Accommodative instability: relationship to progression of early onset myopia


Dr Trine Langaas, Department of Optometry and Visual Science, Buskerud University College, Postbox 235, Frogs Road 41, N-3603 Kongsberg, NORWAY, E-mail:


Background:  In a previous study, we demonstrated that children with early onset myopia had greater instability of accommodation than a group of emmetropic children. Since that study was correlational, we were unable to determine the causal relationship between this and myopic progression. To address this, we examined the children two years later. We predicted that if accommodative instability was causing the myopic progression, instability at Visit 1 should predict the refractive error at Visit 2. Additionally, instability at Visit 1 should predict myopic progression.

Methods:  Thirteen myopic and 16 emmetropic children were included in the analysis. Dynamic measures of accommodation were made using eccentric photorefraction (PowerRefractor) while children viewed targets set at three distances (accommodative demands), namely, 0.25 metres (4.00 D demand), 0.5 metres (2.00 D demand) and 4.00 metres (0.25 D demand).

Results:  Both refractive error and accommodative instability at Visit 1 were highly correlated with the same measures at Visit 2. Children with myopia showed greater instability of accommodation (0.38 D) than children with emmetropia (0.26 D) at the 4.00 D target on Visit 1 and this instability of accommodation weakly predicted myopic progression.

Conclusions:  The results presented in the present study suggest that instability of accommodation accompanies myopic progression, although a casual relationship cannot be established.

The aetiology of early onset myopia is not yet clearly understood, despite the many studies that have suggested mechanisms for this disorder.1–11 For example, evidence suggests that development of myopia is associated with a lag of accommodation.12–14 This was thought to produce myopia through hyperopic retinal defocus, particularly of the longer wavelengths of light.15 In confirmation of this, both human and animal work has demonstrated that hyperopic retinal defocus is a sufficient stimulus to myopia;3–5 however, more recent research has suggested that while a lag of accommodation is a common finding in patients with myopia, this lag in early onset myopia is only present after the myopia has progressed and therefore cannot be the causal mechanism.16,17 Thus, other candidate mechanisms have been sought.

Another mechanism that has been investigated is whether instability of accommodation might drive myopic progression.17–30 Several adult studies19–24 suggest that decreased stability of accommodation is related to either sensitivity to detect errors in accommodation or sensitivity to blur. In one study, young adult myopes had reduced sensitivity to blur compared with emmetropes25 and this was suggested as a risk factor for myopia.26 Decreased sensitivity to blur might lead to a decreased stability of accommodation. This could result in an increase in hyperopic defocus when errors of accommodation are integrated over time, which in turn could result in significant elongation of the eyeball,6,7 providing a direct causal pathway. As a first step in disentangling possible relationships between stability of accommodation and myopia, it is important to determine the temporal relation between decreased stability and myopic progression.

There is some debate in the literature over whether decreased stability of accommodation is found only in adults with late onset myopia26 or whether it can also be found in adults with early onset myopia.31 Differences in accommodative stability between adults with early versus late onset myopia are of potential interest, because there are two possible interpretations of these findings. First, it is possible that differences between these groups could result from differences in aetiology, with decreased stability of accommodation providing a causal mechanism for late onset myopia only. In this case, children who are developing myopia (early onset myopes) would not be expected to show decreased stability. Alternatively, differences in stability between groups with early and late onset adult myopia could relate to the timing of the measurement of instability with respect to the onset of the myopia. If instability of accommodation causes the myopia, it would have to be present before the myopia but would not necessarily persist after progression is complete. In this case, decreased stability would be found in adults with late onset myopia (those whose myopia is still progressing) but might not be found in adults with early onset myopia (for whom myopic progression is complete). In comparison, children who are developing myopia (that is, those who become early onset adults) would be expected to show decreased stability of accommodation.

