Long-term Vision Results Measured With Teller Acuity Cards and a New Light Perception/Projection Scale After Management of Late Stages of Retinopathy of Prematurity

Objectives  To report visual acuity (VA) measured by Teller Acuity Cards (TACs) and a new Light Perception/Projection (LPP) Scale in infants with regressed or treated stage 3, 4, or 5 retinopathy of prematurity (ROP), and to compare VA in eyes that underwent successful vitreoretinal surgery for stage 5 ROP with eyes with persistent retinal detachment.

Methods  Nineteen infants (35 eyes) underwent VA testing using TACs and the LPP scale. The correlation between the methods was determined. Comparisons in VA scores were made in eyes by stage of ROP at the first examination and retinal status at the end of follow-up and between eyes with successful surgical reattachment and persistent retinal detachment.

Results  Scores obtained with the LPP scale and TACs were highly correlated (Spearman rank-order correlation coefficient, 0.92; P<.001). Visual acuity was better in eyes with retinal attachment at the end of follow-up than in eyes with retinal detachment whether the ROP stage at first examination was 4A (n = 6), 4B (n = 16), or 5 (n = 6). In eyes that progressed to stage 5 ROP and had successful surgical retinal reattachment (n = 16), both methods of measurement yielded better visual function than in eyes with persistent retinal detachment. The LPP scale provided scores for eyes without quantifiable grating acuity determined with TAC.

Conclusions  The LPP scale scores were correlated with TAC scores in infants with stages 3, 4, and 5 ROP. Surgical retinal reattachment in stage 5 ROP resulted in better visual function. The LPP scale may be useful in measuring low vision in infants without quantifiable grating acuity and with later stages of ROP.

RETINOPATHY of prematurity (ROP) causes blindness and poor vision in premature infants.¹ The prevention of stage 5 ROP, total retinal detachment, is the most effective way to preserve vision, ² although several case series have reported successful surgical repair and visual rehabilitation in infants with stages 4 and 5 ROP.^3-6 The value of reattaching the retina in stage 5 ROP has been particularly controversial because of poor visual outcome.⁷

Difficulties exist in measuring and interpreting visual acuity in the child with retinal pathologic conditions. The measurement of grating acuity(the ability to resolve stripes of different spatial frequencies) in infants with abnormal maculas or poor macular function may underestimate visual acuity deficits⁸ or visual function⁵ or may not measure very low vision at all in some infants.⁷ Additionally, grating acuity is not always equivalent to recognition acuity⁹ and may not predict visual acuity in the older child.¹⁰ Therefore, the potential for the development of vision at an older age may not be determined.¹¹ In infants with ROP, measuring and interpreting visual acuity may be confounded by other health problems, including developmental delay and hearing deficits. Further, in infants with total retinal detachment, several additional factors—degeneration of retinal elements, detached retina during the critical period of visual development, and the difficulty with the rehabilitation of surgical aphakia associated with retinal repair—can reduce the potential development of foveal vision and, thus, the potential for pattern or recognition vision. In infants with ROP, there is a need to discriminate different levels of low vision when either letter or grating acuity tasks do not provide this information. Although methods that measure low vision exist and are useful in children and some infants, ⁹ in more severely impaired infants, low vision may be missed using current methods, and stratification of levels of very low vision is often not possible.

Because grating acuity may fail to provide information in some infants with very low visual acuity and retinal disease, we were interested in other methods to assess visual function in infants, particularly those with later stages of ROP. We therefore developed the Light Perception/Projection (LPP) Scale to discriminate degrees of visual function at the lowest levels of vision between that measurable with grating targets and the absence of light perception. The LPP scale can be used in most premature infants because it is based on the response to light and is not limited by the patient's age, level of development, or ability or willingness to attempt a certain task. It, therefore, differs from other visual function batteries designed to evaluate behavioral acuity in patients with low vision^9,12,13 and that use some tasks that require a level of child development, such as grabbing an object. It was developed to complement other batteries of visual function.^9,12,14,15

In infants who were referred for management of late stages of ROP, we measured visual acuity by 2 methods: our new LPP scale and a standardized method, ^14,15 using Teller Acuity Cards (TACs) (Vistech Consultants Inc, Dayton, Ohio). We present the correlation between the 2 methods and the visual acuity results based on (1) ROP stage at the first examination and whether the retina was attached at follow-up and (2) stage 5 ROP as the worst stage that developed and whether surgical reattachment of stage 5 ROP was achieved.

The study sample comprised 19 premature infants (Table 1) referred to 1 surgeon (M.E.H.) over a 7-year period for management (1) after laser treatment for stage 3 ROP, (2) of vitreous hemorrhage in association with ROP, or (3) of stage 4 or 5 ROP. The study subjects also had their vision tested by 1 certified vision tester (D.W.R.) using TACs and the LPP scale. All subjects, except 1 (patient 19; Table 1) had previous laser treatment for threshold ROP in both eyes. Patient 19 developed a vitreous hemorrhage and had a vitrectomy and scleral buckle performed in the right eye and developed prethreshold ROP in the left eye that regressed and did not require laser treatment.

