This is the first report of the relation between anisometropia and amblyopia among school-aged members of a Native American tribe with a high prevalence of astigmatism. In agreement with most previous studies,
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 the results of the present study indicated that there was a significant relation between the amount of anisometropia and interocular differences in VA and between the amount of anisometropia and the presence of amblyopia (defined in the present study as an interocular difference in VA of at least 2 logMAR lines). In addition, as previously reported,
6 10 14 16 18 19 20 amblyopia occurred more frequently and at lower amounts of anisometropia in hyperopic compared with myopic anisometropes.
In the present study, the nonastigmatic, hyperopic children with ≥1.00 D of anisometropia showed significantly increased mean IAD. This result is in agreement with the significant increase in mean IAD shown by patients with >1. 00 D of nonastigmatic hyperopic anisometropia in the only other study that compared the magnitude of IAD in anisometropic and nonanisometropic individuals.
18 19 In the present study, the nonastigmatic, myopic children with ≥1.00 D of anisometropia (10/11 of whom had <2.00 D of anisometropia) did not show a significant increase in mean IAD, in comparison with those who had no anisometropia, a result that is in agreement with the previous study’s finding that a significant increase in mean IAD occurred only in nonastigmatic myopes who had anisometropia of >2.00 D.
18 19 In cylinder anisometropes who do not have SE anisometropia, the previous study showed a significant increase in IAD in patients with >1.50 D of anisometropia.
18 19 In contrast, in the present study, although mean IAD was greater in the children with ≥1.50 D of pure cylinder anisometropia than among those in the ISO group
(Fig. 1C) , this difference did not reach statistical significance until the cylinder anisometropia was ≥3.00 D. The lower threshold for amblyopia in the previous study may have resulted from a bias toward a higher prevalence of VA deficits in their relatively small, patient-based sample (
n = 30 with ≥1.50 D of pure cylinder anisometropia; 16 with ≥2.00 D of pure cylinder anisometropia), compared with our larger, school-based sample (
n = 46 with ≥1.50 D; 26 with ≥2.00 D of pure cylinder anisometropia).
In addition to examining anisometropia calculated as interocular differences in sphere and cylinder, we examined anisometropia calculated using two vector-based methods.
39 40 Unlike traditional methods for calculation of anisometropia, vector-based methods take into account interocular differences in axis as well as interocular differences in cylinder magnitude. The method of Thibos et al.
39 calculates interocular differences separately for spherical equivalent (M) and for the Jackson cross-cylinder component (broken into differences along the horizontal/vertical [J0] and oblique [J45] meridians). When we calculated anisometropia in terms of the greater of the two types of meridional differences (J0 versus J45), the magnitude of the IAD
(Fig. 2A)and prevalence of amblyopia
(Fig. 4A)were both related to the amount of anisometropia. However, there was no threshold value at which mean IAD or prevalence of amblyopia differed significantly from the values in the group with no or minimal anisometropia.
In contrast, when anisometropia was calculated with the Harris vector notation,
40 49 50 in which interocular differences are calculated as a single value corresponding to vector difference in three-dimensional space (VDD), results indicated that a significant increase in IAD
(Fig. 2B)and a significant increase in the percentage of children with amblyopia
(Fig. 4B)occurred at a value of 1.41 VDD. This result is particularly important, because it provides the first data on the magnitude of anisometropia that is a risk factor for amblyopia when anisometropia is calculated with a method that incorporates all three components of refractive error: sphere, cylinder, and axis.
It may seem surprising initially that one vector method showed no significant relation between IAD and the amount of anisometropia, whereas the other vector method did. However, this result is most likely related to the fact that the data plotted for the method of Thibos et al.
39 relate only to cylinder and axis anisometropia, whereas the data plotted for the Harris
40 49 50 method reflect anisometropia based on sphere, cylinder, and axis values.
In addition to examining IAD and presence of amblyopia as related to magnitude of anisometropia, we also examined best corrected SA as a function of magnitude of anisometropia. The results indicate that, regardless of whether anisometropia is calculated in terms of clinical notation (spherical equivalent and cylinder), or in terms of either of the two vector methods, significantly reduced best corrected SA occurs in individuals with relatively small amounts of anisometropia. Specifically, a significant reduction in mean SA was found with ≥0.50 D of hyperopic, myopic, or cylindrical anisometropia
(Fig. 5) , a difference between eyes in J0 and/or J45 of ≥0.25 D
(Fig. 6A) , and a difference between eyes of ≥0.71 VDD
(Fig. 6B) . Thus, disruption of best corrected SA occurred at levels of anisometropia that were well below those that put an individual at risk for amblyopia
(Figs. 3 4) . Furthermore, in contrast to results for IAD and amblyopia
(Figs. 1 3) , the magnitude of anisometropia that puts an individual at risk for decreased best corrected SA was identical in the children with hyperopic, myopic, and cylindrical anisometropia
(Fig. 5) .
The SA results of the present study differ substantially from those of the only other study that compared best corrected SA in anisometropes and nonanisometropes.
18 19 In that study, the amount of anisometropia that resulted in decreased best corrected SA was identical with the amount of anisometropia that produced an increase in mean IAD and in prevalence of amblyopia. An important difference between the two studies is that in the present study, SA was measured with a random-dot test (Randot Preschool Stereoacuity Test; Stereo Optics, Inc.), which is free of monocular cues, whereas the previous study used the Titmus stereo test, a non–random-dot test that includes monocular cues that can improve an individual’s ability to detect stereo targets, thereby reducing detection of binocular SA deficits.
