Influence of accommodation on off-axis refractive errors in myopic eyes

Abstract

Introduction: This study aims to understand off-axis refraction during accommodation and to identify whether the relative hyperopia generally observed in myopic eyes changes with accommodation. Method: Twenty bilateral myopes (18 to 33 years) between −0.50 D and −4.25 D (spherical equivalent) and astigmatism less than 1.25 D participated in this study. A soft contact lens was used to correct refractive error for all measurements. Non-cycloplegic autorefraction was measured at the fovea and 20°, 30°, and 40° eccentricities in the nasal and temporal retina at distances of 2 m, 40 cm, and 30 cm. Results: Peripheral refractive error, relative to central refraction, became less hyperopic with increasing eccentricity and with increasing accommodation. Lag of accommodation increased with accommodation ( p < 0.001) shifting the image-shell backward relative to the retina. In the farther periphery, there was either no change in refractive error or increased myopic shifts with accommodation. Astigmatism increased with eccentricity and significantly increased in the farther eccentricities with accommodation ( p < 0.001). Conclusion: Myopes display hyperopic shifts in the center and near peripheral field during near-viewing, while the farther periphery either remains unshifted or demonstrates a myopic shift. These results are due to the combined effect of lag of accommodation and an increased curvature of field during accommodation.

Introduction

Causes and consequences of myopia

The percentage of myopes in the human population has increased dramatically in the last decades, reaching near epidemic proportions in some countries. In Asian industrial centers (e.g., Hong Kong, Singapore), the prevalence approaches 75% of the population depending on age and occupation (Rosenfield & Gilmartin, 1999). It is estimated that the number of myopes worldwide will increase from 1.6 billion today to 2.5 billion by 2020. Much attention is being devoted to counteracting this problem, as myopia can be associated with economic and pathological consequences such as cataract, glaucoma, choroidal neovascularization, and retinal detachment (Saw, Katz, Schein, Chew, & Chan, 1996). The mechanism underlying the development of myopia in the postnatal years is unknown. However, in humans, both genetic (Kurtz et al., 2007; Mutti, Mitchell, Moeschberger, Jones, & Zadnik, 2002; Pacella et al., 1999; Saw et al., 2005) and environmental factors (Konstantopoulos, Yadegarfar, & Elgohary, 2008; Mutti et al., 2002; Saw et al., 2001) contribute to the disorder. 

Animal models of refractive development suggest that axial growth of the postnatal eye, and its refractive state, can be highly influenced by altering the optical demand. Infant non-human primates reared wearing powered spectacle lenses (Graham & Judge, 1999; Hung, Crawford, & Smith, 1995) or soft contact lenses (Whatham & Judge, 2001) develop myopic or hyperopic refractive errors in response to hyperopic and myopic defocus, respectively. This finding, both in primates and other animal species such as chicks (Schaeffel, Glasser, & Howland, 1988; Wildsoet & Wallman, 1995) and tree shrews (Shaikh, Siegwart, & Norton, 1999), provides compelling evidence that the human eye is similarly capable of modifying its refractive state in response to optical experience and minimizing its degree of uncorrected refractive error during the postnatal years before adulthood. Despite this knowledge, as well as evidence for active emmetropization in human infants and children, it is not known why children and adolescents develop myopia. 

Relationship between near-work and myopia

One environmental risk factor that has long been associated with the development of myopia is prolonged near-work (Cohn, 1886). Yet the causal link between near-viewing and myopia has so far proved elusive. One of the more explored possible links is that of accommodation for near viewing. In humans, this adjustment is brought about by the crystalline lens becoming more spherical. Early mechanical theories for accommodation-induced myopia have lost support, in favor of defocus-induced mechanisms such as under-accommodation during near viewing. Hyperopic defocus during near viewing (lag of accommodation) is greater in myopic than non-myopic individuals (Gwiazda, Thorn, Bauer, & Held, 1993; McBrien & Millodot, 1986), which might provide the initial stimulus for myopia. However, a recent large clinical trial in children found only a small (∼0.25 D) retardation of myopia progression when hyperopic defocus produced by the near lag of accommodation was alleviated using progressive spectacle lenses (Gwiazda et al., 2003). Furthermore, Mutti et al. (2006) found that a lag of accommodation accompanied the development of myopia, rather than preceding it, suggesting that the lag of accommodation is a consequence and not a cause of myopia. 

