis type 1, cerebral palsy, or developmental delay. Forced-choice preferential looking (FCPL) is an accepted method for assessing visual acuity in young children. We have developed a preferential looking perimeter (PLP) that uses this spontaneous, natural response to assess visual fields. We report a trial comparing the results of PLP to those from confrontation testing on Author affiliations: aUniversity of Cambridge, The Old Schools, Trinity Lane, Cambridge, United Kingdom; bCambridge University Hospitals NHS Foundation Trust, Cambridge, United Kingdom; cUniversity of York, Heslington, York, United Kingdom Submitted August 10, 2011. Revision accepted January 2, 2012. Correspondence: Louise E. Allen, MBBS, MD, FRCOphth, Consultant Paediatric Ophthalmologist, Addenbrooke’s Hospital, Box 41, Hills Road, Cambridge, CB2 0QQ, UK (email: [email protected]). Copyright Ó 2012 by the American Association for Pediatric Ophthalmology and Strabismus. 1091-8531/$36.00 doi:10.1016/j.jaapos.2012.01.006
Journal of AAPOS
a group of young children, both with and without possible visual field defects, whose age or behavioral and learning difficulties made them unlikely to be able to perform automated static perimetry.
Materials and Methods The Preferential Looking Perimeter The PLP consists of a 1.27 m domestic use flat plasma screen with a modified webcam mounted just below the central point of the screen. This screen size allows 80 of peripheral visual field in the horizontal meridian and 60 on the vertical meridian to be tested when viewed from a distance of 0.5 m. A computer drives the plasma screen and an independent monitor on which the live image from the webcam superimposed with the position of the peripheral targets is seen by the clinician. The central fixation video target subtends 7 and peripheral video targets subtend 5 from the child’s viewing position. The test can be run on each eye separately or with both eyes open to assess the binocular visual field if patching is not possible. A video of the device, named “Kidzeyez,” in operation may be viewed at the University of Cambridge Web site.1 Before the test begins, the child chooses a test stimulus from a range of age-appropriate cartoons and the clinician chooses the template of the peripheral locations to be tested or can design a new one. For the purpose of this trial, a six-point template of peripheral targets initially located at 30 eccentricity superior and inferior to the central target and 35 eccentricity in the
261
262
Allen et al
superotemporal, superonasal, inferotemporal, and inferonasal meridians was used (Figure 1). The test begins with a cartoon video playing on the center of the screen to elicit central fixation. The soundtrack of the cartoon is played on the built-in speakers and gives no clue to the site of the target. The clinician can see the child’s fixation (imaged by the webcam) and a superimposed square, showing the site of the next peripheral stimulus, on the monitor (Figure 1A). While the child is watching the central video, the clinician begins the test. The central video on the plasma screen freezes but the soundtrack continues. The cartoon action continues on the new peripheral site; as this target appears, the superimposed target square changes color on the clinician’s monitor (Figure 1B). The child preferentially looks toward the moving cartoon if it is seen. The examiner studies the child’s eye movements in response to the appearance of the target via the webcam, looking for an accurate saccade toward the target and selects “seen,” “not seen,” or “maybe” depending on the child’s response. The target is illuminated for 5 seconds and then disappears as the cartoon story returns to the central target. Once the child has refixated on the central target, the next stimulus can be enabled. Each template stimulus location is tested in random sequence until there is a consistent response. The 75% “not seen” responses result in the stimulus being shown 5 closer to center later in the test. The 75% response consistency was chosen for simplicity and to minimize the test time (targets only required three consistent responses of four showings to determine the overall response). A normal test using a six-point template takes approximately 2 minutes with each eye; an abnormal test will take several minutes longer. The video of the test is kept for future review and a printout of the test and test series is automatically printed for the patient notes (Figure 2 and e-Supplement 1, available at jaapos.org).
