Slow Binocular Reading in Amblyopic Children is a Fellow Eye Deficit

Abstract

Significance.

Amblyopic children read 25% slower than their peers during binocular silent reading.

Purpose.

We compared binocular reading to fellow eye reading to determine whether slow reading in amblyopic children is due to binocular inhibition i.e., the amblyopic eye is interfering during binocular reading.

Methods.

In a cross-sectional study, 38 children with amblyopia and 36 age-similar control children who completed grades one to six were enrolled. Children silently read grade-appropriate paragraphs during binocular reading and fellow eye reading while wearing ReadAlyzer eye-tracking goggles. Reading rate, number of forward saccades, number of regressive saccades, and fixation duration were analyzed between groups and between viewing conditions. We also examined whether sensory factors (amblyopia severity, stereoacuity, suppression) were related to slow reading.

Results.

For amblyopic children, binocular reading versus fellow eye reading did not differ for reading rate (176±60 wpm vs 173±53 wpm, p=.69), number of forward saccades (104±35 saccades/100 words vs 97±33 saccades/100 words, p=.18), number of regressive saccades (21±15 saccades/100 words vs 22±13 saccades/100 words, p=.75), or fixation duration (0.31±0.06 sec vs 0.32±0.07 sec, p=.44). As expected, amblyopic children had a slower reading rate and more forward saccades than control children during binocular reading and fellow eye reading. Slow reading was not related to any sensory factors.

Conclusions.

Binocular reading did not differ from fellow eye reading in amblyopic children. Thus, binocular inhibition is unlikely to play a role in slow binocular reading, and is instead a fellow eye deficit that emerges from a disruption in binocular visual experience during development.

Amblyopia is the most common cause of monocular visual impairment in children at 3% of the USA population.¹ It most often is associated with binocularly discordant visual input due to strabismus or anisometropia during a critical period of development.¹ Because on average, amblyopia is present in 1 or 2 children in every USA classroom,¹ it is important to study how amblyopia can affect a child’s performance on academic tasks such as reading. We previously reported that during binocular viewing, amblyopic children with strabismus or anisometropia silently read about 25% slower than controls when reading for comprehension.^2,3

But why is binocular reading for comprehension slower in amblyopic children compared to controls when they can rely on their fellow eye that has normal visual acuity? One suggestion is that binocular inhibition is actively occurring at the time of reading. Specifically, although we typically consider suppression to be an effect of the fellow eye on the amblyopic eye,⁴ it is also possible that the amblyopic eye affects fellow eye performance during binocular viewing. The reduced contrast sensitivity of fellow eyes during binocular viewing relative to monocular viewing is one type of binocular inhibition that has been reported in amblyopia.⁵ Alternatively, disruption of binocular vision by discordant visual experience early in life due to anisometropia and strabismus not only places the child at risk for amblyopia, but also disrupts the development and maintenance of binocular cortical connections. In turn, both monocular and binocular cell functioning is affected, and results in deficits not only in the amblyopic eye, but also in the fellow eye. Presence of fellow eye deficits in reading speed even when the amblyopic eye is occluded for monocular testing would rule out inhibition by the amblyopic eye and suggest that binocular reading speed is reduced due to the consequences of binocular dysfunction.

Two studies have evaluated fellow eye reading for comprehension in strabismic amblyopia with inconsistent results. Stifter et al⁶ reported a binocular deficit in oral reading speed for printed sentences, but no difference in fellow eye oral reading speed compared to controls. In contrast, Kanonidou et al⁷ reported that silent reading speed for paragraphs presented on a distant screen was slower in amblyopic adults compared to controls during both binocular reading and when reading with the fellow eye alone. However, neither study measured silent reading speed for age-appropriate paragraphs with 35–50 cm binocular viewing distance as is typical in school settings. Eye movement parameters differ substantially between silent and oral reading, and with viewing distance.⁸ Moreover, articulation during oral reading reduces test-retest reliability of both reading speed and eye movement parameters during reading, so the association between reading skills and eye movement patterns is lower during oral compared to silent reading.⁸

Here, we examine whether slow silent reading for comprehension in children with strabismic and anisometropic amblyopia is due to binocular inhibition or a fellow eye deficit. If binocular inhibition is actively occurring during reading, we expect to see a faster reading rate during fellow eye reading compared with binocular reading. However, if slow reading in amblyopic children is due to a fellow eye deficit due to early binocular disruption, slow reading should be found in both binocular and fellow eye reading conditions.