Peripheral refraction and the development of myopia

Recent studies have shown that the peripheral refraction of the eye might play a role in the progression or the onset of myopia. Relative to the fovea, it was found that a more hyperopic refractive error exists in the horizontal peripheral visual field than exists in emmetropic or hyperopic eyes during distance viewing (Atchison, Pritchard, & Schmid, 2006; Mutti, Sholtz, Friedman, & Zadnik, 2000; Seidemann, Schaeffel, Guirao, Lopez-Gil, & Artal, 2002). These studies suggest peripheral hyperopic defocus encourages excessive axial eye growth and the progression of foveal myopia. This hypothesis was further supported in animal studies involving macaque monkeys. Depriving the peripheral retina of form vision in infant monkeys produces foveal axial myopia (Smith, Kee, Ramamirtham, Qiao-Grider, & Hung, 2005) while laser ablation of the fovea does not prevent emmetropization in infant monkeys (Smith et al., 2007), indicating that the peripheral retina is a sufficient, and perhaps dominant, regulator of the emmetropization process. 

In contrast to current knowledge of peripheral refractive errors during distance viewing, little is known about the patterns of off-axis refraction in myopic eyes during accommodation. Peripheral hyperopia may increase during accommodation and thus act as the missing link between near-work and the development of myopia. Ocular shape becomes more prolate during accommodation (Walker & Mutti, 2002), which might be associated with a hyperopic shift in peripheral refractive error. In contrast to this putative hyperopic shift in the periphery during accommodation secondary to ocular shape changes, the retinal image-shell undergoes peripheral myopic shift due to changes in the Petzval radius of curvature (Smith, Millodot, & McBrien, 1988). 

A recent study found no hyperopic or myopic shift in peripheral refraction out to 30° when changing fixation from distance to 40 cm (Calver, Radhakrishnan, Osuobeni, & O'Leary, 2007). The aim of the present study is to expand upon the above results and further understand the patterns of off-axis refraction present during accommodation in myopic eyes by measuring peripheral refraction at more than one level of accommodation and 30° of eccentricity (i.e. angle of field relative to central/foveal vision). 

As myopes typically demonstrate a lag of accommodation during near viewing, it is unclear what refraction profile across the retina will result from the interaction between under-accommodation and increased curvature of field. An additional goal is to discover this relationship. 

Methods

The study protocol followed the tenets of the Declaration of Helsinki and was approved by the local institutional human research ethics committee prior to study initiation. 

Subjects and contact lens corrections

All subjects were recruited using an advertisement circulated via email to staff and eligible subjects on the database of IER (Institute for Eye Research, Sydney, Australia). 

Subjects were pre-screened for suitability using routine eye examination, which included subjective refraction, visual acuity determination, and general eye health examination with a biomicroscope and direct ophthalmoscope. Healthy subjects with good visual acuity (best-corrected acuity better than 6/7.5), normal binocular function (no manifest strabismus or amblyopia), and bilateral mild to moderate myopia were recruited. A commonly adopted criterion for myopia of spherical equivalent refraction of less than or equal to −0.50 D (Ip et al., 2008) was used. Twenty subjects, (18 to 33 years) predominantly female (75%), were fitted with spherical soft contact lenses, and measurements were made of peripheral refractive errors under differing degrees of accommodative demand, using an autorefractor. The age range was chosen to ensure that at least 3 D to 4 D of accommodation could be achieved. Subjects had bilateral refractive errors (as determined by non-cyclopleged, subjective refraction employing the maximal plus refraction technique) between −0.50 D and −4.25 D (spherical equivalent) and astigmatism less than or equal to 1.25 D. Mean refractive error for this group in the tested (right) eye was −2.17 D ± 1.18 D (SD) with range −0.50 D to −4.125 D. Eyes were assessed and deemed suitable for wearing soft contact lenses for 1 hour. All subjects' eyes were fitted with commercially available spherical soft contact lenses (Acuvue 2 Contact Lenses, etafilcon 2, 58% water content, Vistakon/Johnson & Johnson, USA) to permit best on-axis distance vision and to ensure each subject used the same accommodative effort during near viewing. 

The optical zone diameter of the contact lens was 8 mm. Since the effective aperture size of measurement of the Shin-Nippon is approximately 3 mm, measurement of peripheral refraction, even at 45° eccentricity, is not influenced by the peripheral zones of the contact lens. 

Contact lens power was determined from conventional subjective refraction performed at a distance of 6 m from an LCD screen displaying standardized optotypes. The dioptric value of the correcting contact lens was defined as the spherical equivalent value of the subjective refraction using the criterion of maximum plus to best visual acuity. All eyes attained acuities of at least 6/9 through the best equivalent spherical contact lens. 