This study has a number of strengths. First, it is a large, school-based study, in which approximately 85% of children in grades K-2 and 4 to 6 in schools on the Tohono O’odham reservation were enrolled. Second, the population includes a high percentage of children with high astigmatism (42.7% of the 972 children in the study had astigmatism of ≥1.00 D in one or both eyes; 27.5% had astigmatism of ≥2.00 D), which allowed detailed analysis of the effect of astigmatic anisometropia on prevalence of amblyopia and decreased SA. Third, measurement of refractive error was conducted with cycloplegia and followed a rigorous protocol that included measurement with an unbiased, objective instrument (the Retinomax autorefractor; Righton Manufacturing Co.), followed by verification of autorefractor measurements by retinoscopy and, when possible, by subjective refinement. Fourth, best corrected VA was measured with ETDRS charts, which contain logMAR spacing of optotypes and are the gold standard for assessment of VA in clinical studies of adults.
46 Fifth, best corrected SA was measured with a random dot SA test,
51 which is free of the monocular cues present in non–random-dot stereo tests. Sixth, the subject population included a substantial number of children who had little or no anisometropia, which provided isometropic baseline data concerning IAD, SA, and prevalence of amblyopia in the absence of anisometropia. Seventh, all subjects, even those who did not meet the criteria for prescription of spectacle correction, were tested while wearing glasses providing the best correction. This masked the adults who tested VA and SA from knowing which children had significant refractive error and made testing conditions (wearing of spectacles) equal for all subjects.
A final strength is that the present study provides the first large-sample data relating vector-method calculation of anisometropia to data on IAD, SA, and presence of amblyopia. Because vector methods include cylinder axis in calculations of anisometropia, they provide a more complete description of interocular differences in refractive error than do traditional clinical notation techniques that focus on differences in sphere, spherical equivalent, and/or cylinder. The present results provide the first data indicating the magnitude of vector differences that put children at risk for amblyopia and/or decreased SA.
Despite its strengths, the present study has limitations. First, many of the subjects, especially those with higher amounts of anisometropia, had a history of spectacle wear. Because as little as 12 weeks of spectacle wear can reduce or eliminate amblyopia,
23 31 35 37 it is possible that the prevalence of amblyopia was underestimated and the amount of anisometropia needed to produce amblyopia was overestimated because of the children’s previous glasses wear. However, as shown by the white bars in
Figures 1 2 3 4 5 6 , results for only those children who had no history of glasses wear were similar to those for the group as a whole, although sample sizes of non–glasses-wearing children were small at higher values of anisometropia, which weakens the meaningfulness of these data. In addition, no data were available concerning the children’s compliance with glasses wearing, and therefore, it is possible that the similarity of results of the children with no history of glasses wear to the results of the group as a whole relate to poor compliance with glasses wear.
43
A second limitation relates to likely differences in the variability of recognition acuity versus SA results, which may have contributed to the lesser sensitivity of IAD than of SA to the amount of anisometropia. For recognition acuity testing, children were required to identify all visible letters on the ETDRS chart: first with the right eye, then with the left eye. This procedure took approximately 5 to 10 minutes per child and may have led to inattentiveness, resulting in variability in acuity results. In contrast, the SA test involved binocular testing that required children to identify pictures in six sets of four-plate combinations and could be completed quickly, perhaps resulting in more consistent results for the ISO group than were obtained with recognition acuity testing.
A third limitation is the absence of against-the-rule and oblique axis astigmatism in the subject population. Thus, any conclusions about the effects of astigmatic anisometropia may be applicable only to individuals who have with-the-rule axis orientation in both eyes.
A final limitation is the relatively small number of subjects with myopic anisometropia ≥1.00 D. As a result, we were unable to determine the minimum amount of myopic anisometropia that was associated with significantly increased IAD and amblyopia. However, it was possible to determine the minimum amount of myopic anisometropia that was associated with reduced best corrected SA in this group.
In conclusion, the present study provides data from a school-based population on the amount of interocular refractive error difference that is associated with a significant increase in interocular best corrected recognition acuity difference, and in a reduction in best corrected acuity for random dot stereograms. Results indicate that an increase in interocular difference in best corrected recognition acuity is related to both the amount and the type of refractive error difference between eyes. In addition, for all methods of calculating interocular differences in refractive error, disruption of best corrected random dot SA occurs with smaller interocular refractive error differences than those producing an increase in interocular best corrected recognition acuity differences, suggesting that development of SA is particularly dependent on similarity in refractive error between fellow eyes. Additional research is needed to determine the effect of early and consistent glasses correction on the relation between the amount of anisometropia and best corrected recognition acuity and SA in the school-aged child.
The authors thank the Tohono O'odham Nation, the Indian Oasis/Baboquivari School District, the Bureau of Indian Affairs Office of Indian Education Programs (BIA OIEP, Papago/Pima Agency), the San Xavier Mission School, the parents and children who participated in the study, and our NIH/NEI Data Monitoring and Oversight Committee (Maureen Maguire, PhD [former chair], Robert Hardy, PhD [current chair], Morgan Ashley, Donald Everett, MA, Jonathan Holmes, MD, Cynthia Norris, and Karla Zadnik, OD, PhD).