Correlation of monocular grating acuity at age 12 months with recognition acuity at age 4.5 years: findings from the Infant Aphakia Treatment Study

Abstract

Purpose

To determine whether grating acuity at age 12 months can be used to predict recognition acuity at age 4.5 years in children treated for unilateral congenital cataract enrolled in the Infant Aphakia Treatment Study (IATS).

Methods

Traveling testers assessed monocular grating acuity at 12 months of age (Teller Acuity Card Test [TACT]) and recognition acuity at 4.5 years of age (Amblyopia Treatment Study Electronic Visual Acuity Testing, HOTV) in children treated for visually significant monocular cataract in the IATS. Spearman rank correlation was used to evaluate the relationship between visual acuities at the two ages in the treated eyes.

Results

Visual acuity data at both ages were available for 109 of 114 children (96%). Grating acuity at 12 months of age and recognition acuity at 4.5 years of age were significantly correlated for the treated eyes (r_spearman = 0.45; P = 0.001). At age 4.5 years, 67% of the subjects who had grating acuity at 12 months of age within the 95% predictive limits in their treated eye demonstrated recognition acuity better than 20/200. Similarly, at age 4.5 years 67% of the subjects who had grating acuity at age 12 months below the 95% predictive limits in their treated eye demonstrated recognition acuity of 20/200 or worse.

Conclusions

A single grating acuity assessment at age 12 months predicts recognition acuity in a child treated for unilateral congenital cataract in only two-thirds of cases. Clinicians should consider other factors, such as patching compliance and age at surgery, when using an early grating acuity assessment to modify treatment.

Trial registration: clinicaltrials.gov Identifier NCT00212134.

Grating acuity assessments in infants and young children have evolved from a rigorous laboratory-based protocol to a relatively simple procedure that has been used extensively in clinical populations.^1,2 The Teller Acuity Card procedure was developed to determine an infant’s visual acuity at the time of the test; however, application of this procedure with clinical populations has led to studies evaluating the extent to which early grating acuities can predict later recognition acuities. Research with preterm infants has demonstrated significant, albeit modest, correlations between grating acuity at 12 months’ corrected age and recognition acuity at 4 and 5.5 years of age.^3,4 Two studies of infants who had been treated for congenital cataracts demonstrated significant correlations between early grating acuity and later recognition acuity that were higher than those reported for preterm populations (Maurer D, Lewis T & Brent HP, IOVS 1989;30:ARVO Abstract 408,69).⁵ The greater range of both grating and recognition acuities in these cataract patients compared with a more optically uniform group of preterm infants likely allowed for a stronger correlation between these two measures. The longitudinal nature as well as the large range of visual acuities in IATS affords an opportunity to assess the ability of grating acuity at 12 months of age to identify children treated for unilateral congenital cataract whose visual outcome was better than 20/200 using recognition acuity in the treated eye at 4.5 years of age.

Subjects and Methods

The IATS is a multicenter, randomized clinical trial funded by the National Eye Institute designed to assess the benefits as well as the risks of implanting an IOL at the time of initial cataract surgery compared with leaving the patient aphakic. This study followed the tenets of the Declaration of Helsinki, received approval from the institutional review boards of the participating institutions, and complied with the US Health Insurance Portability and Accountability Act of 1996. The off-label research use of the Acrysof SN60AT and MA60AC IOLs (Alcon Laboratories, Fort Worth, TX) for infants was covered by US Food and Drug Administration investigational device exemption G020021.

Previous publications have reported the study design and methodology⁶ as well as the primary outcome results at 12 months and 4.5 years of age.^7,8 (See eAppendix A for a brief summary.)

Statistical Procedures

Measures of monocular grating acuity and recognition acuity were converted to logMAR values for calculation purposes. Statistical analyses demonstrated no significant difference between median visual acuity for the two treatment groups at either age. Given the lack of significant difference between the two treatment groups and the fact that the current analysis relates to prediction of visual acuity, data from all participants were pooled for the present analyses. The relationship between acuities at 12 months and 4.5 years was determined using Spearman rank correlation.