In a recent paper,1 we demonstrated that children with early onset myopia were found to have decreased stability of accommodation compared with emmetropic children when tested at a time when their myopia had started to progress. This research supported the theory that accommodative instability could be a potential mechanism for driving myopia. Our previous study demonstrated only a correlation between accommodative instability and myopia, thus providing no evidence to determine the direction of causality. In the present paper, we report data from children who were tested in our previous study and were then invited back two years later. Data from each visit was used to investigate possible causal relationships between stability of accommodation and myopic progression. We predicted that children who demonstrated accommodative instability at Visit 1 would continue to show this instability at Visit 2, especially at the nearest target distance where accommodative demand is greatest. In addition, if instability of accommodation drives myopia either directly or as a consequence of insensitivity to blur, we predicted that instability at the first visit would predict myopic progression. The results of the present study suggest that instability of accommodation accompanies myopic progression.



Approval for this research was obtained from the National Committees for Research Ethics in Norway prior to the instigation of the study. The experiment followed the tenets of the Declaration of Helsinki. Experiments commenced after informed consent was obtained from parents/guardians and also from the participants, if they were over the age of 12 years.

For our previous study,1 we invited participants who failed a local optometric screening due to myopia to participate. In addition, we recruited a group of emmetropic participants from local schools. The children from the local schools came from one class chosen to be similar to the age range of the myopic children we had recruited. While children were matched for mean age, this resulted in the range of ages being slightly larger for the myopic than the emmetropic group. In the present study, we invited all participants that had produced useable data in our previous study to return to the laboratory for a two-year follow-up test and 37 of these children returned.

All participants had visual acuity of 6/6 or better. Any participant with anisometropia or astigmatism greater than 1.0 D with manifest strabismus or amblyopia, or with a history of ocular health problems were excluded from the study. Participants were categorised on the basis of their refraction at Visit 2. All participants in the myopic group had a subjective, non-cycloplegic refraction of at least -0.50 D mean spherical equivalent (sphere + 0.5 cylinder) in each eye and were considered to need a full-time prescription for myopia by a qualified optometrist. The emmetropic group consisted of emmetropes and low hyperopes (subjective non-cycloplegic refraction from plano to a mean spherical equivalent up to +0.75 D), who were not considered to require optical correction.

Fifteen myopic participants from the original cohort agreed to be tested in the follow-up study. As myopia was first identified at screening, there is no information available on the age of onset of myopia. As all participants were below the age of 15 at Visit 1, this was classified as early onset myopia.2 Twenty-two emmetropic participants from the original cohort also returned for the follow-up study.


Accommodation was recorded using an eccentric infrared photorefractor, the PlusOptix PowerRefractor II (Plusoptix Inc, Hillsboro Beach, FL, USA). It is a safe device for measuring accommodative response32 and a useful tool for measuring refraction in infants.33 The PowerRefractor includes a dynamic mode of operation in which the refractive status of both eyes is recorded continuously at a sampling frequency of 12.5 Hz. Pupil size was within the range of 3.5 to 7.0 mm, a range that has previously been reported to provide consistent accommodative responses.32 A more detailed description of this procedure has been reported previously.1

Measurement procedures

At each visit, all participants were corrected for distance and wore their subjective refractive correction with full aperture lenses in a trial frame during the test procedure. The trial frame was fitted on the participant so that the line of refraction through the refractometer was not disrupted. Any refractive error in 0.25 D steps (sphere and cylinder) evident during subjective examination was corrected, with the result that all myopes and most of the emmetropic participants wore refractive lenses during the procedures at Visit 1. At Visit 2, all participants viewed the targets through either plano or correcting lenses as appropriate. This ensured that variability in accommodation due to uncorrected refractive error was eliminated and that testing conditions were similar between the two groups. In addition, as both groups were tested through lenses, it is less likely that any difference in variability between the groups is the result of instrument error due to reflections from the trial frame or corrective lens.

The subjects were seated in front of the PowerRefractor with the head stabilised on a head and chin rest. Fixation targets were identical to those used at Visit 1, namely, printed letters equivalent to letter size N5 and scaled for fixation distance (Figure 1).