Instrumentation

Instrument and apparatus for near fixation targets are shown in Figures 1 and 2. The Shin-Nippon NVision-K5001 autorefractor (Shin-Nippon, Tokyo, Japan) was used to measure central and peripheral refractive error during distance and near viewing. This instrument has been reported to have good accuracy and reproducibility (Davies, Mallen, Wolffsohn, & Gilmartin, 2003) and has been used in previous studies to provide objective measurements of peripheral refraction (Atchison, 2003; Atchison et al., 2006; Berntsen, Mutti, & Zadnik, 2008; Radhakrishnan & Charman, 2008) even in substantially aberrated eyes (Charman, Mountford, Atchison, & Markwell, 2006; Ma, Atchison, & Charman, 2005) and for objective measurements of accommodation (Win-Hall & Glasser, 2008; Win-Hall, Ostrin, Kasthurirangan, & Glasser, 2007). 

Figure 1

 

Autorefractor Shin-Nippon NVision K-5001 with custom near fixation targets fixed in position on top.

Figure 2

 

Details of custom fixation targets (top-view). The fixation targets can be moved in/out (towards/away from the eye) to allow measurements at 30 cm and 40 cm.

The autorefractor was modified (using small laser diodes emitting visible red light) to enable fixation points to be presented at different eccentricities, based on the optical center of the instrument. The laser diodes were mounted on the same stage as the near target fixed to the Shin-Nippon. These laser diodes projected a very fine beam at their set angles of eccentricity from the instrument into the subject's visual field. Upon intersection with the projection surface, a 5-mm spot was produced. Illumination of the appropriate laser diode for distance focus and fixation was computer controlled. As the laser diodes were mechanically coupled to the near target and refractometer, the same angles of eccentricity were assured even with movement of the instrument. For far distance measurements, fixation targets were provided at the wall of the clinic room at 0°, 20°, 30°, and 40° viewing on both left- and right-hand sides. Instrument-to-fixation target distance was 2.0 to 2.5 m (referred to throughout this article as 2 m). 

For refraction in near vision, a custom-made fixation target, which allows off-axis refraction, was developed and mounted on the Shin-Nippon autorefractor. Fixation targets consisted of two differently sized high contrast standard letters (E) directly above each other. One letter was developed for 40 cm (actual size 0.872 mm) and the other letter for 30 cm (actual size 0.654 mm) viewing distance. Each letter was equivalent to a visual acuity of 6/9 (20/30) to ensure accommodation stability. As all our subjects were young myopes with good ocular health, we expected all of them to achieve the minimum required visual acuity with correcting contact lenses. Each fixation target could be moved to allow measurements at 30- and 40-cm distance. There were seven near fixation charts—one for central fixation and two for 20°, 30°, and 40° for measurements on the nasal and temporal peripheral retina, respectively. We compared, in a feasibility study, autorefractor measurements of central refraction at 2 m for the two different target types to see whether similar refractive errors could be obtained at distance viewing for the laser diode and the letter target. The difference found was less than 0.125 D (mean difference being 0.07 D and standard deviation ±0.15 D) and was not statistically significant ( p = 0.297). Hence, possible differences in refraction measures at 2 m that might result from different accommodation responses to the different targets were negligible. 

Central and peripheral refractive error measurement

Autorefraction was performed centrally and in the horizontal visual field at 20°, 30°, and 40° from the visual axis on each subject's right eye in a pseudo-randomized order. Automated target presentation and order of fixation targets (i.e. angles of eccentricity) as well as data acquisition were managed using author's custom-designed software (in C language). A pseudo-random sequence for each eye at each target distance was generated within a subprogram using a random-number generator function with automatic seeding. For distance, the required laser-target was automatically illuminated and selected. For near targets, the appropriate next target was indicated by the program to the operator and subject. Target distance was not randomized. However, in order to control for any order effect, a counterbalancing strategy was used. The order of measurement of target distances for half of the subjects, selected at random, progressed from 2 m to 30 cm while the remaining half were measured at 30 cm first, then 40 cm, and finally 2 m. 

The non-measured left eye was not covered. Thus fixation targets were viewed binocularly. All subjects maintained accurate binocular fixation to distance and near targets during the screening process. As no subjects had manifest strabismus, it was assumed that accurate binocular fixation was achieved to each viewing target, and no fixation errors would be outside the normal range of fixation disparities in healthy populations (e.g., it has been reported that 90% of subjects measured had fixation disparities between ±22 arcmin; Zaroff, Knutelska, & Frumkes, 2003). 

Measurements at all distances and eccentricities were conducted under ambient (photopic) room lighting conditions between 150 and 160 lux. According to the manufacturer's reference manual, the Shin-Nippon is capable of reliable measurements through pupil sizes as small as 2.3 mm. Five measurements were taken from each location, each fixation distance, and each subject. 