Predictive values were determined by dichotomizing the grating acuity results based on the lower 95% predictive limit from normative monocular data at 12 months of age: 4.21 cy/deg.⁹ Grating acuity values above the lower 95% predictive limits were considered to be within normal limits and represent good vision, while grating acuities below this level were judged to represent poor vision. Recognition acuity values were categorized into good visual acuity (better than 20/200) and poor visual acuity (equal to or worse than 20/200). This cut-off was based on the legal definition of blindness used in the United States: visual acuity that cannot be corrected to better than 20/200.¹⁰ Negative predictive value (NPV) was defined as the proportion of children with grating acuity within 95% prediction limits (4.21 cycles/degree or better) who had recognition acuity better than 20/200. Positive predictive value (PPV) was estimated as the proportion of eyes with poor grating acuity (worse than 4.21 cycles/degree, 95% predictive limits) who had poor recognition acuity (20/200 or worse). We also estimated the sensitivity and specificity of grating acuity within the 95% predictive limits in identifying those children who will have a good recognition acuity (better than 20/200) at age 4.5 years.

Additional analyses grouped patients by factors that might contribute to and improve predictive values, specifically: (1) grouping patients by age at surgery; and (2) grouping patients according to (a) reported decreases in patching between 12 months and 4.5 years of age, (b) the occurrence of a sight-threatening adverse event after 12 months of age, or both (a) and (b). Comparison of the slopes of the regression lines (logMAR at 12 months vs logMAR at 4.5 years) and analysis of covariance were performed for these supplemental analyses, respectively.

Results

A total of 114 infants were enrolled in the IATS; 57 were randomized to each treatment condition. At age 12 months, monocular grating acuity data were obtained on all 114 participants. At age 4.5 years monocular recognition acuity data were obtained on 112 participants: one child was lost to follow-up around 18 months of age and a second was unable to complete the HOTV testing because of a developmental disorder diagnosed at 9 months of age. Additionally, for the present analyses, we excluded 3 patients: one child had endophthalmitis in the early postoperative period that limited visual potential and resulted in a visual acuity of LP at 4.5 years; another developed a retinal detachment and the eye, subsequently, became phthisical, leaving the child with NLP at 4.5 years of age; the third patient was ultimately diagnosed with Stickler’s syndrome. Thus, data on 109 children were available for analysis.

Comparison of logMAR Visual Acuity at 12 months and 4.5 years

The correlation between grating and recognition acuities for the 109 treated eyes was statistically significant (r_spearman = 0.45, P < 0.001). That is, eyes with better grating acuity at 12 months of age tended to have better recognition acuity at 4.5 years. However, as shown in Figure 1, there is a considerable amount of variability in logMAR values of the treated eye at both ages. It is also apparent that the number of patients whose vision improved (n = 60) between the two ages is almost the same as the number whose vision decreased (n = 49).

FIG 1.

Scatterplots of logMAR (resolution: TACT) at age 12 months versus logMAR (recognition: HOTV) at 4.5 years in the treated eyes of 109 patients. The vertical and horizontal lines provide reference points for the threshold levels specified in this analysis: specifically, 0.85 logMAR at 12 months of age (4.21 cycles/degree, which is the lower 95% confidence limit for norms at this age) and 1.0 logMAR at 4.5 years of age, which is 20/200 Snellen. The diagonal line shows a hypothetical perfect correlation between the two measures of 1.0 and can be used to visually identify the individuals whose vision improved versus the individuals whose vision worsened between the two ages. The Spearman rank correlation for these two measures was 0.45 (P < 0.001).

In order to assess the contribution of age at surgery on these predictive values, we grouped our patients into the following three categories: (1) surgery at 4–6 weeks of age (n = 49); (2) surgery at 7–12 weeks of age (n = 28); and (3) surgery after 12 weeks (n = 34). We compared the slopes of the regression lines (logMAR at 12 months vs logMAR at 4.5 years) across these three age groups and determined that there was no statistically significant difference among the groups (P = 0.218).

Exploring intervening events that might influence long-term recognition acuity, we considered the effects of changes in reported patching and the occurrence of one or more sight-threatening adverse events between 12 months and 4.5 years on the ability of grating acuity at 12 months to predict recognition acuity at 4.5 years. We divided patients into four groups based on specific experiences between 12 months and 4.5 years: (1) reported patching on average ≤30 minutes per day between 12 months and 4.5 years (n = 5), (2) experienced a sight-threatening adverse event (AE) between ages 12 months and 4.5 years (n = 54), (3) both poor compliers and experienced an AE between 12 months and 4.5 years (n = 10), and (4) all others (n = 45). The additional analysis showed no relationship between grating acuity at 12 months of age and recognition acuity at 4.5 years across these four groupings (P = 0.71). For the sake of clarity, we are not plotting these separate groups in the Figure 1.