Trial Group Ethical committee approval was gained through the UK National Research and Ethics Service. Parents provided written informed consent for their children. Children attending the eye clinic aged 3 to 10 years with best-corrected visual acuity of at least 0.4 logMAR in the eye(s) to be tested were invited to participate. All children enrolled were clinically judged to be unsuitable to perform automated static perimetry, due to either age or behavioral and learning difficulties. Automated static perimetry was not attempted in any child and no formal evaluation of developmental age was made. The child wore his/her usual spectacle correction during testing. Occlusive patches were fitted under rather than over the spectacle lens to avoid obstruction of the nasal visual field of the other eye; the spectacle rims themselves were not felt likely to obscure the relatively large target. Provided central fixation of the target was achieved and there was no apparent visual obstruction by part of the face, the child was allowed to use his/her compensatory head posture if present. All children underwent preliminary confrontation visual field assessment using a red pin or finger puppets (depending on the child’s age and developmental abilities) by the pediatric ophthalmologist (LEA). Some children were long-term patients and, although at-
Volume 16 Number 3 / June 2012
FIG 1. Screen captured image of the examiner’s monitor showing the webcam image of the child fixating the central cartoon video. A, The pink superimposed square indicates to the examiner the position of the next peripheral target. B, the target becomes green when shown. The examiner clicks on the icon that represents the child’s response. tempted, complete masking of the child’s pathology was not always possible for this part of the test. Subsequently, PLP was performed by a second examiner (either an orthoptist or a specialist pediatric nurse) who was masked to the child’s diagnosis and the result from confrontation testing. The orthoptist (RVP) was experienced in FCPL techniques but the pediatric nurse (EQ) had no such previous experience. All PLP videos were reviewed with a final interpretation being made at the end of the trial by the ophthalmologist, who was masked to the previous findings and the diagnosis (where possible) to assess interobserver variability. The interpretability, sensitivity, and specificity of final interpretation of PLP compared to the current gold standard of confrontation visual field testing were assessed.
Results A total of 74 children aged 3-10 years (mean, 6.6; median, 7 years) participated in the trial, including 32 controls and
Journal of AAPOS
Volume 16 Number 3 / June 2012
Allen et al
263
FIG 2. Serial visual field assessment in a child with a suprasellar astrocytoma. The initial assessment (red outline) illustrates a bitemporal hemianopia. The second test (blue outline) shows the improvement in the left visual field following chemotherapy.
42 children with neurological or ocular pathology that could cause visual field loss. Of the 32 controls, 30 had normal visual fields using confrontation techniques but in 2 the result was uninterpretable. PLP demonstrated normal visual fields in all 32 of these control children. Table 1 compares the results of the visual field testing methods in the whole trial group. Of the 42 children with possible visual field loss, 18 had field deficit confirmed by confrontation, 5 children had normal visual fields, but a further 19 children had uninterpretable results. PLP demonstrated the same pattern of visual field loss (ie, agreement on quadrantic or hemifield loss or visual field constriction) in all 18 of the children found to have a deficit on confrontation techniques. PLP also demonstrated normal visual fields on the five children in this group with normal fields on confrontation testing. Of the 19 children with possible visual field loss who had uninterpretable confrontation visual fields, PLP showed 7 with normal visual fields, 6 with visual field deficit, and 6 with uninterpretable results (Table 2). There was interobserver agreement in 138 of the 140 PLP tests performed (66 children had each eye tested separately, 2 children were blind in 1 eye, and 6 children underwent binocular testing because they were unable to tolerate patching). Disagreement in the interpretation of the PLP testing was found in one eye each of two patients with pathology that could cause visual field loss. In both cases the initial interpretation had been made by the pediatric nurse, who had no previous experience of FCPL techniques. The first was a 10-year-old child with postpapilledematous optic atrophy who had a normal right visual field and constricted left visual field on confrontation. Her initial PLP interpretation had been of normal visual fields in each eye, but, on review by the ophthalmol-
Journal of AAPOS
Table 1. Comparison of the overall interpretability of visual field tests PLP Predicted visual field (normal or abnormal) Uninterpretable Total Confrontation Predicted visual field (normal or abnormal) Uninterpretable Total
53
0
53
15 68
6 6
21 74
PLP, preferential looking perimeter.
ogist, the left field was felt to be constricted. The second was a 10-year-old with a blind right eye due to an orbital optic nerve glioma, known also to involve the optic chiasm. Her confrontation visual field assessment was not helpful because she would not maintain central fixation. PLP was initially interpreted as normal but, on review, it was clear that the child had a left temporal hemianopia. Video 1 (available at jaapos.org) shows the subject’s repeated hunting saccades into the left temporal field throughout testing. By using this adaptive visual behavior, the subject was able to locate the peripheral target in the left hemifield on occasion and this was misinterpreted as a looking response by the initial observer. Assuming confrontation testing as the current gold standard technique for this group of children and comparing the results with the final interpretation of PLP, the sensitivity of PLP in detecting similar visual field deficit to that identified on confrontation testing was 18/18, that is, 100% (95% CI, 82%-100%), and specificity was 35/35, that is, 100% (95% CI, 90%-100%).