METHODS

Ethics

The research protocol observed the tenets of the Declaration of Helsinki, was approved by the Institutional Review Board of the University of Texas Southwestern Medical Center and conformed to the requirements of the United States Health Insurance Portability and Privacy Act. Informed consent was obtained from a parent and assent was obtained from children ≥10 years of age.

Participants

Children who have completed grades (i.e., school year levels) one to six and who were diagnosed with unilateral amblyopia (anisometropic, strabismic, or combined mechanism) were referred to the Retina Foundation of the Southwest by pediatric ophthalmologists in the Dallas-Fort Worth area between August 2017 and July 2022. Children were between the ages of 7 and 13 years, which is typical for grades one to six in the USA. Diagnosis and ocular alignment were obtained from medical records. Strabismic children were initially diagnosed with esotropia but were aligned to within 4 prism diopters of orthotropia at near. Strabismic children with larger misalignments were not included because misalignments >4 prism diopters could impact tracking reliability of the Readalyzer and be a confounding factor for binocular measurement. Amblyopia was defined as an interocular difference in visual acuity ≥0.2 logMAR, amblyopic eye best-corrected visual acuity ≥0.2 logMAR (≥20/32), and fellow eye best-corrected visual acuity ≤0.1 logMAR (≤20/25). Age-similar control children who had no history of vision disorders and had normal visual acuity and stereoacuity were also enrolled. Exclusion criteria were prematurity ≥8 weeks, co–existing ocular or systemic disease, congenital infections/malformations, developmental delay, and dyslexia/reading disorders. The primary language for all children was English. All children were tested with their habitual spectacle correction.

Procedure

Vision Assessment

A vision assessment was conducted prior to the reading experiment that included; 1) Crowded monocular best-corrected visual acuity with the e-ETDRS protocol, scored in logMAR.^9,10 2) Stereoacuity with the Randot Preschool Stereoacuity and Stereo Butterfly Tests,¹¹ converted to log arcsec (ranging from 1.3 to 3.3 log arcsec). No measurable stereoacuity was arbitrarily assigned a value of 4 log arcsec. 3) Extent of suppression scotoma was quantified with the Worth 4-dot fusion test measured as the farthest distance that 4 dots are reported using up to 7 different distances (16 cm, 33 cm, 50 cm, 67 cm, 100 cm, 200 cm, 300 cm), converted to size of suppression scotoma in log degrees.^12–14

Reading

Binocular reading and fellow eye monocular reading were assessed using the ReadAlyzer (Compevo AB, Stockholm, Sweden) infrared eye movement recording goggles worn over the child’s habitual optical correction. Children sat eye level at a comfortable reading distance (35–40 cm) and were asked to silently read at their normal reading speed for comprehension to replicate conditions typically used for completing schoolwork. Children read three passages; the first grade-level 1 passage was a practice run and the other two passages were at the last grade-level completed. Children read one grade-level paragraph viewing binocularly and one grade-level paragraph viewing with the fellow, nonamblyopic eye (right eye for controls); viewing conditions were counterbalanced. An Ortopad^® eye patch was placed over the amblyopic eye (left eye for controls) on the outside of the goggles during fellow eye reading. Ten ‘Yes/No’ comprehension questions followed each reading to ensure the child comprehended the story. Passages were standardized Readalyzer paragraphs taken from Taylor¹⁵ that used several readability formulas to verify the grade level status of each passage. Passages averaged 12 lines and 100 words. Font size approximately corresponded to 20/160 (0.9 logMAR) for grades one to 3 and 20/100 (0.7 logMAR) for grades four to six. Recordings were acceptable if tracking reliability was ≥70% and comprehension was ≥80%. Tracking reliability may be lowered by excessive blinking or head movements, ≥5% artifact time, and occlusion of an infrared sensor by glasses frames. If a child’s comprehension or reliability did not meet these criteria, they were given another passage of the same grade-level to read. If after a second try, criteria were still not met, the experiment was terminated. Reading rate (words per minute, wpm), the number of forward and regressive saccades (per 100 words), and fixation duration (seconds) were calculated by Readalyzer software.