All patients underwent ophthalmologic examinations, including dilated funduscopy using scleral depression and indirect ophthalmoscopy. Surgical procedures evolved over the course of the study and included scleral buckle, open-sky vitrectomy, ¹⁶ closed vitrectomy with lensectomy, and lens-sparing vitrectomy.¹⁷ Fourteen patients underwent 1 or more surgical procedures in at least 1 eye. Surgical treatment was recommended when ROP appeared to be progressing (ie, increased number of clock hours of elevated ridge or retinal detachment), when the macula was detached, or if vitreous hemorrhage obscured the posterior pole or 6 or more clock hours of the ridge. Eyes that did not show progression of ROP did not require surgical intervention. For the sake of clarity, eyes that only had laser treatment were not defined as having had surgery. All infants were treated by and had follow-up with the same surgeon (M.E.H.).

The reliability of grating acuity measurements is affected by within-tester variability, variability related to measurement procedures, and variability associated with more than 1 tester.¹⁸ To reduce variability, all grating acuity testing was performed by 1 tester (D.W.R.) only in a windowless, 2.4- × 2.4-m room illuminated by 4.2-m candles. The vision tester was unaware of the ROP status of the infants' eyes, whether surgery had been performed, and whether the retina was attached or detached. Vision was tested at different points in the infant's development. The vision scores from the most recent examination, defined as the end of follow-up, are reported.

Monocular followed binocular testing both with the LPP scale and in grating acuity measurement using TACs. The LPP scale (responses to light) was always assessed before grating acuity. Light perception was assessed with a penlight directed toward the child's eye(s) in a lighted room. If there was no response in a lighted room, light perception was tested in a dark room. To test light projection in each of the 4 field quadrants, the penlight was directed toward the child's eye from an oblique, eccentric position 23 to 31 cm from the gaze direction. The light was presented several times in each quadrant until there were definite, repeatable responses or no or equivocal responses. If able, the child was asked to report when the light was off or on, as well as the location of the light, by pointing or responding verbally. Younger and nonverbal patients were observed for orienting eye movements toward the light, with sufficient time allowed for delayed reaction (maximum of about 5 seconds). Positive responses in young infants (or neurologically impaired children) who were not able to make clear eye or head movements included a change in facial expression and/or body movement when the light was presented. The LPP scale was developed by one of us (D.W.R.) because grating acuity was not measurable in many infants with severe ROP. A lower score implies better visual function (Table 2).

Grating acuity was measured with TACs¹⁴ using a method similar to that described in Trueb et al.¹⁵ The examiner showed the cards to the child in free space at 38 cm (or, occasionally, 19 cm) from the child's face. The gratings were presented either off center horizontally or vertically relative to fixation or directly in front of the child's face, ¹⁹ depending on the child's ocular motor abnormalities (strabismus or nystagmus), gaze abnormalities, and visual field extent. The examiner's task was to determine the finest grating the child detected; this grating provided the child's acuity score. Grating presentation began with the largest striped pattern, the low vision card (0.23 cycles/cm), and continued to progressively finer gratings until a grating was reached that the child apparently did not detect. Each grating card was presented several times, varying the position of the grating on each trial. Gratings could be re-presented as necessary until the examiner was confident of the smallest grating the child detected. The tester always was unaware of the location of the grating until she could make a judgment as to whether the child saw that grating. The judgment that a child detected the grating was based on a definite, positive response on all presentations. Positive responses included, at the highest level, pointing toward the possible grating location, or an eye or head-eye movement toward the off-center grating location, or eye widening and/or change in facial expression for centrally presented cards. Indications the child did not detect the grating included similar eye or head-eye movements to both potential grating positions or no change in facial expression or a look away from the centrally presented card.

Grating acuity was expressed as the number of octaves below the lower 99% prediction limit for the child's age²⁰ corrected for gestational age at birth. An octave is a difference by a factor of 2 of the minutes of arc or the Snellen equivalent. For example, there is one octave difference between 20/200 and 20/400 and between 20/50 and 20/100. Visual acuity was expressed in octaves relative to age norms because of the patients' wide age range (6-68 mo) and because normal acuity improves with age. In eyes in which a quantifiable measurement of grating acuity could not be made, an assessment as to the presence of light perception or its absence was made based on prior testing using the LPP scale. Based on a study of 91 infants treated in a neonatal intensive care unit, in which the within-tester difference of 1 octave or less was 93% and of 0.5 octave or less was 71% in test-retest pairs, ²¹ we considered 1 octave to reflect a significant difference in grating acuity.