264
Allen et al
Volume 16 Number 3 / June 2012
Table 2. Pathology and visual field results in the 42 children with possible visual field deficit Pathology
Number
Mean age (yr)
Pattern of visual field
Confrontation number
PLP number
Cerebral palsy/CVI
10
6
Suprasellar/chiasmal tumors
10
6
Hypoxic ischemic encephalopathy
6
3
Idiopathic intracranial hypertension
3
6
Developmental delay
4
3
Posterior fossa tumors/hydrocephalus
5
8
Bilateral disc drusen
1
7
1 1 1 42
10 2 10
Normal Uninterpretable Homonymous hemianopia Normal Uninterpretable Constricted Temporal hemianopia Normal Uninterpretable Homonymous hemianopia Normal Constricted Normal Uninterpretable Constricted Normal Uninterpretable Constricted Uninterpretable Constricted Homonymous hemianopia Uninterpretable Constricted
0 6 4 3 3 2 2 0 3 3 2 1 0 4 0 0 1 4 1 0 1 1 1 42
4 1 5 4 0 2 4 2 1 3 2 1 0 2 2 0 1 4 0 1 1 1 1 42
Sturge-Weber syndrome Tuberous sclerosis Posttraumatic optic atrophy Total
CVI, cortical visual impairment; PLP, preferential looking perimeter.
PLP gave an interpretable and expected result in 15 (71%) of the 21 children in whom confrontation testing was unhelpful. Overall, PLP was interpretable in 31% more patients than confrontation (95% CI, 16%-45%).
Discussion Automated static perimetry is the accepted gold standard in visual field testing and developmentally normal children over the age of 8 years may be able to perform this test adequately.2-4 Aslam and colleagues5 have recently described a computer game–based, child-friendly, flat-screen perimeter.5 This system may be useful for detecting relative field deficit and scotomata in children but requires a learned and conscious response, which limits its use in younger and developmentally delayed children. Such children are often unable to remain still on a chin rest, unable to suppress their natural looking response to the peripheral target, or give a voluntary, learned response when the target is seen. This is the group in whom confrontation visual field testing is currently the only clinical method available for assessing dense and extensive visual field loss and the group for whom PLP may be most useful. Confrontation techniques are generally believed to have low sensitivity and can be particularly difficult in this group, because the child’s adaptive behavior, such as hunting saccadic eye movements and the tendency to track up the examiner’s arm to the target, can make the test difficult to interpret.6 The use of the looking response in determining the peripheral visual field for infants and young children has
been demonstrated previously. In 1986 Mohn and van Hof-van Duin7 first developed a method of kinetic doublearc perimetry using white spherical targets and the assessment of the child’s eye movements. This technique has subsequently been used to determine the normative visual field and to detect visual field deficit in young children.8-11 Mayer and colleagues developed a hemispheric LED static perimeter that uses a forced-choice assessment of the looking response and demonstrated this to be a sensitive and specific method of detecting quadrantanopia, hemianopia, and visual field constriction of infants and young children.11,12 Subsequently Murray and colleagues developed a perimeter, tested in both adults and children, that uses a peripheral Goldmann perimetry–equivalent target on a flat screen and monitors the eye movement response with a tracking device.13 The PLP described in this study similarly uses the natural preferential looking response and has been designed to appeal to children: the child can sit on the parent’s lap without restraint and he/she can choose a cartoon of interest rather than a plain target. Other advantages are that a video of the test is automatically saved and can be reviewed after testing, and a printout for the clinical notes is generated, making it possible to compare serial tests. The hardware required for testing, namely, the flat screen TV, the personal computer system, and the webcam, are commercially available and relatively inexpensive. A disadvantage of the design is that although large areas of dense visual field loss will be detected, milder losses of sensitivity or scotomata will be missed. Another disadvantage is the inability to have a hemispheric screen display,
Journal of AAPOS
Volume 16 Number 3 / June 2012 although the peripheral target size is modified to allow for this. It could be possible to improve the portability of the test by projecting the targets rather than using a plasma screen, with response detection via a wall-mounted camera. Smaller display screens (for example, a laptop) could also be used; although this would reduce the area of the visual field tested, hemianopic field loss could still be detected. This trial has demonstrated PLP to be a sensitive, specific, and more interpretable alternative to confrontation visual field testing. However there were some limitations and weaknesses in the study design: complete masking of diagnosis was not always possible and the interobserver variation was only tested using two examiners. Although this study did not seek to compare the test times for confrontation and PLP techniques, it is likely that the PLP test does take longer than confrontation testing in eyes with abnormal visual fields or children in whom response inconsistency is poor. PLP does not prevent the use of adaptive visual behavior by children with visual field deficit but it does make it easier to recognize and interpret. We considered using an eye tracker during the development of PLP such as that used in the perimeter described by Murray and colleagues13 but felt that the observation of the child’s visual reaction by an experienced examiner made this additional technology unnecessary. The contrast and luminance of the targets of the PLP are not consistent because of the use of a cartoon video; however, children remained alert and interested throughout the test. Rather than extinguishing the central video completely when a peripheral target appeared, we decided rather to freeze the central video to reduce the cue for the child to search for the peripheral stimulus. Occasional freezing of the central image without a peripheral target might have also reduced this tendency but would have prolonged the test time. Parent feedback led to the development of simpler cartoon stimuli for toddlers and the use of a keystroke rather than a noisy mouse click after the trial. It became obvious that children tended to search for a peripheral target in their blind field if they had not seen one for a while, so the system was subsequently modified after the trial to reshow some “already seen” stimuli if several consecutive targets are undetected. Adaptive visual behavior makes visual field assessment difficult and can lead to the misinterpretation of eye movements on confrontation or PLP testing. The 100% sensitivity and specificity result of PLP compared to confrontation testing was based on the final video interpretation by the consultant ophthalmologist. Although interexaminer agreement was good, 2 of 140 tests were initially misinterpreted by the initial observer, demonstrating that the accuracy of the PLP test requires its interpre-
Journal of AAPOS
Allen et al
265
tation by a clinician experienced in FCPL techniques. Many of these children have subsequently had repeat PLP testing as part of their ongoing clinical care, and the results of initial testing were confirmed; most will never be able to undergo confirmatory automated static perimetry because of their cognitive impairment. The PLP should be considered as an alternative to automated static perimetry for the screening and detection of significant visual field loss in young and developmentally delayed children.
