Peripheral refraction for distance and near vision in emmetropes and myopes

Abstract

We investigated the relationship between myopia and peripheral refraction for distance and near vision by measuring peripheral refractive errors in 10 myopic and 10 emmetropic participants at viewing distances of 2.5 and 0.4 m. Measurements were made at the fovea, and at eccentricities of 10°, 20° and 30° in the temporal and nasal hemispheres of the horizontal visual field. Our results showed that peripheral astigmatism increased with increasing eccentricity, but there was no significant difference between refractive error groups except at 30° eccentricity in the temporal retina. Considering the Mean Spherical Equivalent errors, emmetropes became relatively myopic at peripheral eccentricities, but there was little change in myopes. The effect of viewing distance on astigmatism or Mean Spherical Equivalent error was not significant. Our results do not support the view that myopia is associated with changes in peripheral refraction during distance or near vision.

Introduction

Myopia is an increasingly common condition affecting a large number of people worldwide (Grosvenor, 1996) and is associated with ocular pathology including maculopathy, retinal detachment and glaucoma (Saw et al., 2005). Although its aetiology is not completely understood, there is good evidence that the development of myopia is affected by the nature and quality of the retinal image. Experimental work has shown that myopia develops when retinal image quality is impaired either by form deprivation (for example, by pathology or suturing of the eyelids: Norton, 1999) or by optical defocus. Retinal image blur brought about by the introduction of positive or astigmatic defocus (Kee et al., 2004) leads to the development of myopia in monkeys and it has also been suggested by some studies that the more subtle effects of ocular aberrations play a role in the development of myopia in humans (Collins et al., 1995; He et al., 2002; Paquin et al., 2002; Kirwan et al., 2006).

Most investigations of the effects of image degradation on myopia have concentrated on the fovea but recent work has found that degradation of the peripheral retinal image, in the presence of a clear foveal image, also leads to myopia. Kee et al. (2002) found that hyperopic defocus in the retinal periphery was associated with axial myopia in monkeys, and Smith et al. (2005) found that by degrading the quality of the peripheral retinal images of monkeys using diffusing lenses, they were able to induce myopia. Once the diffusers were removed, there was a regression to emmetropia. This suggests that peripheral retinal image quality may also be important in myopia development. The vulnerability of the peripheral retina to defocus is perhaps not surprising. Although resolution is lower in the retinal periphery than at the fovea, other peripheral visual functions such as detection acuity (Artal et al., 1995) are highly developed and can be compromised by peripheral defocus (Wang et al., 1997). It is also known that local regions of the retina have the capacity to regulate local eye growth in chicks (Wallman et al., 1987). Although peripheral retinal images are affected by the same set of aberrations as foveal images, their magnitudes are different: coma and astigmatism become more significant when light enters the eye eccentrically and, although coma is not easy to measure, many investigators have studied the nature of peripheral refraction for distance vision. Early studies by Ferree et al. (1931) identified characteristic types of peripheral astigmatic profiles, and peripheral refraction has also been measured as a function of refractive error (Millodot, 1981; Dunne et al., 1993; Seidemann et al., 2002; Atchison et al., 2006) and accommodation (Smith et al., 1988).

Peripheral refractive errors are commonly described in terms of the astigmatism or Mean Spherical Equivalent error at a peripheral location. Both of these errors vary in well-known ways, but with some differences between myopes and emmetropes. Peripheral astigmatism tends to increase with eccentricity, but exhibits some temporal–nasal asymmetry. Seidemann et al. (2002), using a PowerRefractor, found that emmetropes and myopes exhibited more astigmatism in the temporal retina than the nasal retina, although their measurements taken with a double-pass technique suggested that emmetropes might suffer from greater peripheral astigmatism than myopes. Millodot (1981) and Dunne et al. (1993), using auto-refractors, found that the temporal retina was more astigmatic than the nasal retina in both emmetropes and myopes. This asymmetry has also been demonstrated by other studies concentrating exclusively on emmetropes (Rempt et al., 1971; Lotmar and Lotmar, 1974). Atchison et al. (2006) found that the asymmetry in peripheral astigmatism was due to an asymmetrical variation in the J₁₈₀ component, which was more pronounced in emmetropic observers.