Statistical Analyses

Primary Analyses. Independent t-tests were conducted to determine differences between groups (amblyopia, control) and paired samples t-tests were conducted to determine differences within the amblyopia group between viewing condition (binocular, fellow eye) for reading rate, number of forward saccades, number of regressive saccades, and fixation duration. Secondary Analyses. For the amblyopic group, Mann Whitney U tests and Wilcoxon Signed Ranks tests were conducted to determine between groups and within-groups differences for sensory factors contributing to reading rate; amblyopia severity (mild, 0.2−0.3 logMAR; moderate/severe, 0.4−1.2 logMAR); stereoacuity measurable (present, not present); extent of suppression scotoma (bifoveal-macular fusion, −0.15 to 0.45 log deg; peripheral-no fusion, 0.60−1.2 log deg). All tests were corrected for multiple comparisons and p values were adjusted using Holm’s sequential Bonferroni procedure, which corrects for Type I error as effectively as the traditional Bonferroni method while retaining more statistical power.¹⁶ Effect size was calculated using Cohen’s d.

Sample size was based on our prior study showing that control children read an average rate of 204 words per minute and that amblyopic children read at an average rate of 148 words per minute. For α=.05 and 1−β of .80, the required sample size was 20 children per group. To account for an anticipated 5% dropout rate, and to conduct secondary analyses determining factors associated with slow reading, we planned to enroll 30 children per group (60 total children).

There is evidence that development of reading ability varies with socioeconomic status and early exposure to literacy in children.¹⁷ To control for socioeconomic status and education level between groups, we used the nationally-normed Child Opportunity Index scores (from 1 to 100) for the education domain and for the social and economic domain in the Child Opportunity Index 2.0 ZIP Code data.¹⁸ Using ZIP codes (US postal codes), the Child Opportunity Index is an instrument that describes and quantifies the neighborhood conditions that children in the USA experience today, ranking them from lowest to highest opportunity (higher scores = better opportunity).

RESULTS

Group characteristics for amblyopic and control children are provided in Table 1. Data from 6 children (2 control, 4 amblyopic) were excluded from analysis due to <80% comprehension, and from 11 children (6 control, 5 amblyopic) due to <70% tracking reliability. A total of 74 children met the comprehension and tracking reliability criteria; 38 amblyopic children (mean age±SD=10.0±1.5 years) and 36 control children (10.3±1.7 years). Based on Mann Whitney U-tests, groups did not differ in nationally-normed Child Opportunity Index education (U=563, P=.26) or social and economic (U=661, P=.97) scores, last grade completed (U=622, P=.50), or comprehension for binocular reading (U=554, P=.13) or fellow eye reading (U=679, P=.95). No sex differences were found for any reading measure for the amblyopic or control group (all Ps >.10).

Table 1.

Group characteristics.