Acknowledgments The authors thank Cambridge Enterprise for its financial and promotional support in the development of the preferential looking perimeter and Dr. Richard Parker, Centre for Applied Medical Statistics, University of Cambridge, for his statistical support. References 1. Allen L. Researchers develop new test for children with vision loss. University of Cambridge. Research news. www.cam.ac.uk/research/ news/researchers-develop-new-test-for-children-with-vision-loss/. Accessed May 14, 2012. 2. Safran AB, Laffi GL, Bullinger A, et al. Feasibility of automated visual field examination in children between 5 and 8 years of age. Br J Ophthalmol 1996;80:515-18. 3. Akar Y, Yilmaz A, Yucel I. Assessment of an effective visual field testing strategy for a normal pediatric population. Ophthalmologica 2008;222:329-33. 4. Wabbels BK, Wilscher S. Feasibility and outcome of automated static perimetry in children using continuous light increment perimetry (CLIP) and fast threshold strategy. Acta Ophthalmol Scand 2005; 83:664-9. 5. Aslam TM, Rahman W, Henson D, Khaw PT. A novel paediatric game-based visual-fields assessor. Br J Ophthalmol 2011;95:921-4. 6. Pandit RJ, Gales K, Griffiths PG. Effectiveness of testing visual fields by confrontation. Lancet 2001;358:1339-40. 7. Mohn G, van Hof-van Duin J. Development of the binocular and monocular visual fields of human infants during the first year of life. Clin Vis Sci 1986;1:51-64. 8. Wilson M, Quinn G, Dobson V, Breton M. Normative values for visual fields in 4- to 12-year-old children using kinetic perimetry. J Paediatr Ophthalmol Strabismus 1991;28:151-3. 9. Quinn GE, Dobson V, Hardy RJ, et al. Visual field extent at 6 years of age in children who had high-risk prethreshold retinopathy of prematurity. Arch Ophthalmol 2011;129:127-32. 10. Dobson V, Brown AM, Harvey EM, Narter DB. Visual field extent in children 3.5-30 months of age tested with a double-arc LED perimeter. Vision Res 1998;38:2743-60. 11. Mayer DL, Fulton AB, Cummings MF. Visual fields of infants assessed with a new perimetric technique. Invest Ophthalmol Vis Sci 1988;29:452-9. 12. Cummings MF, van Hof-van Duin J, Mayer DL, Hansen RM, Fulton AB. Visual fields of young children. Behav Brain Res 1988; 29:7-16. 13. Murray IC, Fleck BW, Brash HM, Macrae ME, Tan LL, Minns RA. Feasibility of saccadic vector optokinetic perimetry: A method of automated static perimetry for children using eye tracking. Ophthalmology 2009;116:2017-26.
To compare the sensitivity, specificity, and interpretability of a newly developed semiautomated static perimeter based on the preferential looking response to the results of confrontation visual field testing in a group of young and/or developmentally delayed children with and without visual field deficit.
METHODS
The preferential looking perimeter (PLP) uses observation of the child’s natural eye movement response to an appearing target to determine the peripheral visual field. We compared preferential looking perimetry to confrontation testing in 74 children 3-10 years of age (mean, 6.6 years; median, 7 years), including 32 controls and 42 children with neurological and ocular disorders that could cause significant visual field deficit.
RESULTS
Using confrontation testing as the gold standard, the PLP was 100% sensitive and 100% specific (95% CI, 90%-100%), with excellent interobserver agreement. An interpretable result could be achieved in 15 (71%) of the 21 children in whom confrontation testing was unhelpful.