It has been reported that there are only small or no differences between the head and eye movement methods for measuring eccentric refraction using an autorefractor (Radhakrishnan & Charman, 2008; Seidemann et al., 2002). However, in this study, eye movement needed to be controlled and minimized for measuring eccentric refraction. A phenomenon well known to contact lens practitioner is that of contact lens “lag.” Due to the interplay of mechanical actions of the contact lens, tear-film, cornea, and eyelid, when an eye fitted with a contact lens undergoes version, the contact lens tends to decenter relative to the cornea in a direction opposite to the direction of version (to the observer, it appears as if the contact lens is “lagging” behind the eye's motion). As the amount of contact lens lag differs from eye to eye, this effect needs to be eliminated to ensure measurements at a given eccentricity are effected through the same point on the contact lens for all eyes. Thus, the subject was asked to maintain straight-ahead eye fixation with head movement in the direction of the fixation target. In addition to ensuring measurements were made through the optical zone of the contact lens, this strategy also eliminates any variability in power change resulting from the combination of contact lens aberrations and lens decentration. While the subject looked at the fixation targets foveally, the autorefractor beam entered the eye at an eccentric angle equal to the angle of the subject's eye gaze. 

Statistical analysis

All data were stored in relational databases and imported into SPSS 15.0 (Chicago, USA) for statistical analyses. Sphero-cylindrical refractive errors were first converted into power vector form ( M, J ₁₈₀, and J ₄₅), according to conventional formulae for astigmatic decomposition prior to analysis, to allow grouping of data (Thibos, Wheeler, & Horner, 1997). 

Relative peripheral refractive error (RPRE) was also calculated. Foveal refraction components were subtracted from the corresponding peripheral measurements to quantify change with respect to the foveal values according to Equation 1. This parameter was calculated for each viewing distance so that patterns of peripheral refractive error relative to central refractive error, at each accommodation level could be determined.  

$\mathrm{R}\mathrm{P}\mathrm{R}{\mathrm{E}}_{i.a}=M_{i.a}-M_{0.a},$

(1)

where M is the mean-sphere refractive error measured at the viewing distance denoted by suffix .a (where .a represents 2, 40, or 30 cm viewing) and at the eccentricity denoted by suffix i (where i represents −40°, −30°, −20°, central, 20°, 30°, or 40°). Suffix 0 denotes measurement taken at the central/foveal direction. 

The above variables were analyzed between accommodation distances, as a function of eccentricity (main effects), and also across accommodation levels as a function of eccentricity (interaction effect), using a linear mixed model ANOVA. Post hoc multiple comparisons were adjusted using the Bonferroni correction. A type I error rate of less than or equal to 5% ( p ≤ 0.05) was considered to be statistically significant. 

Results

Results are summarized in Table 1. An ANOVA table of main and interaction effects is given in Table 2. 

Table 1

 

Results for mean sphere M, horizontal and oblique astigmatic components of power vectors J ₁₈₀ and J ₄₅, respectively, and relative peripheral refractive error RPRE (mean ± SD) for 20 eyes at 7 eccentricities and three viewing distances.

		Nasal retina			Central	Temporal retina
		−40°	−30°	−20°	0°	20°	30°	40°
2 m	M	0.91 ± 0.82	0.54 ± 0.70	0.36 ± 0.58	0.10 ± 0.31	−0.18 ± 0.39	0.04 ± 0.69	0.31 ± 1.12
	J 180	−1.15 ± 0.44	−0.64 ± 0.32	−0.24 ± 0.31	−0.01 ± 0.22	−0.63 ± 0.26	−1.23 ± 0.35	−1.88 ± 0.51
	J 45	−0.28 ± 0.29	−0.27 ± 0.19	−0.23 ± 0.19	−0.15 ± 0.12	0.01 ± 0.21	0.00 ± 0.30	−0.06 ± 0.48
	RPRE	0.81 ± 0.84	0.44 ± 0.72	0.26 ± 0.55	0.00 ± 0.00	−0.28 ± 0.46	−0.07 ± 0.80	0.21 ± 1.25
40 cm	M	−1.43 ± 1.10	−1.65 ± 0.77	−1.71 ± 0.51	−1.70 ± 0.24	−1.94 ± 0.46	−1.94 ± 0.75	−1.89 ± 1.26
	J 180	−1.40 ± 0.50	−0.65 ± 0.35	−0.23 ± 0.37	0.02 ± 0.23	−0.66 ± 0.35	−1.44 ± 0.49	−2.36 ± 0.57
	J 45	−0.17 ± 0.24	−0.18 ± 0.23	−0.15 ± 0.26	−0.07 ± 0.14	−0.02 ± 0.21	−0.06 ± 0.40	−0.15 ± 0.45
	RPRE	0.27 ± 1.11	0.06 ± 0.75	−0.01 ± 0.47	0.00 ± 0.00	−0.24 ± 0.49	−0.24 ± 0.83	−0.19 ± 1.33
30 cm	M	−2.34 ± 1.02	−2.41 ± 0.76	−2.35 ± 0.49	−2.41 ± 0.29	−2.69 ± 0.39	−2.75 ± 0.70	−2.79 ± 1.25
	J 180	−1.46 ± 0.49	−0.69 ± 0.37	−0.22 ± 0.37	0.06 ± 0.21	−0.65 ± 0.36	−1.42 ± 0.44	−2.48 ± 0.58
	J 45	−0.26 ± 0.22	−0.23 ± 0.20	−0.12 ± 0.23	−0.04 ± 0.17	0.02 ± 0.21	0.00 ± 0.25	−0.11 ± 0.41
	RPRE	0.07 ± 1.06	0.00 ± 0.78	0.05 ± 0.42	0.00 ± 0.00	−0.28 ± 0.46	−0.34 ± 0.75	−0.38 ± 1.31