	Amblyopic (n=38)	Control (n=36)	Amblyopic vs. Control
Sex: F, n (%)	21 (55)	14 (39)	P=.16
Etiology, n (%)
Strabismus,	9 (24)	n/a
Anisometropia	10 (26)	n/a
Strabismus + anisometropia	19 (50)	n/a
Treatment Statusa, n (%)
Continuing	11 (29)	n/a
Completed	27 (71)	n/a
Age, Mean±SDb years	10.1±1.5	10.3±1.7	P=.49
(range)	(8.1 to 13.4)	(7.2 to 13.6)
Last grade completed, Mean±SD	3.3±1.6	3.6±1.7	P=.50
(range)	(1 to 6)	(1 to 6)
Comprehension, Mean±SD %
Binocular reading	94±8	91±9	P=.13
(range)	(80 to 100)	(80 to 100)
Fellow eye reading	94±8	94±8	P=.95
(range)	(80 to 100)	(80 to 100)
Child Opportunity Index education score	69±20	73±20	P=.26
(range)	(27 to 94)	(30 to 95)
Child Opportunity Index social/economic score	74±20	75±19	P=.97
(range)	(24 to 98)	(24 to 99)
Amblyopic eyec best-corrected visual acuity
Mean±SD logMARd	0.4±0.2	0.0±0.1	P<.001
(Snellen equivalent)	(20/50±2 lines)	(20/20±1 lines)
(range)	(0.2 to 1.2)	(−0.2 to 0.1)
Fellow eyee best-corrected visual acuity
Mean±SD logMAR	0.0±0.1	0.0±0.1	P=.25
(Snellen equivalent)	(20/20±1 lines)	(20/20±1 lines)
(range)	(−0.2 to 0.1)	(−0.2 to 0.1)
Stereoacuity, Mean±SD log arcsec	3.5±.9	1.6±.1	P<.001
(range)	(1.6 to 4)	(1.3 to 1.8)
Suppression, Mean±SD log deg	0.4±0.5	−0.2±0.0	P<.001
(range)	(−0.2 to 1.2)	(−0.2 to −0.2)

^a.Treatment consisted of patching, atropine drops, and binocular treatment

^b.SD = standard deviation

^c.For control children, the left eye is listed for amblyopic eye best-corrected visual acuity

^d.logMAR, logarithm of the minimum angle of resolution

^e.For control children, the right eye is listed for fellow eye best-corrected visual acuity

^f.n/a, not applicable

Reading Rate

No within-groups differences in reading rate were found for binocular reading versus fellow eye reading for the amblyopic group (176±60 wpm vs 173±53 wpm, t₃₇=0.4, P=.69, d=0.1). As expected, the amblyopic group had a slower binocular reading rate (t₇₂=2.5, P=.016, d=0.6) and fellow eye reading rate (t₇₂=2.7, P=.010, d=0.6) than the control group (binocular, 217±83; right eye, 213±75).

Forward Saccades

No within-groups differences in the number of forward saccades were found for binocular reading versus fellow eye reading for the amblyopic group (104±35 saccades/100 words vs 97±33 saccades/100 words, t₃₇=1.4, P=.18, d=0.2). As expected, the amblyopic group had more forward saccades during binocular reading (t₇₂=3.1, P=.003, d=0.7) and fellow eye reading (t₇₂=3.2, P=.002, d=0.8) than the control group (binocular, 81±26 saccades/100 words; right eye, 76±22 saccades/100 words).

Regressive Saccades

No within-groups differences in the number of regressive saccades was found for binocular reading versus fellow eye reading for the amblyopic group (21±15 saccades/100 words vs 22±13 saccades/100 words, t₃₇=0.3, P=.75, d=0.1). As expected, the amblyopic group produced a similar amount of regressive saccades during binocular reading (t₇₂=1.4, P=.16, d=0.3), but unexpectedly, significantly more regressive saccades during fellow eye reading (t₇₂=2.5, P=.016, d=0.6) than the control group (binocular, 16±12 saccades/100 words; right eye, 15±10 saccades/100 words).

Fixation Duration

No within-groups difference in fixation duration was found for binocular reading versus fellow eye reading for the amblyopic group (0.31±0.06 sec vs 0.32±0.07 sec, t₃₇=0.8, P=.44, d=0.1). As expected, the amblyopic group had similar fixation duration during binocular reading (t₇₂=0.5, P=.61, d=0.1) and fellow eye reading (t₇₂=1.5, P=.14, d=0.3) as the control group (binocular, 0.32±0.06 sec; right eye, 0.35±0.09 sec). See Figure 1 for group means for all reading measures.