CONCLUSIONS
PLP is a useful new technique for assessing significant visual field loss in young or developmentally delayed children, with many advantages over confrontation testing. ( J AAPOS 2012;16:261-265)
A
variety of different neurological and ocular disorders can result in major visual field deficits, such as hemianopia and field constriction in young children. Recognition of a visual field deficit leading to early treatment can sometimes prevent subsequent loss of central vision, aid in the management of unpredictable tumors, such as optic glioma, and allow educational support tailored to the child’s visual deficit. In practice it can be difficult to assess accurately and interpret confrontation visual fields, particularly in children most in need of testing, namely, those with neurofibromatosis type 1, cerebral palsy, or developmental delay. Forced-choice preferential looking (FCPL) is an accepted method for assessing visual acuity in young children. We have developed a preferential looking perimeter (PLP) that uses this spontaneous, natural response to assess visual fields. We report a trial comparing the results of PLP to those from confrontation testing on Author affiliations: aUniversity of Cambridge, The Old Schools, Trinity Lane, Cambridge, United Kingdom; bCambridge University Hospitals NHS Foundation Trust, Cambridge, United Kingdom; cUniversity of York, Heslington, York, United Kingdom Submitted August 10, 2011. Revision accepted January 2, 2012. Correspondence: Louise E. Allen, MBBS, MD, FRCOphth, Consultant Paediatric Ophthalmologist, Addenbrooke’s Hospital, Box 41, Hills Road, Cambridge, CB2 0QQ, UK (email: [email protected]). Copyright Ó 2012 by the American Association for Pediatric Ophthalmology and Strabismus. 1091-8531/$36.00 doi:10.1016/j.jaapos.2012.01.006
Journal of AAPOS
a group of young children, both with and without possible visual field defects, whose age or behavioral and learning difficulties made them unlikely to be able to perform automated static perimetry.
Materials and Methods The Preferential Looking Perimeter The PLP consists of a 1.27 m domestic use flat plasma screen with a modified webcam mounted just below the central point of the screen. This screen size allows 80 of peripheral visual field in the horizontal meridian and 60 on the vertical meridian to be tested when viewed from a distance of 0.5 m. A computer drives the plasma screen and an independent monitor on which the live image from the webcam superimposed with the position of the peripheral targets is seen by the clinician. The central fixation video target subtends 7 and peripheral video targets subtend 5 from the child’s viewing position. The test can be run on each eye separately or with both eyes open to assess the binocular visual field if patching is not possible. A video of the device, named “Kidzeyez,” in operation may be viewed at the University of Cambridge Web site.1 Before the test begins, the child chooses a test stimulus from a range of age-appropriate cartoons and the clinician chooses the template of the peripheral locations to be tested or can design a new one. For the purpose of this trial, a six-point template of peripheral targets initially located at 30 eccentricity superior and inferior to the central target and 35 eccentricity in the
261
262
Allen et al
superotemporal, superonasal, inferotemporal, and inferonasal meridians was used (Figure 1). The test begins with a cartoon video playing on the center of the screen to elicit central fixation. The soundtrack of the cartoon is played on the built-in speakers and gives no clue to the site of the target. The clinician can see the child’s fixation (imaged by the webcam) and a superimposed square, showing the site of the next peripheral stimulus, on the monitor (Figure 1A). While the child is watching the central video, the clinician begins the test. The central video on the plasma screen freezes but the soundtrack continues. The cartoon action continues on the new peripheral site; as this target appears, the superimposed target square changes color on the clinician’s monitor (Figure 1B). The child preferentially looks toward the moving cartoon if it is seen. The examiner studies the child’s eye movements in response to the appearance of the target via the webcam, looking for an accurate saccade toward the target and selects “seen,” “not seen,” or “maybe” depending on the child’s response. The target is illuminated for 5 seconds and then disappears as the cartoon story returns to the central target. Once the child has refixated on the central target, the next stimulus can be enabled. Each template stimulus location is tested in random sequence until there is a consistent response. The 75% “not seen” responses result in the stimulus being shown 5 closer to center later in the test. The 75% response consistency was chosen for simplicity and to minimize the test time (targets only required three consistent responses of four showings to determine the overall response). A normal test using a six-point template takes approximately 2 minutes with each eye; an abnormal test will take several minutes longer. The video of the test is kept for future review and a printout of the test and test series is automatically printed for the patient notes (Figure 2 and e-Supplement 1, available at jaapos.org).