Table 2

 

ANOVA table of main effects and interactions for mean sphere M, horizontal and oblique astigmatic components of power vectors J ₁₈₀ and J ₄₅, respectively, and relative peripheral refractive error RPRE, showing F and p.

	Variables
	M	J 180	J 45	RPRE
Main effects
Distance	753.4; <0.001	12.88; <0.001	0.95; 0.39	11.36; <0.001
Eccentricity (viewing angle)	7.93; <0.001	478.4; <0.001	10.80; <0.001	9.33; <0.001
Interaction
Distance × Eccentricity	1.14; 0.33	4.23; <0.001	0.90; 0.55	1.02; 0.43

Power vector M (spherical equivalent)

Mean sphere, M (spherical equivalent of refractive error), from autorefraction is shown in Figure 3 for all three accommodation levels as a function of eccentricity. Negative values exist for the 40-cm and 30-cm viewing distances since accommodation introduces a relative negative (myopic) state as measured by the autorefractor. For far distance viewing, M was more hyperopic relative to foveal refractive error at all eccentricities except 20° and 30° on the temporal retinal side. 

Figure 3

 

Power vector M (spherical equivalent) with eccentricity and three levels of accommodation. Negative eccentricities denote measurements in the nasal retina (temporal visual field). Error bars indicate ± SEM.

M varied significantly with eccentricity across accommodation levels ( p < 0.001) and as a function of eccentricity at each viewing distance ( p < 0.001). This change in M with eccentricity was similar across the different levels of accommodation ( p = 0.33). 

Relative peripheral refractive error (RPRE)

Relative peripheral refractive error (RPRE), as a function of eccentricity and level of accommodation, is shown in Figure 4. For far distance viewing, RPRE became more myopic, relative to the fovea, at 20° and 30° in the temporal retina but more hyperopic at all other eccentricities measured ( p < 0.001). Peripheral refractive error became progressively and significantly ( p < 0.001) more myopic/less hyperopic relative to central refractive error as accommodation increased. Substantial asymmetry was observed with the temporal retinal image at 40° being more hyperopic by 0.60 D than the corresponding nasal periphery. At 2 m, corresponding temporal and nasal eccentricities were statistically significantly different ( p = 0.022, 0.012, 0.001 for 40°, 30°, and 20°, respectively, by paired t-tests). While asymmetry was also observed at 40-cm and 30-cm distances, no corresponding eccentricities were statistically significantly different. 

Figure 4

 

RPRE with eccentricity and three levels of accommodation. Negative eccentricities denote measurements in the nasal retina (temporal visual field). Error bars indicate ± SEM. X-values have been displaced slightly for clarity.

Power vector J ₁₈₀

No effect of accommodation on the decrease in J ₁₈₀ with eccentricity below 30° was observed. For larger eccentricities (for angles of field of 30° and 40° temporal retina, and 40° nasal retina), there was a significant decrease in J ₁₈₀ with increasing accommodation ( p < 0.001). Magnitude of the J ₁₈₀ (vertical/horizontal astigmatism) component ( Figure 5) exhibited temporal-nasal asymmetry with a greater decrease (increasing against-the-rule astigmatism) on the temporal retinal side. At 40° eccentricity, J ₁₈₀ for the temporal retinal image was greater than the nasal counterpart by 0.73 D. At all viewing distances, corresponding temporal and nasal eccentricities were statistically significantly different (all p < 0.0001 by paired t-tests). 

Figure 5

 

Variation in power vector component J ₁₈₀ (vertical/horizontal astigmatism) with eccentricity and three levels of accommodation. Negative eccentricities denote measurements in the nasal retina (temporal visual field). Error bars indicate ± SEM. X-values have been displaced slightly for clarity.