These studies have also found variations in Mean Spherical Equivalent errors with eccentricity. Seidemann et al. (2002) found a peripheral myopic shift in both emmetropes and myopes when using their double-pass method, but found that this was greater in emmetropes than myopes at wide eccentricities. Several studies (Millodot, 1981; Mutti et al., 2000; Atchison et al., 2006) have found that Mean Spherical Equivalent errors display a myopic shift in hyperopes, in contrast to a slight hyperopic shift in myopes. Atchison et al. (2006) showed that the hyperopic shift in myopes was more pronounced with increasing myopia. The myopic shift found in emmetropes was also apparent in the original study by Ferree et al. (1931).

Myopia has also been shown to be associated with near work (McBrien and Adams, 1997; Mutti et al., 2002; Allen and O’Leary, 2006). Accommodative response in myopes is poorer than in emmetropes and it has been suggested that extensive and prolonged near vision may stimulate the development of myopia (Gwiazda et al., 1993) and cause the progression of myopia in already-myopic participants (Gwiazda et al., 1995; Abbott et al., 1998; Allen and O’Leary, 2006).

The potential relationships between myopia, peripheral refraction and near work, and the differences in peripheral refraction of emmetropic and myopic eyes, justify the study of the peripheral refraction of the eye for near vision as well as distance vision, yet a detailed investigation does not seem to have been attempted before. Smith et al. (1988) showed that peripheral astigmatism and field curvature increase when large amounts of accommodation are exerted, but directed their study towards emmetropic observers only. Walker and Mutti (2002) found that relative peripheral refraction became more hyperopic with accommodation, but did not look at astigmatism and did not separate their participants into different refractive error groups, although the mean refractive error of their cohort suggests that their participants tended to be myopic. If peripheral retinal defocus is related to the development of myopia, and if myopes experience under-accommodation for near vision, peripheral refractive errors may be more pronounced for near vision in myopes. We investigated this theory by measuring the peripheral astigmatic and Mean Spherical Equivalent errors in emmetropic and myopic participants for both distance and near vision, using a Shin-Nippon auto-refractor (Ajinomoto Trading, Inc, Tokyo, Japan). The results were used to assess the effects of near vision on peripheral refraction in both groups.

Methods

We measured the peripheral refractive errors of 10 emmetropic (six males and four females) and 10 myopic (three males and seven females) participants at two fixation distances. Results of a pilot study (Radhakrishnan et al., 2004) showed that an experiment using a sample of this size would give a statistical power of 83%, assuming an effect size of 0.25 D. The mean Mean Spherical Equivalent errors were −0.02 ± 0.16 D (emmetropes) and −3.50 ± 1.83 D (myopes). The mean age of the cohort was 25.3 ± 5 years and all participants were in good ocular health as determined by optometric examination. Only those with <1.25 D of astigmatism were included in the study.

All participants were drawn from the student population at Anglia Ruskin University, Cambridge. Their informed consent was obtained prior to the experiment and the tenets of the Declaration of Helsinki were followed throughout. Project protocols were approved by the Anglia Ruskin University Research Ethics Committee.

Peripheral refraction

Peripheral refraction measurements were taken with a Shin Nippon auto-refractor at viewing distances of 2.5 and 0.40 m. Cycloplegia was avoided in order to preserve accommodation for the near vision measurements and data were obtained from right eyes only. The subject fixated a high contrast letter target subtending a visual angle of 15 min of arc which was mounted on a screen at the appropriate distance from the cornea. Targets were positioned on the screen so that they were separated by 10°, 20° and 30° with respect to the line of sight in straight-ahead gaze, in both temporal and nasal meridians. Three refraction measurements were taken at each eccentricity. Each sphero-cylindrical refractive error measurement was decomposed into vector components using the following equations derived by Thibos et al. (1997):

• ( (1a))

• ( (1b))

• ( (1c))

where J₁₈₀ and J₄₅ are the 90° to 180° astigmatic components and the 45° to 135° astigmatic components, respectively, M is the Mean Spherical Equivalent error, and S, C and θ are the spherical, cylindrical and cylindrical axis components of the sphero-cylindrical refractive error.