Figure 1.

Targets used to measure accommodation at 4.00 D (A), 2.00 D (B) and 0.25 D (C). Letters were typed in Arial font with double spacing between letters and rows, and were scaled to be N5 for each target distance.

The participants were instructed to look at each of the three targets in a set order: 4.00 D, 2.00 D, 0.25 D, 4.00 D, 0.25 D and 2.00 D. This pseudo-random target order ensured that all participants viewed all distances twice with the same variable step changes across the trial. The accommodative response was recorded continuously during the procedure. Data were viewed online during collection to determine when the participant was accommodating steadily at each target and to ensure that data were being collected. The experimenter waited until a steady accommodative response was achieved for an estimated two to three seconds on the real-time output at each distance before asking the participant to move to the next target.

The data were stored offline and transferred to spreadsheets for later data analysis. Blinks were first eliminated from the data, as in our previous study.1 Out of the initial pool of 37 participants that returned for follow-up testing, the data for eight participants (one myope and seven emmetropes) had to be discarded during offline analysis, because the data revealed that the participant had not achieved a steady accommodative response lasting more than one second at more than two target distances. This was usually the result of inattention and/or blinks in the data, which reduced the duration of the accommodative response. In all, 29 participants (13 myopes and 16 emmetropes, at the time of Visit 2) were included in the analysis.

As only one eye per participant was used in the analyses, visual inspection of the data for each eye was used to determine the eye with the most stable accommodation and the data from this eye were analysed. The starting point of each fixation was chosen as the first point at which the accommodative response remained at a consistent new plane for at least one second and the end-point was chosen as the point before which there was a change of at least 0.75 D in accommodation towards the next target plane, as explained previously.1 For each participant, sections of accommodative data that lasted a minimum of three seconds (minimum of 40 data points per fixation) and did not contain blinks or eye movements were chosen for each fixation target. The standard deviation of this section of data was used as an estimate of stability. Although the number of data points at each distance and pupil size was not constant across participants, these measures were not considered here as they were shown to have no effect on stability of accommodation in our previous study.1

Table 1 shows the mean age, spherical equivalent non-cycloplegic refraction and time between visits for the myopes and emmetropes at each visit. At both visits, participants were categorised by their current refractive status using the same strict criteria for inclusion in the myopic group. One low myope at the first visit was found to be emmetropic at the second visit and hence the number of subjects differed between visits. There was no significant difference in age at either visit or time between visits, but as by definition, the refractive error was significantly different between the two groups. Our sample of emmetropic children included four who had a subjective, non-cycloplegic refraction of 0.75 D. We considered the possibility that these children might have had different accommodative stability patterns than the rest of the emmetropic group; however, comparison of means showed that there were no differences in performance between these children and the rest of the group and therefore these children were retained in the analysis.

Table 1. Subject details
 Visit 1Visit 2
Emmetropes (15)Myopes (14)Emmetropes (16)Myopes (13)
Mean ± SERangeMean ± SERangeMean ± SERangeMean ± SERange
Age (years)13.45 ± 0.1010 to 1413.87 ± 0.4712 to 1415.72 ± 0.1112 to 1716.02 ± 0.5814 to 17
Refraction (D)0.10 ± 0.07-1.00 to 0.75-1.37 ± 0.24-4 to -0.50.11 ± 0.09-0.5 to 0.75-2.27 ± 0.28-4.5 to -0.5
Between visits (years)    2.27 ± 0.03 2.15 ± 0.04 


We used the stability of accommodation across time as the dependent variable in most analyses, which were performed using ANOVA and regression analyses. We have reported F-values, p-values and observed power for these analyses. Pairwise comparisons (Bonferroni corrected) were used for post-hoc comparisons. Bland Altman analyses were used to examine relationships between measures across visits.