Power vector J ₄₅

Variation in the J ₄₅ (oblique astigmatism) component as a function of eccentricity and accommodation is shown in Figure 6. There was a small (<0.50 D) but significant decrease in J ₄₅ with eccentricity (repeated measures ANOVA, p < 0.001). However, there was no effect of accommodation on J ₄₅ ( p = 0.39), nor was there any interaction between accommodation level and the change in J ₄₅ with eccentricity ( p = 0.55). While magnitudes were small, temporal–nasal asymmetry was observed in J ₄₅ with temporal retina image values being more positive (e.g., 0.274 D at 30° eccentricity and at 2 m). Asymmetry in J ₄₅ reached significance sporadically and occurred only at the lower eccentricities ( p = 0.003 and p < 0.0001 for 30° and 20° at 2 m, p = 0.034 for 20° at 40 cm, and p = 0.002 and 0.008 for 30° and 20° at 30 cm). 

Figure 6

 

Variation in power vector component J ₄₅ (oblique astigmatism) with eccentricity and at three levels of accommodation. Negative eccentricities denote measurements in the nasal retina (temporal visual field). Error bars indicate ± SEM. X-values have been displaced slightly for clarity.

Lag of accommodation

Mean (± SEM) of accommodation for all 20 myopic eyes (determined from the on-axis refractive error measurement) was 1.70 ± 0.24 D for the 40-cm near fixation target and 2.40 ± 0.29 D for the 30-cm near fixation target. Consequently, lag of accommodation, defined as the difference between the accommodative demand and the accommodative response, resulted in mean hyperopic defocus of 0.80 D for 40 cm and 0.93 D for the 30-cm target. 

Discussion

Validity of relative peripheral refraction findings

To our knowledge, there are no reports in the literature detailing the accuracy of the Shin-Nippon refractometer for peripheral refraction measurements at large angles of eccentricity in life human eyes. Such a study would need to overcome significant technical challenges. As demonstrated by theoretical modelling (Charman & Atchison, 2008), the accuracy of peripheral refraction measurements may be adversely influenced by the presence of significant off-axis aberrations such as radial astigmatism and coma as well as the elliptical shape of the pupil especially at large angles of eccentricity. However, there have been empirical studies (Atchison, 2003; Berntsen et al., 2008) showing good comparability between the Shin-Nippon refractometer and the COAS aberrometer, the latter of which can be used to analyze, separately, the effect of the various aberration components. Further, the Shin-Nippon refractometer has previously been shown to produce reliable objective measurements of peripheral refraction (Atchison, 2003; Atchison et al., 2006; Berntsen et al., 2008; Radhakrishnan & Charman, 2008) even in substantially aberrated eyes (Charman et al., 2006; Ma et al., 2005). 

In the present study, the key scientific objective was not to quantify the absolute peripheral focal positions, but to determine how they change, relative to central measures, during accommodation. It is reasonable to assume that for the same peripheral angle of eccentricity, any small bias caused by the effects modelled by Charman and Atchison (2008) should differ negligibly between different states of accommodation. Hence, findings of this study, with regard to the relationship between peripheral and central refraction under differing levels of accommodation should remain valid. 

Curvature and radial astigmatism of the eye

Relative peripheral refraction (spherical equivalent) changed with increasing amounts of accommodation ( Figure 4), producing a relative myopic shift in peripheral image points upon accommodation. It is interesting to speculate on the source of this myopic shift with accommodation. From an optical design stand point, the change in peripheral refraction could be attributable to an increase in the curvature of field of the eye on accommodation. That is, the image surface representing the mean sphere power increased in curvature (decrease in radius). An increase in curvature of field with accommodation has been reported previously (Smith et al., 1988). Smith and colleagues suggest that the retinal image-shell would undergo peripheral myopic shift due to changes in the Petzval radius of curvature. In the present study, accommodation caused increasing negative values in the peripheral refraction, rendering it relatively more myopic. 

In addition to increasing the curvature of field, accommodation also increased peripheral astigmatic refractive error, which was primarily in the horizontal/vertical astigmatism ( J ₁₈₀) component ( Figure 5), increasing the amount of against-the-rule astigmatism. This increase in peripheral astigmatism with increasing accommodation agrees with previous reports (Smith et al., 1988). 

Both findings, of increased field curvature and against-the-rule astigmatism, are consistent with theories pertaining to off-axis optical aberrations. In the eye, third-order aberration theory relating to astigmatism and curvature of field (Conrady, 1929) predicts that as eccentricity increases, the increasing radial astigmatism aberration should mimic increasing amounts of against-the-rule astigmatism. Our result showing increasingly negative J₁₈₀ with eccentricity is consistent with this prediction. Further, the change in peripheral J₁₈₀ relative to central J₁₈₀ provides an estimate of the amount of radial astigmatism incurred at each eccentricity. 