Trial Group Ethical committee approval was gained through the UK National Research and Ethics Service. Parents provided written informed consent for their children. Children attending the eye clinic aged 3 to 10 years with best-corrected visual acuity of at least 0.4 logMAR in the eye(s) to be tested were invited to participate. All children enrolled were clinically judged to be unsuitable to perform automated static perimetry, due to either age or behavioral and learning difficulties. Automated static perimetry was not attempted in any child and no formal evaluation of developmental age was made. The child wore his/her usual spectacle correction during testing. Occlusive patches were fitted under rather than over the spectacle lens to avoid obstruction of the nasal visual field of the other eye; the spectacle rims themselves were not felt likely to obscure the relatively large target. Provided central fixation of the target was achieved and there was no apparent visual obstruction by part of the face, the child was allowed to use his/her compensatory head posture if present. All children underwent preliminary confrontation visual field assessment using a red pin or finger puppets (depending on the child’s age and developmental abilities) by the pediatric ophthalmologist (LEA). Some children were long-term patients and, although at-
Volume 16 Number 3 / June 2012
FIG 1. Screen captured image of the examiner’s monitor showing the webcam image of the child fixating the central cartoon video. A, The pink superimposed square indicates to the examiner the position of the next peripheral target. B, the target becomes green when shown. The examiner clicks on the icon that represents the child’s response. tempted, complete masking of the child’s pathology was not always possible for this part of the test. Subsequently, PLP was performed by a second examiner (either an orthoptist or a specialist pediatric nurse) who was masked to the child’s diagnosis and the result from confrontation testing. The orthoptist (RVP) was experienced in FCPL techniques but the pediatric nurse (EQ) had no such previous experience. All PLP videos were reviewed with a final interpretation being made at the end of the trial by the ophthalmologist, who was masked to the previous findings and the diagnosis (where possible) to assess interobserver variability. The interpretability, sensitivity, and specificity of final interpretation of PLP compared to the current gold standard of confrontation visual field testing were assessed.
Results A total of 74 children aged 3-10 years (mean, 6.6; median, 7 years) participated in the trial, including 32 controls and
Journal of AAPOS
Volume 16 Number 3 / June 2012
Allen et al
263
FIG 2. Serial visual field assessment in a child with a suprasellar astrocytoma. The initial assessment (red outline) illustrates a bitemporal hemianopia. The second test (blue outline) shows the improvement in the left visual field following chemotherapy.
42 children with neurological or ocular pathology that could cause visual field loss. Of the 32 controls, 30 had normal visual fields using confrontation techniques but in 2 the result was uninterpretable. PLP demonstrated normal visual fields in all 32 of these control children. Table 1 compares the results of the visual field testing methods in the whole trial group. Of the 42 children with possible visual field loss, 18 had field deficit confirmed by confrontation, 5 children had normal visual fields, but a further 19 children had uninterpretable results. PLP demonstrated the same pattern of visual field loss (ie, agreement on quadrantic or hemifield loss or visual field constriction) in all 18 of the children found to have a deficit on confrontation techniques. PLP also demonstrated normal visual fields on the five children in this group with normal fields on confrontation testing. Of the 19 children with possible visual field loss who had uninterpretable confrontation visual fields, PLP showed 7 with normal visual fields, 6 with visual field deficit, and 6 with uninterpretable results (Table 2). There was interobserver agreement in 138 of the 140 PLP tests performed (66 children had each eye tested separately, 2 children were blind in 1 eye, and 6 children underwent binocular testing because they were unable to tolerate patching). Disagreement in the interpretation of the PLP testing was found in one eye each of two patients with pathology that could cause visual field loss. In both cases the initial interpretation had been made by the pediatric nurse, who had no previous experience of FCPL techniques. The first was a 10-year-old child with postpapilledematous optic atrophy who had a normal right visual field and constricted left visual field on confrontation. Her initial PLP interpretation had been of normal visual fields in each eye, but, on review by the ophthalmol-
Journal of AAPOS
Table 1. Comparison of the overall interpretability of visual field tests PLP Predicted visual field (normal or abnormal) Uninterpretable Total Confrontation Predicted visual field (normal or abnormal) Uninterpretable Total
53
0
53
15 68
6 6
21 74
PLP, preferential looking perimeter.
ogist, the left field was felt to be constricted. The second was a 10-year-old with a blind right eye due to an orbital optic nerve glioma, known also to involve the optic chiasm. Her confrontation visual field assessment was not helpful because she would not maintain central fixation. PLP was initially interpreted as normal but, on review, it was clear that the child had a left temporal hemianopia. Video 1 (available at jaapos.org) shows the subject’s repeated hunting saccades into the left temporal field throughout testing. By using this adaptive visual behavior, the subject was able to locate the peripheral target in the left hemifield on occasion and this was misinterpreted as a looking response by the initial observer. Assuming confrontation testing as the current gold standard technique for this group of children and comparing the results with the final interpretation of PLP, the sensitivity of PLP in detecting similar visual field deficit to that identified on confrontation testing was 18/18, that is, 100% (95% CI, 82%-100%), and specificity was 35/35, that is, 100% (95% CI, 90%-100%).
