It was recently suggested that emmetropization targets a cycloplegic refractive error of +1D by the age of 2 years,1–3 which is theorized to protect against myopization. After 2 years the eye is in homeostasis, a period during which the axial growth and lens power loss are matched to keep the refractive error near +1D.1,2 Emmetropization does not occur at the same rate in all eyes, however, causing the distribution of the refractive error to gradually become narrower by the age of 6 years.1,4,5 Especially in populations growing up with a high exposure to outdoors light, this process can even continue until as late as 12 years,6 with a clustering of the refractive error around moderate hyperopia without increases in myopia, showing a late emmetropization.

Without cycloplegia, the refractive target of emmetropization is much closer to 0D,7 suggesting that, under physiological conditions, the low hypermetropic refraction is neutralized by the accommodative tonus. This tonus of about +0.75D may be linked to night myopia,8,9 the phenomenon causing someone who is emmetropic in photopic circumstances to experience blurred distance vision in the dark due to a lack of visual stimuli. Hence, emmetropes may experience a night myopia of–0.75D, corresponding with the change from the cycloplegic baseline refractive error as confirmed consistently in young participants.

Although the course of refractive development is quite well known in children, it remains unclear what the normal refractive target is in young adults (aged 20 – 40 years) due to an overall lack of studies in this age group. Moreover, existing studies are non-cycloplegic and often focused on subgroups prone to myopia that are not representative for the general population (e.g., people in a clinic or university students). This has led to the idea10–12 that there may be a myopic phase in healthy young adults, but as myopia generally does not reverse, a slight myopic shift seen in young adults in cross-sectional samples seems unrealistic on a prospective basis. Finally, many clinicians seem to be under the impression that eyes of this age do not undergo any active development, even though myopia can still emerge or progress further in this group showing that the mechanism governing eye growth is still active.13 The high prevalence of myopia in young adults causes the refractive distribution to skew, so that neither the average nor the median of the refractive error can provide an accurate estimate of the target refractive error unbiased by myopia. Consequently, such estimates in young adults show a wide range between low hypermetropia14,15 and low myopia.10–12 After the age of 40 years the refractive error shows a gradual return towards low hypermetropia until 70 years,10–12,15 followed by a final shift back towards emmetropia that is induced by cataract.11,12,15,16

The uncertainty about the average refractive error in young adults leaves an important question about the long-term target of normal refractive development. To address this issue, the cycloplegic and non-cycloplegic refractive errors of the Tehran Eye Study (TES) are reanalyzed using Bigaussian analysis, which is able to negate the influence of myopia that affected an earlier analysis.17 The TES data are ideal for this purpose as they represent the only large population study in the literature with cycloplegic data for young adults.

The histogram in Fig. 1 shows the distribution of the cycloplegic spherical equivalent refractive in bins 0.25D wide. Fitting a Gaussian function and the sum of two Gaussian functions gave coefficients of determination of r2= 0.962 and r2= 0.994, respectively. The F-test suggests that the Bigaussian representation is better able to capture the specific shape of the distribution (F = 72.21, p < 0.001). The Bigaussian function is therefore used in the following.

Changes with age

The global mean refractive error under cycloplegia decreases with age during childhood until it reaches a mildly myopic minimum (–0.26 ± 1.75 D) at 20 – 24 years, after which it increases again (Table 1). This agrees with the previous literature and reflects the influence of myopia incidence in adolescents and the known hypermetropization in older eyes due to changes in the gradient index of the lens.21,22 Applying Bigaussian fits to histograms with bins 1D wide for each age group, the emmetropized peak remains inside a range of [+0.39D, +0.50D] for 15 – 49-year-olds and later increases towards hypermetropia (Table 1, Fig. 2). The global median seems to follow a similar course as the position of the emmetropized peak.

Table 1.

Mean cycloplegic refractive errors (mean ± StDev) of the emmetropized and dysregulated peaks determined for each age group using Bigaussian fits.

Age range	N	Globalmean	Global median	Emmetropized peak	Dysregulated peak	r2
[0, 4]	103	1.17 ± 0.71	1.25	1.14 ± 0.57	0.58 ± 1.65	0.989
[5, 9]	262	1.13 ± 0.98	1.13	1.03 ± 0.55	1.11 ± 1.75	0.998
[10, 14]	440	0.62 ± 1.12	0.75	0.71 ± 0.44	0.42 ± 1.31	0.999
[15, 19]	485	0.21 ± 1.38	0.50	0.50 ± 0.40	–0.09 ± 1.27	0.998
[20, 24]	375	–0.26 ± 1.75	0.25	0.41 ± 0.46	–1.01 ± 1.74	0.993
[25, 29]	228	–0.11 ± 1.93	0.38	0.39 ± 0.52	–0.81 ± 2.05	0.994
[30, 34]	259	0.17 ± 1.58	0.38	0.53 ± 0.49	–0.34 ± 1.01	0.995
[35, 39]	271	0.18 ± 1.33	0.38	0.44 ± 0.41	0.02 ± 1.31	0.994
[40, 44]	265	0.19 ± 1.74	0.50	0.48 ± 0.54	–0.11 ± 1.65	0.996
[45, 49]	242	0.01 ± 1.69	0.38	0.44 ± 0.58	–0.46 ± 2.10	0.981
[50, 54]	226	0.31 ± 2.23	0.50	0.63 ± 0.56	0.41 ± 1.42	0.983
[55, 59]	132	0.68 ± 1.94	0.75	0.71 ± 0.66	1.04 ± 2.04	0.979
60+	288	0.80 ± 2.92	0.94	1.14 ± 1.00	0.01 ± 2.64	0.985
All	3576	0.34 ± 1.76	0.50	0.72 ± 0.42	0.43 ± 1.42	0.991

