Prevalence and Associations of Anisometropia with Spherical Ametropia, Cylindrical Power, Age, and Sex in Refractive Surgery Candidates

Abstract

Purpose.: To analyze the prevalence and associations of anisometropia with spherical ametropia, astigmatism, age, and sex in a refractive surgery population.

Methods.: Medical records of 27,070 eyes of 13,535 refractive surgery candidates were reviewed. Anisometropia, defined as the absolute difference in mean spherical equivalent powers between right and left eyes, was analyzed for subjective (A_subj) and cycloplegic refraction (A_cycl). Correlations between anisometropia (>1 diopter) and spherical ametropia, cylindrical power, age, and sex, were analyzed using χ² and nonparametric Kruskal–Wallis or Mann–Whitney tests and binomial logistic regression analyses. Power vector analysis was applied for further analysis of cylindrical power.

Results.: Prevalence of A_subj was 18.5% and of A_cycl was 19.3%. In hyperopes, logistic regression analysis revealed that only spherical refractive error (odds ratio [OR], 0.72) and age (OR, 0.97) were independently associated with anisometropia. A_subj decreased with increasing spherical ametropia and advancing age. Cylindrical power and sex did not significantly affect A_subj. In myopes all explanatory variables (spherical power OR, 0.93; cylindrical power OR, 0.75; age OR, 1.02; sex OR, 0.8) were independently associated with anisometropia. Cylindrical power was most strongly associated with anisometropia. Advancing age and increasing spherical/cylindrical power correlated positively with increasing anisometropia in myopic subjects. Female sex was more closely associated with anisometropia.

Conclusions.: This large-scale retrospective analysis confirmed an independent association between anisometropia and both spherical ametropia and age in refractive surgery candidates. Notably, an inverse relationship between these parameters in hyperopes was observed. Cylindrical power and female sex were independently associated with anisometropia in myopes.

Anisometropia, which is defined as a difference in ocular refraction between both eyes, can be classified into physiological anisometropia and higher anisometropia, which can result in amblyopia and strabismus. In general, differences up to 1 diopter (D) are presumed to lie within the norm. ¹  

The correlation of refractive error between right and left eyes is high. ² The prevalence of anisometropia is 0 to 28%. ^3,4 Several studies have documented a positive association between the level of anisometropia and the degree of spherical ametropia, ^{5 –9} astigmatism, ⁵ or age. ^1,10  

Only Tong et al. ⁷ and Qin et al. ¹¹ have assessed to which extent anisometropia and spherical ametropia, astigmatism, and age are independently associated. This is an important issue, given that Guggenheim et al. ¹² reported an association between spherical and cylindrical refractive errors. In the study by Tong and colleagues, ⁷ multiple logistic regression was used to identify explanatory variables (risk factors) that were independently associated with the presence of anisometropia in Singaporean schoolchildren ranging in age from 7 to 9 years. They found that the presence of myopia (odds ratio [OR], 4.60), age (OR, 1.19 per year), and sex (OR, 1.19 in females) were all independently associated with anisometropia. Qin et al. ¹¹ showed that there is an independent association between anisometropia and both spherical ametropia and astigmatism in a group of 90,884 subjects attending optometry practices in the United Kingdom. 

Unlike general population-based studies, our study is an analysis of refractive surgery candidates. Our purpose was to examine the prevalence of anisometropia in a refractive surgery population and to explore its independent association with spherical ametropia, astigmatism, age, and sex. 

Methods

The study population consisted of individuals attending Care Vision refractive clinics in Germany and Austria between April 2006 and August 2010 for treatment of their ametropia. The refractive clinics were not selected according to defined epidemiologic sampling criteria. Due to the selection bias there is a deficit in both the number of pre- and teenage subjects (younger than 18 years) and older individuals (older than 65 years). Most of the subjects were candidates for undergoing refractive surgery with an excimer laser (either LASIK or PRK). Subjects exceeding the range for laser vision correction were candidates for phakic intraocular lens surgery or clear lens extraction. 

We examined the medical records of all cases in detail. A general ophthalmological examination had been performed for each case. In addition to general medical and ophthalmic histories, preoperative measurements included uncorrected distance visual acuity (UDVA), corrected distance visual acuity (CDVA), subjective refraction, cycloplegic refraction, tonometry, pupillometry, corneal topography, and pachymetry (Orbscan or Pentacam), slit-lamp examination of the anterior segment, and funduscopy. We converted all refractive data to the minus cylinder form to prevent confusion during analysis. Ethics approval for the study was granted by the local ethics committee. The research adhered to the tenets of the Declaration of Helsinki. 