264
Allen et al
Volume 16 Number 3 / June 2012
Table 2. Pathology and visual field results in the 42 children with possible visual field deficit Pathology
Number
Mean age (yr)
Pattern of visual field
Confrontation number
PLP number
Cerebral palsy/CVI
10
6
Suprasellar/chiasmal tumors
10
6
Hypoxic ischemic encephalopathy
6
3
Idiopathic intracranial hypertension
3
6
Developmental delay
4
3
Posterior fossa tumors/hydrocephalus
5
8
Bilateral disc drusen
1
7
1 1 1 42
10 2 10
Normal Uninterpretable Homonymous hemianopia Normal Uninterpretable Constricted Temporal hemianopia Normal Uninterpretable Homonymous hemianopia Normal Constricted Normal Uninterpretable Constricted Normal Uninterpretable Constricted Uninterpretable Constricted Homonymous hemianopia Uninterpretable Constricted
0 6 4 3 3 2 2 0 3 3 2 1 0 4 0 0 1 4 1 0 1 1 1 42
4 1 5 4 0 2 4 2 1 3 2 1 0 2 2 0 1 4 0 1 1 1 1 42
Sturge-Weber syndrome Tuberous sclerosis Posttraumatic optic atrophy Total
CVI, cortical visual impairment; PLP, preferential looking perimeter.
PLP gave an interpretable and expected result in 15 (71%) of the 21 children in whom confrontation testing was unhelpful. Overall, PLP was interpretable in 31% more patients than confrontation (95% CI, 16%-45%).
Discussion Automated static perimetry is the accepted gold standard in visual field testing and developmentally normal children over the age of 8 years may be able to perform this test adequately.2-4 Aslam and colleagues5 have recently described a computer game–based, child-friendly, flat-screen perimeter.5 This system may be useful for detecting relative field deficit and scotomata in children but requires a learned and conscious response, which limits its use in younger and developmentally delayed children. Such children are often unable to remain still on a chin rest, unable to suppress their natural looking response to the peripheral target, or give a voluntary, learned response when the target is seen. This is the group in whom confrontation visual field testing is currently the only clinical method available for assessing dense and extensive visual field loss and the group for whom PLP may be most useful. Confrontation techniques are generally believed to have low sensitivity and can be particularly difficult in this group, because the child’s adaptive behavior, such as hunting saccadic eye movements and the tendency to track up the examiner’s arm to the target, can make the test difficult to interpret.6 The use of the looking response in determining the peripheral visual field for infants and young children has
been demonstrated previously. In 1986 Mohn and van Hof-van Duin7 first developed a method of kinetic doublearc perimetry using white spherical targets and the assessment of the child’s eye movements. This technique has subsequently been used to determine the normative visual field and to detect visual field deficit in young children.8-11 Mayer and colleagues developed a hemispheric LED static perimeter that uses a forced-choice assessment of the looking response and demonstrated this to be a sensitive and specific method of detecting quadrantanopia, hemianopia, and visual field constriction of infants and young children.11,12 Subsequently Murray and colleagues developed a perimeter, tested in both adults and children, that uses a peripheral Goldmann perimetry–equivalent target on a flat screen and monitors the eye movement response with a tracking device.13 The PLP described in this study similarly uses the natural preferential looking response and has been designed to appeal to children: the child can sit on the parent’s lap without restraint and he/she can choose a cartoon of interest rather than a plain target. Other advantages are that a video of the test is automatically saved and can be reviewed after testing, and a printout for the clinical notes is generated, making it possible to compare serial tests. The hardware required for testing, namely, the flat screen TV, the personal computer system, and the webcam, are commercially available and relatively inexpensive. A disadvantage of the design is that although large areas of dense visual field loss will be detected, milder losses of sensitivity or scotomata will be missed. Another disadvantage is the inability to have a hemispheric screen display,
Journal of AAPOS
Volume 16 Number 3 / June 2012 although the peripheral target size is modified to allow for this. It could be possible to improve the portability of the test by projecting the targets rather than using a plasma screen, with response detection via a wall-mounted camera. Smaller display screens (for example, a laptop) could also be used; although this would reduce the area of the visual field tested, hemianopic field loss could still be detected. This trial has demonstrated PLP to be a sensitive, specific, and more interpretable alternative to confrontation visual field testing. However there were some limitations and weaknesses in the study design: complete masking of diagnosis was not always possible and the interobserver variation was only tested using two examiners. Although this study did not