We first determined whether refractive error and variability of accommodation were repeatable when measured two years apart using correlation analyses. Figure 2a shows the correlation between refractive error at Visits 1 and 2. This was highly correlated between visits suggesting that children retained their refractive category across this time period (r2= 0.81, p < 0.0001). Figure 2b shows the correlations between accommodative variability at Visits 1 and 2 for each target distance. Again, there were significant correlations between measures taken two years apart for all target distances (4.00 D: r2= 0.55, p < 0.0001; 2.00 D: r2= 0.39, p < 0.0001; 0.25 D: r2= 0.19, p = 0.016).

Figure 2.

Correlations between measures taken at Visit 1 and Visit 2 (two years later). Figure 2A shows the correlation between refractive errors at Visits 1 and 2. There is a significant correlation between visits. Figure 2B shows the correlation between accommodative stability at each visit for each target distance. Again, there were significant correlations between visits.

A Bland Altman analysis was used to investigate the repeatability of refractive error and accommodative stability across visits. There was no significant bias between measures of refractive error at Visits 1 and 2 (t = 1.55, p > 0.05). A regression analysis showed that there was a significant positive correlation between the bias and the mean refractive error (r2= 0.26, p = 0.005). This indicates that children who were myopic at the first visit continued to show myopic progression and therefore were more myopic at the second visit. The refractive error changed by a mean of -0.75 D (or -0.325 D per year) for the myopic group and -0.07 D (or -0.035 D per year) for the emmetropic group.

The second Bland Altman analysis for the stability of accommodation at each target distance showed a significant increase in the bias for the 4.00 D target (t = -2.36, p = 0.025) demonstrating that children were more stable at Visit 1 (mean = 0.31 D) than at Visit 2 (mean = 0.36 D). There was no significant bias for either the 2.00 D or 0.25 D target (2.00 D: t = 0.026; 0.25 D: t = -0.67). For the 4.00 D target, a regression analysis showed that the bias became more negative with increasing variability (r2= 0.31, p = 0.002), suggesting that children with unstable accommodation at Visit 1 showed greater instability at Visit 2 than more stable children. This trend approached significance for the 2.00 D target (r2= 0.13, p = 0.054) but was not found for the 0.25 D target (r2= 0.009).

We had predicted that accommodative instability measured at Visit 1 would be related to refractive status at Visit 2 if instability accompanied myopic progression. As variability was significantly greater for the 4.00 D target than for the 2.00 D or 0.25 D targets at each visit, only the 4.00 D target variability was used in this analysis (Figure 3). An ANOVA was used to compare accommodative instability at 4.00 D for Visit 1 between children who were myopic or emmetropic at Visit 2. There was a significant difference between groups on stability of accommodation (F1,27= 7.91, p < 0.009, observed power = 0.77). This result lends support to a relationship between accommodative instability at Visit 1 and the refractive status two years later.

Figure 3.

Stability of accommodation for the myopic and emmetropic groups when categorised by their refraction at Visit 2. There is a significant group difference in stability of accommodation at Visit 1.

The mean accommodative instability in the myopic participants for the 4.00 D target at Visit 1 was 0.38 D (range: 0.23 to 0.57 D), while the mean for the emmetropic participants was 0.26 D (range: 0.12 to 0.36 D). Thus, the highest accommodative instability in the emmetropic participants was below the mean for the myopic participants. It was also found that no emmetrope became myopic during the course of the present study. Additionally, if good accommodative responses are defined to be the average stability for an emmetropic participant (0.26 D), then only one myopic participant had stable accommodation. This myope did not show myopic progression over the course of the study.

Finally, to determine whether instability at the first visit could be used to predict the degree of myopic progression, we regressed the instability at Visit 1 against progression of myopia (Figure 4). This showed a small but significant predictive effect of instability on myopic progression (r2= 0.15, p = 0.046). This was partly driven by three children whose myopic progression was -1.50 D (0.75 D per year). This is larger than that reported in a previous study.34 Therefore, we repeated the regression analysis removing these children. The regression remained significant (r2= 0.14, p = 0.05). Thus, the greater myopic progression was found for children who had shown the greater accommodative instability for the 4.00 D target at Visit 1.