Acuities within the 99% prediction limits for the child's age corrected for gestational age at birth were recorded as normal. Acuities below the prediction limit were recorded in octaves. For the purpose of analysis, when only light perception was present, this was estimated and scored as 9 octaves below normal. No perception of light was scored as 12 octaves below normal.

Stages 3, 4, and 5 ROP and threshold disease were defined according to the International Classification of ROP.²² Stage 4A was partial retinal detachment present posterior to the ridge that did not involve the fovea, whereas stage 4B ROP had foveal involvement. Stage 5 ROP was a total retinal detachment as determined by clinical examination. Some stage 5 eyes had attachment of the peripheral retinal trough. The stage at first examination was the diagnosis prior to surgical intervention or further observation if surgery was not indicated. The worst stage was the highest stage that developed and included those eyes with stage 4A or 4B ROP that progressed despite the first surgery. The worst stage did not reflect the anatomic outcome, because some eyes with stage 4 and 5 ROP were reattached surgically and recorded as having retinal attachment at the end of follow-up(Table 1). Retinal attachment at the end of follow-up was defined as retinal attachment at least 1 month after surgery and included some eyes with a peripheral fold or macular dragging. Retinal attachment also included those eyes that did not develop or have progression and, therefore, did not undergo surgery. This included 3 eyes (patients 4[OD], 10 [OD], and 12 [OD]) with stage 4A ROP and 3 eyes (patients 11 [OD] and 17 [OU]) with stage 4B ROP.

Periodic retinal examinations were made. Spectacle correction or contact lens treatment was initiated as determined by the patient's pediatric ophthalmologist. A trial of patching was used to treat amblyopia. Patients were also enrolled in early intervention programs that emphasized enriching environments.

For comparison of each subject's TAC and LPP scale scores, we determined the Spearman rank-order correlation coefficient. This analysis was used rather than the Bland-Altman method of analysis of differences and averages, which is applied when 2 methods are measuring the same physiological quantity (such as blood glucose level or intraocular pressure).²³ This was because, in this case, the 2 methods do not seem to measure the same quality on the same scale. Establishment of the degree of correlation was therefore the primary concern in the present comparison of these 2 methods.

The TAC scores varied with the ROP stage at the first examination; better grating acuity was measured in eyes with stages 3 and 4A ROP and worse or no grating acuity in patients with stages 4B and 5 ROP. On the LPP scale, eyes with scores of 0 or 1 (localization of light in a lit room) were more likely to have had stage 3 or 4 ROP than stage 5 ROP.

When we compared the LPP scale and TAC scores (Figure 1), all eyes with an LPP score of 0 or 1 (response to a light in a normally lit room) had a measurable grating acuity, whereas there was no measurable grating acuity present in eyes with an LPP score of 3 or higher. Only some eyes with an LPP score of 2, indicating response to a light in a dark room with light projection in all 4 quadrants, had a measurable grating acuity. The scores obtained with the LPP scale were highly correlated with TAC scores (Spearman rank-order correlation coefficient, 0.92; P<.001; n = 35 eyes).

Figure 2 presents the number of eyes with retinal attachment or detachment at the end of follow-up by the ROP stage at the first examination and the numbers of eyes that underwent surgical procedures. Retinal attachment was present in 7 eyes with regressed stage 3 ROP, 4 eyes with stage 4A ROP (1 surgically repaired), 13 eyes with stage 4B ROP (10 surgically repaired), and 3 eyes with stage 5 ROP (all surgically repaired). Two eyes with stage 4A ROP at the first examination, 3 with stage 4B ROP, and 3 with stage 5 ROP progressed, and surgery failed to repair retinal detachment.

The TAC scores became higher (ie, worse grating acuity) as stage of ROP at the first examination increased (Figure 3). However, eyes that had either spontaneous or surgically induced retinal attachment had better TAC scores than eyes with the same stage at the first examination but that had retinal detachment at the end of follow-up. This was especially apparent in eyes with stages 4A and 4B ROP. Similar relationships were found with the LPP scale. A higher stage of ROP at the first examination was associated with a poorer (higher) LPP score (Figure 3A). A better LPP score was seen in eyes with all stages of ROP that had retinal attachment compared with those with persistent retinal detachment, except in 1 eye (patient 18 [OS]; Table 1). This patient had had a vitrectomy for stage 5 ROP, still had a retinal detachment at 6 months' corrected age, and had penlight perception(LPP scale score, 2) and quantifiable grating acuity (octaves below normal, 3).

Figure 4 presents visual acuity measured with TACs and the LPP scale in 16 eyes that had stage 5 ROP as the worst stage and that either had successful surgical reattachment of the retina or failed retinal reattachment. Visual acuity measured with both scales was better in eyes with stage 5 ROP that had surgically reattached retinas. Three of 8 eyes with reattached retinas had no quantifiable grating acuity measured by TACs, whereas the LPP scale provided information for these eyes.