seek to compare the test times for confrontation and PLP techniques, it is likely that the PLP test does take longer than confrontation testing in eyes with abnormal visual fields or children in whom response inconsistency is poor. PLP does not prevent the use of adaptive visual behavior by children with visual field deficit but it does make it easier to recognize and interpret. We considered using an eye tracker during the development of PLP such as that used in the perimeter described by Murray and colleagues13 but felt that the observation of the child’s visual reaction by an experienced examiner made this additional technology unnecessary. The contrast and luminance of the targets of the PLP are not consistent because of the use of a cartoon video; however, children remained alert and interested throughout the test. Rather than extinguishing the central video completely when a peripheral target appeared, we decided rather to freeze the central video to reduce the cue for the child to search for the peripheral stimulus. Occasional freezing of the central image without a peripheral target might have also reduced this tendency but would have prolonged the test time. Parent feedback led to the development of simpler cartoon stimuli for toddlers and the use of a keystroke rather than a noisy mouse click after the trial. It became obvious that children tended to search for a peripheral target in their blind field if they had not seen one for a while, so the system was subsequently modified after the trial to reshow some “already seen” stimuli if several consecutive targets are undetected. Adaptive visual behavior makes visual field assessment difficult and can lead to the misinterpretation of eye movements on confrontation or PLP testing. The 100% sensitivity and specificity result of PLP compared to confrontation testing was based on the final video interpretation by the consultant ophthalmologist. Although interexaminer agreement was good, 2 of 140 tests were initially misinterpreted by the initial observer, demonstrating that the accuracy of the PLP test requires its interpre-
Journal of AAPOS
Allen et al
265
tation by a clinician experienced in FCPL techniques. Many of these children have subsequently had repeat PLP testing as part of their ongoing clinical care, and the results of initial testing were confirmed; most will never be able to undergo confirmatory automated static perimetry because of their cognitive impairment. The PLP should be considered as an alternative to automated static perimetry for the screening and detection of significant visual field loss in young and developmentally delayed children.
Acknowledgments The authors thank Cambridge Enterprise for its financial and promotional support in the development of the preferential looking perimeter and Dr. Richard Parker, Centre for Applied Medical Statistics, University of Cambridge, for his statistical support. References 1. Allen L. Researchers develop new test for children with vision loss. University of Cambridge. Research news. www.cam.ac.uk/research/ news/researchers-develop-new-test-for-children-with-vision-loss/. Accessed May 14, 2012. 2. Safran AB, Laffi GL, Bullinger A, et al. Feasibility of automated visual field examination in children between 5 and 8 years of age. Br J Ophthalmol 1996;80:515-18. 3. Akar Y, Yilmaz A, Yucel I. Assessment of an effective visual field testing strategy for a normal pediatric population. Ophthalmologica 2008;222:329-33. 4. Wabbels BK, Wilscher S. Feasibility and outcome of automated static perimetry in children using continuous light increment perimetry (CLIP) and fast threshold strategy. Acta Ophthalmol Scand 2005; 83:664-9. 5. Aslam TM, Rahman W, Henson D, Khaw PT. A novel paediatric game-based visual-fields assessor. Br J Ophthalmol 2011;95:921-4. 6. Pandit RJ, Gales K, Griffiths PG. Effectiveness of testing visual fields by confrontation. Lancet 2001;358:1339-40. 7. Mohn G, van Hof-van Duin J. Development of the binocular and monocular visual fields of human infants during the first year of life. Clin Vis Sci 1986;1:51-64. 8. Wilson M, Quinn G, Dobson V, Breton M. Normative values for visual fields in 4- to 12-year-old children using kinetic perimetry. J Paediatr Ophthalmol Strabismus 1991;28:151-3. 9. Quinn GE, Dobson V, Hardy RJ, et al. Visual field extent at 6 years of age in children who had high-risk prethreshold retinopathy of prematurity. Arch Ophthalmol 2011;129:127-32. 10. Dobson V, Brown AM, Harvey EM, Narter DB. Visual field extent in children 3.5-30 months of age tested with a double-arc LED perimeter. Vision Res 1998;38:2743-60. 11. Mayer DL, Fulton AB, Cummings MF. Visual fields of infants assessed with a new perimetric technique. Invest Ophthalmol Vis Sci 1988;29:452-9. 12. Cummings MF, van Hof-van Duin J, Mayer DL, Hansen RM, Fulton AB. Visual fields of young children. Behav Brain Res 1988; 29:7-16. 13. Murray IC, Fleck BW, Brash HM, Macrae ME, Tan LL, Minns RA. Feasibility of saccadic vector optokinetic perimetry: A method of automated static perimetry for children using eye tracking. Ophthalmology 2009;116:2017-26.