Figure 1.

Reading measures. Bar graphs depicting group means for reading rate (A), number of forward saccades (B), fixation duration (C), and number of regressive saccades (D) during binocular reading (white bars) and fellow eye (right eye for controls) reading (grey bars) for amblyopic and control children. Error bars represent ± standard error of the mean (SEM). *P < .05.

Factors Associated with Slow Reading

Including only amblyopic children, we furthered evaluated whether sensory factors (amblyopia, mild vs moderate/severe; stereoacuity, present vs not present; extent of suppression scotoma, bifoveal-macular fusion, vs peripheral-no fusion) were related to slow reading by comparing performance between groups and between viewing conditions. No sensory factors were related to slower reading rate within the amblyopic group (Table 2). Relative visual performance under binocular and monocular conditions may be different between amblyopic children with fusion and some degree of stereopsis and amblyopic children with no sensory fusion. Thus, we also conducted Kruskall-Wallis Analysis of Variance (ANOVAs) and Wilcoxon Signed Ranks tests to determine differences in binocular and fellow eye reading rate for amblyopic children with measurable stereopsis and bifoveal-macular fusion (n=10), amblyopic children with no measurable stereopsis but with bifoveal-macular fusion (n=15), and amblyopic children with no measurable stereopsis or fusion (n=13). No significant differences were found between groups or between viewing conditions (all Ps ≥.21).

Table 2.

Between-groups and within-groups comparisons of binocular and fellow eye reading rate for sensory factors in amblyopic children.

Sensory Factora	n	Binocular Reading Rate wpmb, Mean (SD)c	Fellow Eye Reading Rate wpm, Mean (SD)	Binocular vs. Fellow Eye
Control	36	217 (83)	213 (75)
Amblyopiad
Mild	22	173 (63)	161 (43)	Z=1.0, P=.32, d=0.4
Moderate/Severe	16	180 (57)	190 (62)	Z=0.6, P=.54, d=0.3
Mild vs. Moderate/Severe		U=164, P=.73, d=0.1	U=127, P=.15, d=0.5
Stereoacuityd
Present	12	190 (82)	177 (68)	Z=0.9, P=.37, d=0.1
Not Present	26	169 (47)	172 (46)	Z=0.2, P=.84, d=0.1
Present vs. Not Present		U=135, P=.53, d=0.2	U=155, P=.96, d=0.0
Extent of Suppressiond
Bifoveal-macular fusion	23	168 (64)	162 (51)	Z=0.7, P=.46, d=0.3
Peripheral - no fusion	15	188 (53)	190 (52)	Z=0.3, P=.73, d=0.2
Bifoveal-macular fusion vs. peripheral - no fusion		U=136, P=.27, d=0.4	U=115, P=.08, d=0.6

^a.No difference in last grade completed between sensory categories (all Ps ≥.43)

^b.wpm, words per minute

^c.SD = standard deviation

^d.Categories include amblyopic children only; control children are excluded.

DISCUSSION

While we did not explicitly measure binocular inhibition, we can infer that binocular inhibition is not the cause of slower reading for comprehension in amblyopic children because we did not find better fellow eye reading than binocular reading when the amblyopic eye was occluded. This is not surprising given the lack of association of reading rate with amblyopic eye visual acuity, stereoacuity, or Worth 4-dot suppression. It is becoming increasingly clear that visual and ocular motor deficits can occur in the fellow eye of amblyopic individuals, and that other daily tasks involving vision, such as motion perception and fine motor ability, are often impaired when viewing with both eyes.^19–22 In the current study, even fellow eye reading was slower in children with amblyopia. Fellow eye deficits compared to controls may be indicative of binocular dysfunction that arises from binocularly discordant visual experience early in life in amblyopic children. Namely, this discordant experience can impact the development of visual functions in the fellow eye via unbalanced visual input to binocular neurons that respond to stimulation regardless of the eye of origin. Visual area V2 and middle temporal visual area (MT+/V5) contain neurons that respond to stimulation from either eye.^23,24 Neurons in these areas show abnormalities in spiking and receptive field structure, with unusually high neuronal variability in animal models of amblyopia.^23–25