The TAC system measures grating acuity and is useful in assessing the vision of infants and nonverbal patients who are able to detect pattern stimuli. However, there is a need to score the lowest levels of vision that exist between TAC-measured acuity and the absence of light perception in early infancy when treatment for ROP is most urgent. This gap is particularly pertinent in the management of later stages of ROP, ^5,6 in which retinal detachment further reduces the potential for visual development. Visual function batteries have been designed to evaluate patients who have much reduced vision.^9,12,13 In the battery reported by Droste et al, ⁹ a variety of behaviors (eg, reaching, avoiding obstacles), all performed under standard office lighting, is assessed, along with fixation, following, and optokinetic nystagmus. Scoring is based on 3 trials on 8 measures for patients mature enough to be tested by all tasks, thus potentially providing a sensitive scale between 0 and 24 points. Another approach is the Visual Ability Score (161 patients; mean age, 36 months), an inventory of visual behavior at home as reported by parents.¹² We designed the LPP scale to discern different levels of low vision below that measured with grating targets and to be used in infants whose developmental stage is below that required to perform some of the tasks of other low vision batteries.^9,12 It is based on the quality of responses to light and is not limited by the patient's age, level of development, or willingness or ability to attempt certain tasks.

In our sample, we found that LPP scores correlated well with TAC scores(Spearman rank-order correlation coefficient, 0.92; P<.001; n = 35 eyes). Although TAC and LPP scale scores were not independent in that the same examiner obtained them in each infant, the magnitude of the correlation between the scores supports the conclusion that the 2 scores are similar and not mainly based on within-tester correlation of the scores for each infant. The LPP scale is easy to administer, requires simple materials, and can be performed with minimal staff training, whereas testing with TACs requires good patient cooperation and experienced trained vision testers. In children with grating acuity, TAC results provide much more information regarding the level of visual function than does the LPP scale. However, the LPP scale can stratify the level of low vision in infants who do not have pattern or grating acuity and may be particularly relevant to the management of later stages of ROP involving retinal detachment and the repair of it. It may measure different aspects of visual function than grating acuity by including localization of light in quadrants of visual field under different lighting conditions. In a patient too young or physically incapacitated for a functional vision battery⁹ and when grating acuity testing is not available, careful observation of localization of light using the LPP scale may be useful.

The value of low levels of vision measured with the LPP scale for development and later quality of life remains unknown, but studies suggest that children may benefit from lower levels of vision than do adults.⁵ Our study supports the importance of prevention of stage 5 ROP for visual function. In addition, our data show that surgically reattached retinas of eyes with stage 5 ROP performed better than eyes with persistently detached retinas when tested with both TACs and the LPP scale. The LPP scale provided a better method to discern differences in the lowest levels of vision. Further statistical analyses would not be meaningful because there were too few subjects to analyze and account for intraindividual correlation between eyes. Although the goal is to prevent stage 5 ROP, there remain those circumstances when an infant is not seen by a retinal specialist early enough to treat progressing stage 4 ROP. For ethical and practical reasons a randomized clinical trial of surgical intervention for stage 5 ROP is not possible. However, the results from this study, using both TACs and our LPP scale, suggest that surgical intervention in stage 5 ROP may result in better visual function and should be considered in some infants.

There is a need for careful assessment of low vision in infants, for example, when managing the infant after surgical intervention for stages 4 and 5 ROP. In the infant with retinal detachment at critical periods of visual development, much is unknown about the visual system and the potential for postoperative visual recovery. Retinal detachment, the timing of its repair, and its chronicity are factors affecting postoperative visual function. Preservation of visual function, even at a level not readily appreciated by a single vision test, may be valuable to other aspects of child development. Unless future studies find no benefit from surgical retinal reattachment on a child's quality of life and development, efforts to repair retinal detachment in infants with late stages of ROP should be strongly considered.

Corresponding author and reprints: M. Elizabeth Hartnett, MD, Department of Ophthalmology, University of North Carolina, 5109D Bioinformatics Building, 130 Mason Farm Rd, CB 7040, Chapel Hill, NC 27599-7040 (e-mail: hartnet@med.unc.edu).

Submitted for publication March 26, 2002; final revision received February 4, 2003; accepted March 20, 2003.

This study was supported in part by US Public Health Service departmental core grant P30EY02377 from the National Eye Institute, National Institutes of Health, Bethesda, Md, and an unrestricted departmental grant to the Louisiana State University Eye Center from Research to Prevent Blindness, Inc, New York, NY.

This study was presented at the Association for Research in Vision and Ophthalmology annual meeting; May 6, 2002; Ft Lauderdale, Fla.