The refractive error at a particular retinal location was expressed in vector terms as the mean of the values obtained from these three readings, and the means were used to recover the magnitude of the cylindrical component C′ using the equation

• (2)

The astigmatism at such a location was taken to be the magnitude of the cylindrical component C′ recovered from the vector components. Whereas Atchison et al. (2006) made extensive use of the vector components J₁₈₀ and J₄₅, for our data analysis we paid greater attention to the values of M and C′ described above. This is because we are interested in the magnitude of astigmatism at different locations on the retina, rather than studying the changes in astigmatic composition between different locations. The effect of peripheral astigmatism on myopia development is independent of astigmatic orientation (Kee et al., 2004), so the astigmatic magnitude may be more meaningful than its vector components. The use of vector components is helpful in studying the composition of astigmatism, but is less useful in other ways. For example, if an eye suffered from astigmatism of −1.00 DC × 180 at one location and −1.00 DC × 90 at another, a vector analysis using Equations (1) and (2) would imply a change of −2.00 D. This would suggest an increase in astigmatism even though the astigmatism is the same at both locations. A change from −1.00 DC × 180 to −1.00 DC × 45 would imply a change of −1.41 D. In both of these examples, the vector method conveniently demonstrates that the composition of astigmatism has changed, but inconveniently implies that its magnitude has changed and is therefore less useful for our study.

All measurements were made in a randomised order. Measurements for near vision were made with the myopic participants corrected using trial lenses; distance vision measurements were made with the participants corrected as well as uncorrected. The trial lenses were spherical and were manufactured in a flat form, from glass and treated with an anti-reflection coating compatible with the wavelength of the infrared radiation from the auto-refractor (840 nm). This minimised measurement errors arising from reflections from the lens surfaces. Participants fixated the peripheral targets by rotating their eyes, keeping their heads still. This ensured that measurements were taken along the optical axis of the correcting lens, minimising the effect of the lens on the data.

Results

The mean (±S.D.) foveal refractive error components for the different groups are shown in Table 1 and were used for subsequent analyses. To confirm that the correcting lenses had no effect on foveal or peripheral refractive error components measured in the myopic participants, the data for the corrected and uncorrected myopes were compared using a series of repeated measures anovas performed on the distance J₁₈₀, J₄₅, recovered astigmatic and Mean Spherical Equivalent values of the corrected and uncorrected myopes. The anovas were designed with eccentricity as a within-groups variable and refractive error group as the between-groups variable to investigate the effect of any interactions between group and eccentricity. There were seven levels of eccentricity: 0°, and 10°, 20°, 30° both temporally and nasally. This design permits temporal–nasal asymmetry to be considered without complicating the analysis by introducing retinal hemisphere as a separate variable. For the J₁₈₀ component (F_6,102 =16.43, p = 0.000001), J₄₅ component (F_6,102 = 14.20, p = 0.00001) and astigmatism, there were significant effects of eccentricity (F_6,102 = 11.84, p = 0.000001) only. With respect to the Mean Spherical Equivalent error there were no significant main effects or interactions. These analyses show that the presence of a correcting lens produced no significant effect on the absolute or component values of the peripheral astigmatism, or on the Mean Spherical Equivalent error in the myopic group.

Table 1.  Values (dioptres) of the mean ± S.D. foveal refraction components for the refractive error groups in this study

Refraction component	Refraction
	Emmetropes	Myopes (corrected)	Myopes (uncorrected)
Only measurements at distance were taken for the myopic participants.
C′
Distance	−0.40 ± 0.36	−0.67 ± 0.60	−0.61 ± 0.48
Near	−0.66 ± 0.35	−0.77 ± 0.69	–
M
Distance	−0.02 ± 0.16	−0.34 ± 0.44	−3.50 ± 1.83
Near	−1.74 ± 0.31	−1.63 ± 0.78	–
J180
Distance	+0.06 ± 0.25	+0.23 ± 0.37	+0.20 ± 0.29
Near	+0.21 ± 0.24	+0.29 ± 0.37	–
J45
Distance	−0.03 ± 0.10	+0.01 ± 0.13	−0.08 ± 0.16
Near	−0.06 ± 0.21	−0.02 ± 0.21	–

As peripheral performance in myopes was not affected by the correction, all subsequent analysis compared the peripheral refractions in corrected myopes with those in emmetropes for both distance and near, and these data are shown in Figures 1–6. Although we are less concerned with changes to the vector components, changes in the values of J₁₈₀ and J₄₅ at distance are shown in Figures 1 and 2, respectively, to enable comparison with other studies. When plotting Figures 1–6, the distance refraction components measured at the fovea (Table 1) were subtracted from the corresponding peripheral measurements to illustrate the way in which these components change with respect to the distance foveal values. The magnitudes of the peripheral J₁₈₀, J₄₅, recovered astigmatic and Mean Spherical Equivalent values of the two groups were examined using a series of repeated measures anovas with two levels of viewing distance and seven levels of eccentricity as within-groups variables, as before. Main effects were only considered in the absence of significant interactions, and any necessary post hoc testing was performed using Tukey’s HSD test. Because of the large number of factors involved, only statistically significant interactions and main effects are presented.