Reading requires a coordinated sequence of eye movements, including saccades to move the eyes forward and backward (regressive saccades) through lines of text.^26,27 Ocular motor dysfunction is a hallmark of amblyopia, including abnormal saccade initiation and execution and fixation instability.^28–32 Pertinent to this study, the fellow eye of amblyopic individuals has larger fixation instability than controls, and more variable saccade amplitude.^29,32–34 Fellow eye instability during both monocular and binocular viewing suggests that discordant binocular visual experience interferes with the development of ocular motor control and points to a common neural mechanism that controls the stability of both the fellow and amblyopic eye. We previously reported that slow binocular reading in amblyopia is related to making more forward saccades than controls,² and to fixation instability in the fellow eye.³ In the current study, we also found more forward saccades during binocular reading as well as fellow eye reading in the amblyopic group compared to controls. Unlike our binocular reading studies,² here we found significantly more regressive saccades for amblyopic children than controls during fellow eye reading but not during binocular viewing. The apparent increase in regressive saccades during fellow eye reading may be due to fusion maldevelopment nystagmus (FMNS) that is amplified when one eye is occluded;^34,35 the Readalzyer is unable to distinguish between reading-related saccades and FMNS saccades. Thus, slower reading in amblyopic children compared to controls is likely due to ocular motor impairments, but the nature of these impairments remains unclear. Amblyopic children may be planning smaller, more cautious movements, or they could be making regular saccades that are inaccurate in landing so they need to make small, corrective saccades. Either of these might be related to fixation instability, which has been correlated previously with reading rate.³ It is also possible that reduced visual span,³⁶ crowding,³⁷ or higher-order impairments in visual attention³⁸ are impacting reading speed; all of which could also be disrupted by fixation instability.

Our study had limitations. Reading may be impacted by socio-economic status and parental education.¹⁷ While we did not collect individual data from each participant regarding these two possible confounders, we controlled for them using the nationally-normed child opportunity index scores for the education domain and for the social and economic domain in the Child Opportunity Index 2.0 ZIP Code data.¹⁸ The intelligence quotient (IQ) was also not assessed in our cohort of children. Yet, there is no evidence that amblyopia is related to lower IQ scores. IQ is strongly related to comprehension difficulty.³⁹ If our group of amblyopic children had lower IQ, we would have seen increased fixation duration and increased regressions as these measures are correlated with increased comprehension difficulty⁴⁰ Instead, these measures were similar to controls, with amblyopic children having only increased forward saccades during reading. Because the data were restricted to children who completed grades 1 to 6, the findings are not generalizable to older children and adults. However, data from adults point to reading deficits that persist later in life.⁴¹ Lastly, results with the Worth 4-dot test used to assess suppression must be interpreted with caution. The intensity and visibility of the dots may be affected by differences in room lighting, and the red-green anaglyphic glasses may be dissociative.⁴² However, a child’s fusion status during natural, binocular reading may not be the same as their fusion status measured with the Worth-4 dot test due to its dissociative nature. Thus, the Worth 4-dot limitations likely have no bearing on how fast children read. In our study, most children had bifoveal-macular fusion (23/36; 64%) and there was no significant difference in reading rate between amblyopic children with bifoveal-macular fusion or with peripheral-no fusion.

Reading is essential for academic achievement, and slow reading for comprehension has the potential to impact performance on time standardized tests that are used for school admissions, to assess academic achievement, and determine eligibility for academic tracks in the USA. Parents of amblyopic children may be unaware that their child’s reading is affected by their amblyopia because their child has 20/20 vision in the fellow eye. Whether recovery of normal visual acuity with amblyopia treatment improves reading rate is yet to be determined. Academic accommodations, such as more time for tests, for timed standardized tests typically administered in the USA may need to be considered for children with amblyopia who are slower at reading than controls so that they perform at their maximum potential.