Off-axis J ₁₈₀ astigmatism (against-the-rule) increased in the farther periphery (30° temporal retina and 40° nasal and temporal retina) with accommodation to near targets ( Figure 5) in our myopic subjects. However, the central retina and near periphery showed no change. It has not been determined whether increased peripheral astigmatism during near viewing could influence emmetropization. Research using animal models has shown that on-axis astigmatism imposed on infant animals can disturb eye growth in a quasi-predictable way (Kee, Hung, Qiao-Grider, Roorda, & Smith, 2004). However, imposed astigmatism does not appear to prevent emmetropization taking place (McLean & Wallman, 2003). All human eyes, regardless of their central, on-axis, refractive state, have peripheral astigmatism that increases in magnitude with eccentricity. The orientation of this peripheral astigmatism varies according to the meridian of the eye that is measured. No research has yet shown to our knowledge, in either human studies or animal models, that manipulating the orientation or magnitude of peripheral astigmatism alone can influence central refractive error. However, our finding that astigmatism in the farther periphery increases with accommodation adds to the growing list of ocular and refractive parameters that deserve further investigation for their potential causal relation with the development of myopia. 

Does the eye change shape during near viewing?

Mechanically induced ocular, and consequently retinal, shape changes cannot be ruled out as an influencing factor in increased peripheral myopia with accommodation. At least one report has suggested that the axial length of the eye changes during accommodation (Walker & Mutti, 2002). This might be a result of a range of mechanisms including ciliary muscle contraction during accommodation or increased tension in extra-ocular muscles during convergence (e.g., Greene, 1980) Assuming constancy of eye volume or scleral surface area during accommodation, such a change implies a concomitant change in the shape factor of the retinal surface which, in turn, can alter the peripheral refraction. The relative contributions of mechanical and optical factors to the change in peripheral refraction with accommodation need to be elucidated in future research. 

The effect on retinal imagery remains the same regardless of the cause of the change in peripheral refraction on accommodation. Hence, the relevance of the findings of this study to understanding the contribution of near focus to myopia development is unaltered. 

Asymmetry and change with accommodation

Asymmetry of peripheral refraction has been observed since early investigations (Ferree, Rand, & Hardy, 1932; Millodot, 1981; Rempt, Hoogerheide, & Hoogenboom, 1971) and has also been observed in recent studies using the Shin-Nippon autorefractor (Atchison et al., 2006; Berntsen et al., 2008). Asymmetry is typically observed with the temporal retinal image (nasal field) being relatively more hyperopic. Further, asymmetry in peripheral refraction is accompanied by greater amounts of astigmatism for the temporal than the nasal retina (Atchison et al., 2006; Berntsen et al., 2008; Millodot, 1981; Seidemann et al., 2002). Our results for 2-m viewing distance are in agreement with the above findings. 

While the origin of the asymmetry in peripheral refraction has not been identified conclusively, one possible explanation (Berntsen et al., 2008) may be found in the asymmetry of the optical system of the eye. 

More specific to the objectives of this study, our results suggest that accommodating for near viewing tends to decrease the nasal–temporal asymmetry in mean-sphere refraction but increase asymmetry in vertical astigmatism J ₁₈₀. At 40° eccentricity, nasal–temporal asymmetry in M changed from 0.60 D at 2 m viewing distance to 0.462 D and 0.447 D at 40 cm and 30 cm viewing distances, respectively. In contrast to this decrease, J ₁₈₀ increased from 0.73 D to 0.96 D and 1.02 D when accommodation changed from 2 m to 40 cm and 30 cm, respectively. These findings suggest that the changes in asymmetry in M and J ₁₈₀ brought about by accommodation have different origins from those associated with asymmetry in distance peripheral refraction. Further studies are required to identify other contributing factors. 

Changes in the dioptric image-shell with accommodation

Since the optical condition of the subject remained the same throughout measurements—with the eye fixating on a central target and no change in central or peripheral optics (e.g., the same contact lens is worn throughout by an eye), it is reasonable to assume that the lag of accommodation at each accommodation level is the same for all eccentricities (including central). From this, the dioptric image-shell for each viewing distance can be calculated. This is shown in Figure 7. 

Figure 7

 

The dioptric image-shell, showing the combined effects of increasing curvature of field and lag of accommodation with viewing distance. The position of the retina is represented by the y = 0 line. Negative values indicate a myopic defocus. Negative eccentricities denote measurements in the nasal retina. Error bars indicate ± SEM. X-values displaced slightly for clarity.