Figure 1.  Variation in J₁₈₀ component with eccentricity normalised with respect to the distance foveal value. Negative eccentricities denote measurements from the temporal retina (nasal field). Error bars represent ±1 S.E.M.

Figure 2.  Variation in J₄₅ component with eccentricity normalised with respect to the distance foveal value. Negative eccentricities denote measurements from the temporal retina (nasal field). Error bars represent ±1 S.E.M.

Figure 3.  Variation in astigmatism C′ with eccentricity, normalised with respect to the distance foveal value. Negative eccentricities denote measurements from the temporal retina (nasal field). Error bars represent ±1 S.E.M.

Figure 4.  Variation in Mean Spherical Equivalent error M with eccentricity normalised with respect to the distance foveal value. Negative eccentricities denote measurements from the temporal retina (nasal field). Error bars represent ±1 S.E.M.

Figure 5.  Variation in distance peripheral refraction of the most and least myopic meridians normalised with respect to the distance foveal value. Negative eccentricities denote measurements from the temporal retina (nasal field). Error bars represent ±1 S.E.M.

Figure 6.  Variation in near peripheral refraction of the most and least myopic meridians normalised with respect to the distance foveal value. Negative eccentricities denote measurements from the temporal retina (nasal field). Error bars represent ±1 S.E.M.

The magnitude of the J₁₈₀ component (Figure 1) shows some temporal–nasal asymmetry, with measurements taken from the temporal retina being greater than those taken from the nasal retina. The only statistically significant effect was an interaction between distance and eccentricity (F_6,102 = 2.38, p = 0.03). Post hoc analysis showed that there were significant differences between distance and near values at 20º (p = 0.03) and 30° temporally (p = 0.02), where the respective J₁₈₀ values were approximately 0.31 and 0.65 D smaller for near than for distance. There were no significant main effects or interactions involving refractive error group.

The magnitude of the J₄₅ component (Figure 2) changes in an approximately linear manner. With regard to the peripheral J₄₅ components (Figure 2), there were significant main effects of distance (F_1,17 = 16.10, p = 0.0009) and eccentricity (F_6,102 = 35.15, p = 0.000001), but not group. The effect of distance was much smaller for this component than for the J₁₈₀ component: at eccentricities of 20° and 30° temporally, the magnitude of the J₄₅ component changed by only 0.03 and 0.06 D respectively. Significant differences between temporal and nasal J₄₅ values were present at eccentricities of 20° and 30° (p = 0.0002 in each case).

Figure 3 shows that the magnitude of astigmatism increased with eccentricity, and this increase was greater in the temporal retina, especially in the emmetropic participants. Emmetropic participants in this study displayed about 2.6 D more astigmatism for an eccentricity of 30° in the temporal retina, and about 1.1 D more astigmatism at the same eccentricity in the nasal retina, than at the fovea. For the myopic participants, these values were more symmetrical at about 1.1 D at 30° either side of the visual axis. This effect was confirmed by an interaction between group and eccentricity (F_6,102 = 3.13, p = 0.007), with significant differences between myopes and emmetropes present at 30° in the temporal retina (p = 0.0002), and significant differences between the emmetropes’ astigmatism at 30° temporally and the same eccentricity nasally (p = 0.011). There was a reduction in temporal retinal astigmatism at near, and this was confirmed by a significant interaction between distance and eccentricity (F_6,102 = 8.09, p = 0.000001), with a significant difference found between distance and near astigmatism at a retinal eccentricity of 30° temporally (p = 0.0001). There was also a significant difference between temporal and nasal astigmatism for distance vision at 30° (p = 0.0001).