Grating Acuity at 12 Months as a Screening Assessment

Given the significant correlation between grating and recognition acuities in treated eyes, we hypothesized that we would be able to predict which patients were likely to have better recognition acuity at 4.5 years based on grating acuity measures at 12 months of age. NPV was 67% and PPV was 67%, with a sensitivity of 70% and a specificity of 64% (Table 1). Thus, although children with poorer grating acuity in the treated eye at 12 months of age tended to have poorer recognition acuity at 4.5 years of age, agreement between grating acuity and recognition acuity was only fair (κ = 0.34; 95% CI, 0.16–0.52), suggesting that grating acuity was only able to discriminate between those who would have better recognition vision at 4.5 years of age slightly more than would be expected by chance (Table 1). Approximately two-thirds of children (70%; 95% CI, 0.56–0.81) who had grating acuity within the 95% predictive limits at 12 months had recognition acuity better than 20/200 at 4.5 years of age (sensitivity). A similar proportion (64%; 95% CI, 0.50–0.77) of those with grating acuity below the 95% predictive limits at 12 months of age had recognition acuity of 20/200 or worse at 4.5 years (specificity).

Table 1.

Comparison of visual acuity across age by two categories^a

Visual acuity at 12 months	Visual acuity at 4.5 years, no. (%)
	Better than 20/200	Equal to or worse than 20/200	Total
Within 95% predictive limits	39 (70)	19 (36)	58
Below 95% predictive limits	17 (30)	34 (64)	51
Total	56	53	109

^aSensitivity, 70% (95% CI, 0.56–0.81), ie, good acuity at 12 months and at 4.5 years; specificity, 64% (95% CI, 0.50–0.77), ie, poor acuity at 12 months and at 4.5 years; positive predictive value = 66.7 (34/51); negative predictive value = 67.2 (39/58).

Additional Observations

Visual inspection of Figure 1 reveals that all patients whose grating acuity was worse than 1.5 logMAR ended up with 20/200 or worse recognition acuity. A final analysis of these data is shown in Table 2. We have defined good recognition acuity as equal to or better than 20/50, which is generally considered a level of acuity required for reading and driving. Using this criterion, NPV was 45% and PPV was 68%, with a sensitivity of 70% and a specificity of 56% (Table 2).

Table 2.

Comparison of visual acuity across age by two modified categories^a

Visual acuity at 12 months	Visual acuity at 4.5 years, no. (%)
	Equal to or better than 20/50	Worse than 20/50	Total
Within 95% predictive limits	26 (70)	32 (44)	58
Below 95% predictive limits	11 (30)	40 (56)	51
Total	37	72	109

^aSensitivity, 70% ((95% CI, 0.54–0.82)), ie, good acuity at 12 months and at 4.5 years; specificity, 56% (95% CI: 0.44–0.66) ie, poor acuity at 12 months and at 4.5 years; positive predictive value = 78.4 (40/51); negative predictive value = 44.8 (26/58).

Discussion

Among IATS participants, grating acuity at 12 months was significantly correlated with recognition acuity at 4.5 years in the treated eyes. This finding likely stems, at least in part, from the large variability of both grating and recognition acuity in these eyes, with ranges from better than mean normative grating acuity to pattern vision only and 20/20 to pattern vision only. However, the correlation of 0.45 indicates that only 20% (0.45²) of the variability in the 4.5-year recognition acuity can be accounted for by the grating acuity at 12 months. In addition, a single measure of grating acuity at 12 months of age did not accurately predict which children would have good recognition acuity vision and which would not at age 4.5 years.

Previous research with typically developing infants has demonstrated poor prediction of grating acuity findings over periods as short as 5 months.¹¹ Specifically, Courage and Adams¹¹ followed 27 full-term infants during their first postnatal year, assessing binocular grating acuity at least twice with a mean testing interval of 5 months. In their population, an early estimate of grating acuity was not predictive of a later estimate.