It can be seen from this graph that with decreasing viewing distance the amount of hyperopic defocus at the fovea due to the lag of accommodation increases. Moreover, the amount of hyperopic defocus in the near periphery also increases with decreasing viewing distance. This occurs even though the image-shell is becoming more curved due to the increasing curvature of field and perhaps ocular (and retinal) shape changes that occurs with accommodation. In absolute terms, central and near peripheral refraction became hyperopic ( Figure 7), due to the central accommodative lag. Farther in the periphery, however, the combined effect of curvature of field and accommodative lag at different viewing distances make the absolute level of defocus either unchanged or more myopic (see 30° and 40° nasal retina and 40° temporal retina in Figure 7 with accommodation). In the presence of accommodative lag, there is no point on the image surface at any eccentricity or distance of viewing which is myopic in absolute terms in this study. Thus, there is interplay between curvature of field, any ocular shape changes and accommodative lag in bringing about the net peripheral refraction at any given distance of accommodation and eccentricity. Naturally, this interplay is dependent on the individual's lag of accommodation which is modulated by numerous factors including the fineness of details of the object viewed (we have used a relatively coarse target in this study) both centrally and peripherally. In addition, in this study, the contact lens correction was based on sphero-cylindrical refraction carried out at 6 m employing the criterion of maximum plus power to best visual acuity. The difference between this measurement condition and those used during the study proper may explain the relatively large lag of accommodation measured. It is clear that in determining the etiology of myopia progression, we cannot analyze single factors in isolation. 

Can near viewing promote myopia development?

The present study indicates for myopes that hyperopic defocus (power vector M) generally exists in the horizontal retinal periphery during distance viewing ( Figures 3 and 4). This agrees with previously published data (Atchison et al., 2006; Mutti et al., 2000; Seidemann et al., 2002) and is attributed to the more prolate (Mutti et al., 2000) or less oblate (Atchison et al., 2005) shape of the myopic eye in comparison to emmetropic and hyperopic eyes. However, this peripheral hyperopia did not increase per se as accommodation increased during near viewing, due to the increasing curvature of field. Yet neither did refractive error at all tested eccentricities become more myopic, due to the hyperopic shift produced by the lag of accommodation. The result was that near viewing increased hyperopic defocus at the fovea and in the near periphery, while the farther periphery either remained unchanged or underwent a minor myopic shift. Our results agree with those from previous findings (Calver et al., 2007) and confirm that changes in the retinal periphery upon accommodation in myopes are unlikely to be more than 1 D (for low myopes of less than or equal to −3.50 D). Although the image-shells (Figure 7) are intersecting in the farther periphery, it could be that the near periphery gives the greater impulse for eye growth—which remains hyperopic relative to the fovea at all levels of accommodation tested (Figure 7). Peripheral hyperopic defocus has recently been reported to be present before the development of myopia in a population of schoolchildren (Mutti et al., 2007). Thus, it is a potential indicator of future development or progression of myopia. 

Impact of accommodative peripheral refraction on vision correction

Given the relationship between relative peripheral refraction and central refractive state, there have been recent suggestions that the progression of myopia may be moderated by the use of ophthalmic corrective devices that control the relative peripheral refraction (Smith, Greeman, Ho, & Holden, 2006). In lieu of studies demonstrating the efficacy of such an approach, our results suggest that any device designed to produce a relatively more myopic peripheral refractive state at distance focus is likely to maintain (or slightly further increase) the relative myopic peripheral refraction even when the eye is focused for near viewing. However, as discussed in a previous section, for individual eyes and viewing conditions, the amount of accommodative lag must be taken into account. In this study, lag of accommodation was measured with the eye viewing through conventional soft contact lenses. Ophthalmic devices that modify peripheral refractive states may, in doing so, alter accommodative stimulus and modify accommodative lag. Whether peripheral refraction during near viewing display absolute myopia or hyperopia in any myopic individual will be determined by the level of accommodation, including lag (or lead) of accommodation. A large central lag of accommodation will create central hyperopia as well as peripheral hyperopia, even though peripheral refraction will be more myopic than central refractive error. Absolute peripheral myopia will result if there is a lead of accommodation or a small or absent lag of accommodation. Cleary, further studies are required to understand the interplay between these factors and their implications for myopia treatment strategies. 

Conclusions

Myopes display hyperopic shifts in the center and near peripheral field during near-viewing, while the farther periphery either remains unshifted or demonstrates a myopic shift. The net peripheral refractive state depends on the interplay between the change in curvature of field during accommodation and the amount of accommodative lag. 

Acknowledgments

Financial support for this study was received from the Zentralverband der Augenoptiker (ZVA), the Deutsch Akademischer Austtausch Dienst (DAAD), and the Institute for Eye Research. This work was carried out at the clinics and laboratories of the Institute for Eye Research, Sydney, Australia, as part of Frederick Zimmermann's Practicum Study from the University of Applied Sciences, Berlin, Germany. The authors express their gratitude to Dr. Judith Flanagan (Institute for Eye Research) for her helpful suggestions on the manuscript. 

Commercial relationships: none. 

Corresponding author: Arthur Ho. 

Email: [e-mail]. 

Address: Institute for Eye Research, Level 4 Rupert Myers Building, UNSW Sydney, NSW, 2052, Australia.