Changes in Mean Spherical Equivalent errors are shown in Figure 4. For distance, there is little variation with eccentricity in the myopes, but the emmetropes display an increase in Mean Spherical Equivalent error on both sides of the fovea, in particular in the temporal retina. For near vision, the myopes showed a slight tendency to relative peripheral hyperopia in the temporal retina, but in the emmetropes this effect occurred in the nasal retina. There was a significant interaction between group, distance and eccentricity (F_6,102 = 2.45, p = 0.03), with significant differences between the emmetropes and myopes at 20° and 30° temporally for distance (p = 0.0019 in each case), and at 30° temporally for near (p = 0.0016). The emmetropes also displayed greater asymmetry, having significant differences between temporal and nasal Mean Spherical Equivalent errors at 30° eccentricity for both distances (p = 0.036) and near (p = 0.011).

The effect of eccentricity on the peripheral refraction of the least myopic and most myopic meridians is shown in Figures 5 and 6. At distance (Figure 5), there is a slight tendency towards relative peripheral hyperopia in the least myopic meridian in either group, but the most myopic meridians show a further myopic shift in the periphery. Although this is true of both groups, the effects are more asymmetrical in the emmetropes. With respect to the most myopic meridian, the emmetropic group experienced a strong myopic shift in the temporal retina. Changes in peripheral refraction at near are plotted in Figure 6 and are expressed with respect to a baseline measurement at the fovea at distance. At near there is little change in least myopic refractive error with eccentricity, although some asymmetry is suggested. For example, the myopes demonstrated a greater hyperopic shift at 30° in the temporal retina than in the nasal retina. Results for the least myopic and most myopic meridians of the two groups were analysed using repeated measures anovas with viewing distance and eccentricity as within-groups variables. For the least myopic meridians, there was a significant interaction between viewing distance and eccentricity (F_6,102 = 2.26, p = 0.04). Post hoc tests showed significant differences between distance and near refractive errors at all eccentricities (p = 0.0001). The most myopic meridian in the emmetropic group became more myopic in the temporal retina for near vision and there was a significant interaction between group, distance and eccentricity (F_6,102 = 2.59, p = 0.022) for this meridian. Post hoc analysis demonstrated differences between emmetropes and myopes at both 20° and 30° in the temporal retina for distance, and at 30° in the temporal retina for near vision (p = 0.0002 in each case).

Discussion

In common with previous studies of peripheral refraction, our experiments show that the retinal periphery is more astigmatic than the fovea, with some evidence of temporal–nasal asymmetry which was more pronounced in emmetropes than in myopes. Emmetropic participants in this study displayed greater astigmatism for an eccentricity of 30° in the temporal retina than the nasal retina, but myopic participants showed less asymmetry. Seidemann et al. (2002) found that emmetropes and myopes both exhibited peripheral astigmatism of about 1.5 D at 25° temporally and about 0.5 D at the same eccentricity nasally. When measurements were extended to eccentricities of 30° using the double-pass technique, myopes displayed about 2 D of astigmatism and emmetropes about 2.5 D. Millodot (1981) obtained measurements of approximately 1.5 and 1 D for eccentricities of 30° temporally and nasally, respectively, in both emmetropes and myopes, and Dunne et al. (1993) reported higher values of approximately 3.5 and 1.5 D, respectively. Other studies on emmetropes (Rempt et al., 1971; Lotmar and Lotmar, 1974) found values of about 1–2 D at a similar angle. The differences in asymmetry between our two groups resulted in significant inter-group differences in astigmatism, but only for an eccentricity of 30° in the temporal retina. When our results are broken down into vector components, the magnitude of the J₁₈₀ component displays the temporal–nasal asymmetry that Atchison et al. (2006) have previously reported. In the present study, the J₁₈₀ component of our emmetropic participants was greater in the temporal field than at the same eccentricity in the nasal field, which appears to be similar to the results of Atchison et al. (2006). For the myopic participants there was less asymmetry, agreeing with the results of Atchison et al. (2006). The magnitude of the J₄₅ component varied approximately linearly with field angle, although with a steeper slope than that reported by Atchison et al. (2006). Our results, therefore, are similar to those of other studies in terms of the magnitude of astigmatism and the existence of temporal–nasal asymmetry although the asymmetry measured in our study was greater than has sometimes been found before, but lower than that measured by Dunne et al. (1993). The asymmetry in the measurements of the J₁₈₀ component suggests that this component may be responsible for the asymmetry in the total astigmatism. Astigmatic asymmetry can be attributed to a tilted or displaced crystalline lens, or angle lambda (Barnes et al., 1987; Dunne et al., 1993). Atchison et al. (2006) suggested that corneal asymmetry also plays an important role in variations in peripheral refraction, particularly the J₁₈₀ component. Bearing in mind the asymmetry present in the J₁₈₀ values of the emmetropic participants, it is possible that corneal asymmetries are different in myopic and emmetropic eyes, possibly because of the effects of extensive reading on the corneal shape (Buehren et al., 2005).