Another study¹² with typically developing infants (n = 45) demonstrated longitudinal improvements in monocular grating acuity. All participants were tested once before 12 months of age and at least twice more up to 23 months of age. These data indicated that infants who initially demonstrated above-average acuity estimates tended to maintain above average grating acuity. On the other hand, infants with an initial below-average acuity demonstrated more erratic patterns of grating acuity development. Thus, the authors conclude that above-average acuity estimates during infancy had more predictive value than those initially below average.

Other studies have used multiple assessments of grating acuity in well-defined clinical populations. Mash and Dobson³ reported correlations ranging from 0.22 to 0.66 in preterm infants followed longitudinally between grating acuity and recognition acuity assessments at 48-months corrected age. They found excellent NPV (88–95%) and limited PPV (67%) with the best predictive value when the grating acuity test was conducted at 24–36 months and worst when the grating acuity test was conducted at 11–17 months. In other words, conducting the grating acuity assessment too early may be an important factor in limiting predictive validity. Birch and colleagues⁵ reported correlations ranging from 0.69 to 0.88 between grating acuity measurements obtained at 24, 36, and 48 months of age and recognition acuity measured at 5–8 years of age for 14 children who had been treated for a unilateral congenital cataract. These children were selected as ones with good to excellent compliance with patching and contact lens and glasses wear, as well as no serious postoperative adverse events. Two distinct contrasts between Birch’s study and IATS are: 1) The IATS cohort included children with limited compliance and sight-threatening adverse events both before and after the 12-month grating acuity test; 2) we compared recognition acuity at 4.5 years with grating acuity at 12 months, while Birch’s recognition acuity assessments were conducted at 5–8 years of age and compared with grating acuity tests at 24, 36, and 48 months. (See eAppendix B for additional comments regarding other analyses.)

There are several limitations pertinent to this particular consideration of the IATS data. It is well established that final visual acuity after unilateral congenital cataract surgery depends mainly on age at surgery, severity of the cataract, postsurgical adverse events or complications, and adherence to treatment for amblyopia, which includes both refractive correction and occlusion.^13–15 IATS was designed to assess the advantages and disadvantages of treating young infants with an IOL or leaving them aphakic after cataract surgery. The study was powered to identify a 0.2 logMAR difference between the two treatment groups at both 12 months and 4.5 years of age. Other variables contributing to final visual outcome were either controlled for through randomization (eg, age at surgery) or monitored closely (eg, adherence to patching and refractive correction).

Our adherence data are based on parental report using a combination of 7-day diaries and 48-hour retrospective interview data. These protocols and results have been explained in earlier publications.^16–18 Nonetheless, the rigor of these data can be challenged, because we did not use an objective occlusion or spectacle-wear dose-monitoring system. Another constraint of our protocol is that we tested only recognition acuity at age 4.5 years. The protocol for this clinic visit included numerous assessments, and we determined that obtaining a grating acuity estimate at this visit would be unrealistic. Our comparison of grating with recognition acuity between the two ages was practical, albeit not optimal. Literature from both pediatric and adult patients with amblyopia has clearly demonstrated that there are large discrepancies between grating and recognition acuities in amblyopic eyes, such that grating acuity underestimates the severity of the amblyopia.^19–23 A more precise comparison would have been to obtain grating acuity at both 12 months and 4.5 years.

A major strength of this study is that 100% of the enrolled patients were tested at 12 months of age and 99% were tested at 4.5 years of age. Furthermore, visual acuity was assessed by a limited number of testers, who were masked to treatment and to the results of previous visual acuity testing: two at 12 months of age and two at age 4.5 years. Data from patients 109 (96%) were used in the present analysis.

The challenging visual rehabilitation for patients with a unilateral congenital cataract warrants ongoing, reliable, and valid assessments of visual function. Providing parents with quantitative information throughout the years of rehabilitation can help motivate them to continue the demanding regimen.¹⁴ Our data suggest that grating acuity assessments at 12 months of age have limited utility in predicting the long-term visual outcome of these eyes. Nonetheless, this quantitative information may be helpful for providing anticipatory guidance regarding potential visual limitations for the child^24,25 and possibly for tailoring the prescribed patching.²⁶