Fig. 2.

Mean value of the emmetropized peak in each of the age groups for the cycloplegic (red) and non-cycloplegic (black) data. Error bars represent the 95 % confidence interval of the mean.

(0.14MB).

The non-cycloplegic refractive error shows a similar behavior, with an emmetropic period between 20 – 40 years and a hypermetropic increase at 45 – 49 years. The latter reflects the loss in accommodation and the internal changes in lens gradient index that produce a loss of intrinsic lens power with age21,22 (Fig. 2).

Comparison between cycloplegic and non-cycloplegic refractive error

At younger ages there is a positive correlation between the difference between cycloplegic and non-cycloplegic refractive errors and the average of both (Fig. 3). This positive correlation is lost as the subjects become older and the difference between cycloplegic and non-cycloplegic refractive error decreases. There are, however, many subjects with big differences between both measurements.

Fig. 3.

Bland-Altman plots of each age group, showing the difference between cycloplegic and non-cycloplegic refractive error as a function of the average of both values.

(1.03MB).

This work used Bigaussian fits to improve the previous estimate17 of the changes in refractive errors in the Teheran Eye Study. The results clearly confirm that the cycloplegic refractive error in non-myopes between 20 – 50 years old is about +0.5D (Fig. 2), extending Morgan et al.’s3 report to this effect for children well into adulthood. Meanwhile, the non-cycloplegic refractive error remains at plano until 45 years due to latent accommodation.

The curves in Fig. 2 may be compared to the refractive error changes in the literature (Fig. 4). Among those, Sorsby's 1960 paper14 with the cyclopleged data for 20-year-old army recruits is especially important, reporting a peak “between 0 and +1D”. Interpreting this as the mean of both values, +0.5D, this closely agrees with the current results. Furthermore, the distribution of Sorsby's population closely matches that of the current data (Fig. 5). The cycloplegic refractive errors reported by Slataper15 appear very high overall, especially before 20 years of age, but show a stable period between 25 – 45 years at about +0.6D. Finally, the meta-regression curve by Rozema,2 based solely on cyclopleged reports from the literature, gives a value of +0.35D at 20 years, again agreeing with the current result. Four other studies were considered unsuitable for comparison with the current analysis: one study23 with a large myopic influence, and three others10–12 presenting non-cycloplegic, clinical populations with a large myopic influence.

Fig. 4.

Current results (red lines) compared to literature data.

(0.2MB).

The current analysis was performed using the Teheran Eye Study data, which are especially suited for this analysis as they were the only large-scale cycloplegic data in the age range of interest. Moreover, the TES data were collected 22 years ago before the peak of the myopia epidemic in Asia, leading to a relatively high proportion of emmetropes (Fig. 1). Previously, these data helped demonstrate that non-cycloplegic refraction cannot be reliably used to determine myopic and hypermetropic refractive errors due to accommodation, making cycloplegic refraction the gold standard for epidemiological and clinical studies.24 Consequently, recent efforts in myopia prevention25 focus on detecting children who are plano under cycloplegia26 as they lack a physiological hypermetropic reserve that protects them from developing myopia.27 The current results suggest that during normal refractive development the eye tries to preserve this cycloplegic hypermetropic reserve at +0.5D, making this the target of emmetropization by early adulthood. Note that this is not the same as a hypermetropic reserve under normal accommodative tonus without cycloplegia, where the refraction is near plano for distance viewing and accommodation provides clear near vision. Furthermore, it important to distinguish the current results for objective refraction from cycloplegic and non-cycloplegic subjective refraction that, in the hands of experienced clinicians, could lead to somewhat different refractive values.

Fig. 5.

Distribution of the spherical equivalent of Sorsby's 1960 data and the TES data.

(0.12MB).

Finally, a Bigaussian fitting method was used in this work to distinguish between the emmetropized and dysregulated sections of the population.20 This method is easily implemented in Matlab or R and may be of use in future studies to disregard the influence of myopia seen in some earlier studies based on global mean or average values.28