Considering the Mean Spherical Equivalent errors, our results show that emmetropes become relatively myopic at peripheral eccentricities, whereas there is little change in myopes. The peripheral myopic shift found by Seidemann et al. (2002) was approximately 2 D in emmetropes as opposed to 1 D in myopes at 30° eccentricity when using a double-pass technique, but their measurements obtained with the PowerRefractor showed little variation with eccentricity. Millodot (1981) found that there was about 1 D of myopic shift at 30° in hyperopes. Mutti et al. (2000) measured peripheral refraction in myopes and emmetropes and also found a hyperopic shift in myopes but a myopic shift in emmetropes, and Atchison et al. (2006) reported similar findings. The emmetropic participants in our study behaved in a manner similar to those in previous experiments, but our myopic participants did not display a significant peripheral hyperopic shift. However, the hyperopic shift measured by Millodot (1981) was only approximately 0.5 D at an eccentricity of 30° although that found by Mutti et al. (2000) was slightly larger at 0.8 D at the same eccentricity in the nasal field. At the same eccentricity, Atchison et al. (2006) found a hyperopic shift of about 1 D for their group of −4 D myopes but a smaller shift of approximately 0.5 D for their −3 D myopes. From these studies it appears that any hyperopic shift experienced by myopic participants is likely to be <1 D so we may not expect much change in refraction in myopes. Our results, therefore, tend to agree with other studies with regard to myopes as well as emmetropes.

Previous studies have suggested that emmetropisation is directed towards the least myopic meridian of an eye (Seidemann et al., 2002). Our results support this view in that eccentricity has little effect on the refractive error of the least myopic meridians of either group. On the other hand, the most myopic meridians of each group become more myopic with increasing eccentricity and this effect is stronger in the emmetropes, as observed by Millodot (1981) and Seidemann et al. (2002). Some recent research has examined the impact of astigmatism on emmetropisation in animals, but the results have been equivocal. Whereas Kee et al. (2004) found that emmetropisation was aimed at the least hyperopic meridian in monkeys, and Schmid and Wildsoet (1997) found similar effects in chickens, Seidemann et al. (2002) found that the opposite applied in humans. Our results, also for humans, support the latter view.

There is little evidence of a general change in the magnitude of the astigmatism at near although, in the emmetropes, astigmatism at an eccentricity of 30° in the temporal retina is lower for near vision than for distance, and this applies to the magnitude of the astigmatism as well as its vector components. Astigmatism in the temporal retina is lower at both distance and near in the myopes than in the emmetropes, but astigmatism in the nasal retina is not affected by viewing distance and is similar in both groups. It is possible that astigmatic changes may arise because of the effects of accommodation on the shape of the cornea (Pierscionek et al., 2001) or the crystalline lens, or the effects of lid pressure on corneal shape during near vision (Buehren et al., 2005). The change in Mean Spherical Equivalent errors with eccentricity does not differ much from distance to near, apart from the under-accommodation at near. That there is no clear effect of viewing distance is consistent with the data of Smith et al. (1988) who found that accommodation had little effect on astigmatism and field curvature in emmetropes when the eccentricity was <40°. Although Walker and Mutti (2002) found that accommodation induced a hyperopic shift in relative peripheral Mean Spherical Equivalent error, they did not classify participants by refractive error, and their data suggest that the magnitude of the shift displayed some inter-subject variability. Our results tend to support Smith et al.’s (1988) observation in that we found little change in astigmatism with viewing distance, apart from the changes in the temporal retina: when averaged across both groups and both hemispheres, astigmatism at 30° fell by only 0.2 D for near vision. In addition, the Mean Spherical Equivalent errors were not generally affected by viewing distance, although for near vision the myopes demonstrated a hyperopic shift in the temporal retina compared with the fovea.