Figure 4.

Scatterplot showing the relationship between variability for near at Visit 1 and myopic progression. There was a significant correlation between these measures (r2= 0.15, p = 0.046).


In a previous study,1 it was found that a group of children with early onset myopia demonstrated less stability of accommodation than a group of emmetropic children. It was postulated that this accommodative instability could be a causal factor in myopic progression. Alternatively, the increased instability might have resulted from the myopia and/or myopic progression.

A Bland Altman regression analysis of the repeatability of refractive error demonstrated that the more myopic children changed their refraction more (that is, became more myopic) than emmetropic children (-0.75 D or -0.325 D per year for the myopic group and -0.07 D or -0.035 D per year for the emmetropic group). This compares favourably with previous studies, which found that progression after the onset of myopia is typically -0.39 to -0.52 D per year34 for early onset myopes.

Accommodative instability measured at each visit showed a significant bias for the 4.00 D target only, with children becoming less stable at the second visit. Bland Altman regression analysis showed that there was a relationship between stability and bias, with a greater bias in children who were less stable. Thus, children with greater instability at the first visit showed the greatest change in stability across visits and were less stable when measured at Visit 2.

As in our previous study,1 the accommodation was least stable for the 4.00 D target compared with the 2.00 D and 0.25 D targets in both groups. This could be the result of the larger accommodative responses required to overcome the accommodative demand of the 4.00 D target in comparison to the 2.00 D and 0.25 D targets. Also, accommodation is typically most stable around the tonic accommodation level.35

In the present study, we were able to determine how accommodative variability measured two years earlier at Visit 1 was related to the refractive group at Visit 2. We had predicted that if the accommodative instability accompanied the myopic progression, then it would be greater in the myopic than the emmetropic groups. As predicted, the myopic group was found to have had significantly less stable accommodation at Visit 1 than the emmetropic group.

A regression analysis looking at the relationship between the stability of accommodation for the 4.00 D target at Visit 1 and myopic progression was used to determine whether instability was predictive of myopic progression. The effect in the regression analysis presented here was only just significant but it might indicate that instability in accommodation over time for 4.00 D targets might contribute to myopic progression. This preliminary finding could be tested further by measuring stability of accommodation in a group of children who are at risk of developing myopia and determining whether accommodative instability precedes the myopia.

A previous study reported that adults with late-onset but not those with early onset myopia26 have instability of accommodation. One explanation for these findings is that this difference could have arisen from differences in the aetiology of early onset versus late-onset myopia, with instability of accommodation only found in late-onset myopia. If this were correct, we would not have expected to see instability of accommodation in our early onset myopes, assessed while their myopia was progressing. The alternative hypothesis was that the instability of accommodation is present during myopic progression but later stabilises. This latter hypothesis is supported by the findings reported here.

There are several limitations to the present study. First, the sample size was small, and this prevented us from being able to make a comparison between myopes that progressed (eight myopes) and those that did not (five myopes). Additionally, as our participants were recruited after the onset of their myopia, we were unable to determine whether the accommodative instability preceded the myopia. Finally, some of the myopic children recruited were low myopes. It is possible that instability of accommodation is related to the size of the refractive error. Further research investigating stability of accommodation before the onset of myopia in a larger group of children with a wide range of myopia is required to verify our findings.


The results presented in the present study suggest that instability of accommodation accompanies myopic progression. Decreased stability of accommodation is one possible factor that might predict myopic progression but further research is needed to establish any causal relationship.


The authors thank Ellen Svarverud, Katrine H Larsen, Ann Ystenæs, Cecilie Bjørset, Irene Langeggen, Marit H Fjerdingstad and Jorunn Lid for help with testing the children. Preliminary results were presented at the Association for Research in Vision and Ophthalmology conference, Fort Lauderdale, Florida, 2009.


This study was supported by grant no. 176541/V10 from The Norwegian Research Council.