One of the principal aims of this experiment was to investigate the relationship between myopia and peripheral refraction at near. Our results demonstrate that, apart from the relative under-accommodation experienced by the myopes, viewing distance has little general effect on peripheral refraction. The temporal–nasal asymmetry is present for near vision as it is for distance vision, with the astigmatism in the temporal retina of the emmetropic participants tending to reduce for near vision. Changes in astigmatism in the temporal retina in the myopes, and in the nasal retina in both groups, are much less affected by viewing distance. Therefore, although the retinal defocus because of under-accommodation is greater in myopes, changes in peripheral refraction during near vision are unlikely to exacerbate significantly their peripheral retinal image blur. These results do not support the idea that near peripheral refraction is related to myopic development.

We designed our study so that the subjects’ refractive errors were corrected with trial lenses during the measurement. The subjects’ heads remained stationary throughout the procedure, ensuring that all measurements were taken along the optical axes of the correcting lenses, avoiding induced aberrations. This strategy appears to have been successful as the results obtained with the correcting lenses were not significantly different to those obtained without the lenses. In reality, a myopic observer wearing their refractive correction would find that their peripheral retina received light that had passed obliquely through the correcting lens. This would affect the peripheral retinal image. For example, in the case of a −3.5 D myope (equivalent to the mean Mean Spherical Equivalent of the myopes in our study), light entering the eye at an object space field angle of 30°, passing through a −3.5 D spectacle lens made from 1.5 refractive index glass with a commercial front surface curve of +2 D (Norville Group Limited, 2005) and situated 16 mm from the entrance pupil, would experience approximately −0.4 D astigmatism and −0.2 D mean oblique error caused by the lens. This is smaller than the mean peripheral errors measured in this experiment. Therefore, we believe that correcting lenses would have a small effect on peripheral refraction: the eye’s own peripheral astigmatism would be much more significant.

The peripheral refractive errors of the most and least myopic meridians of both groups changed in similar directions with eccentricity and viewing distance, although those in the emmetropic group demonstrated more asymmetry. The fact that the refractive errors of each group changed in similar directions suggests that myopia is not associated with general differences in peripheral refraction. It is important to remember that our analysis emphasises changes in the magnitude of astigmatism rather than the orientations of the principal meridians and it might be argued that such orientational changes may differ between the two groups. However, this is unlikely as the results of Seidemann et al. (2002) show that the orientation of the most myopic meridian is hardly affected by eccentricity and Kee et al. (2004) found that, in monkeys, the introduction of astigmatism stimulated changes in axial length irrespective of astigmatic orientation.

As the only significant differences in the peripheral refraction between the two groups arise for wide eccentricities in the temporal retina, it could be argued that peripheral retinal image quality plays little role in the development of myopia. Although this is possible, there is some evidence to the contrary. Local changes in the retinal image influence local eye growth in chicks (Wallman et al., 1987), and Smith et al. (2005) found in monkeys that interfering with peripheral retinal image quality produced a change in axial length. It is also important to note that the effect of any variable on the development of myopia really requires a longitudinal study rather than the present cross-sectional investigation. Ideally, longitudinal or cross-sectional studies should also make use of larger sample sizes than this one so that the results can be more reliably translated to the general myopic population.

Optical systems suffer from other off-axis aberrations, such as coma, and these may play a part in the development of myopia. Although the off-axis optical performance of the human eye has been measured using double-pass methods (Navarro et al., 1998; Guirao and Artal, 1999), these techniques have not yet been used to compare the coma in myopic eyes with that in emmetropic eyes. The dependence of coma on eccentricity (Guirao and Artal, 1999) suggests that it may have an effect on peripheral retinal image quality. While concentrating on the visual axis, some studies have found that myopic eyes have higher levels of third-order coma-like aberrations compared with emmetropes (He et al., 2002) and others have shown that the dominance of coma increases as the magnitude of myopia in a population increases (Paquin et al., 2002). the measurement of off-axis coma would therefore seem to be justified.

Conclusions

Apart from differences at wider eccentricities in the temporal retina, and some asymmetry, peripheral refraction varies in a similar way in myopes and emmetropes for both distance and near vision. Our results do not support the hypothesis that myopia is related to changes in peripheral refraction during near vision.