Classification Criteria

The myopic group included only subjects with a spherical equivalent (SE) < 0 D, but also contained subjects with mixed astigmatism (and SE < 0; n = 395). The hyperopia group included subjects with SE > 0 D. We defined anisometropia as the absolute difference in SE powers between the right and the left eyes, because a recent analysis has demonstrated that classification of anisometropia on the basis of mean SE power (A_MSE) or spherical ametropia (A_SPH) has no effect on the significance of the test results. ¹¹ It was calculated separately for subjective and cycloplegic refraction. Each eye was categorized into one of four anisometropia severity groups. The four groups were defined as nonanisometropia (<1.00 D); mild anisometropia (1.00–1.99 D); moderate anisometropia (2.00–2.99 D); and severe anisometropia (≥3.00 D). Subjects were divided into one of six age categories: group 1, <20 years old; group 2, 20–29 years old; group 3, 30–39 years old; group 4, 40–49 years old; group 5, 50–59 years old; and group 6, ≥60 years old. In addition, analyses were also performed on only those subjects ranging in age from 20 to 40 years (inclusive). 

The vectorial approach (J₀, J₄₅) was first described by Thibos et al. ¹³ Advantages of the power vector representation of a spherocylinder lens for numerical and graphical analysis of optometric data are described for the statistical distribution of refractive errors. Therefore subjective refractions in conventional script notation [(S)phere, (C)ylinder, axis (α)] were converted to power vector coordinates by the following formulas: J₀ = (−C/2) cos (2α) and J₄₅ = (−C/2) sin (2α). 

Subjects were grouped according to MSE power in the less ametropic eye in 1.0-D intervals. In the nonvectorial method, the analysis was performed so as to limit the confounding between the magnitudes of spherical and cylindrical powers that occurs with power vector or power matrix methods. ¹²  

Statistical Analysis

After the data were compiled, they were entered into a spreadsheet program (Excel; Hamburg Refractive Data Base) and statistically analyzed using predictive analytic software (SPSS version 17.0; SPSS, Inc., Chicago, IL). To avoid unwanted bias, anisometropia was correlated with ametropia in the least ametropic eye (for further details see supplements in Qin et al. ¹¹ ) and only data for the less ametropic eye were included in this study. For statistical analysis, data description was based on medians and quartiles of the respective anisometropia calculations. Arithmetic means and SD were calculated for further descriptive comparison. Because of the nonnormal distribution of the data, nonparametric tests were used throughout the study. The χ² test or, depending on expecting values of the cells, the Fisher's Exact Test (expecting value <5) were used for comparisons of categorical data between groups. Due to the complexity of the Fisher's Exact Test it could technically not always be applied. The Kruskal–Wallis or Mann–Whitney tests were used for comparisons of continuous measures between groups. The two-sample Kolmogorov–Smirnov test was used to compare distributions. A value of P < 0.05 was considered statistically significant. 

Separate analyses were performed for subjective and cycloplegic refractive data. Because the cohort was divided into a large number of subgroups, only subgroups with a sample size of >30 subjects were included. For the logistic regression analysis, the presence or absence of anisometropia (>1.00 D) was used as the dependent variable, whereas age, spherical ametropia, and cylindrical ametropia were considered as continuous independent explanatory variables, along with sex as a binary independent variable. The regression models were computed separately for myopic and hyperopic subjects. To identify risk factors that were independently associated with anisometropia in this population of refractive surgery candidates, binomial logistic regression analysis was performed. Separate models were computed for subjects of all ages (age coded as a categorical variable) and for subjects younger than 40 years. These models were used to examine the associations in the whole sample and in subjects before the onset of cataract-induced refractive changes, as proposed by Qin et al. ¹¹ To evaluate these associations more closely, additional logistic regression models were computed in which age was analyzed as a categorical variable with 10-year intervals. In these groups, ORs for anisometropia were calculated for each decade compared with the preceding decade in subjects ranging in age from >20 to >60 years, with adjustment for spherical ametropia, astigmatism, and sex, whereas other variables were controlled for. 

Results

Demographics and Prevalence of Anisometropia

Complete data on the refractive state of both eyes of 13,535 subjects (27,070 eyes) were retrieved and enrolled into the statistical analysis. Mean overall age was 35.94 years (range, 18–74 years). In all, 12,097 myopic subjects with a mean age of 34 ± 9.31 years and 1438 hyperopic subjects with a mean age of 44.29 ± 11.69 years were analyzed. We first explored our data set based on subjective refraction, and then repeated calculations based on cycloplegic refraction. The median level of anisometropia (A_subj) in the whole study population was 0.37 D (0.12–0.75 D; 25th to 75th quartiles) and the prevalence of A_subj > 1 D was 18.5%. Mild anisometropia was observed in 13.9%, moderate anisometropia in 3.2%, and severe anisometropia in 1.4% of the subjects. The median level of A_cycl was 0.37 D (0.12–0.75) and mean overall prevalence of A_cycl > 1 D was 19.3%. 

Further analysis of myopic and hyperopic refractive candidates revealed an 18.7% prevalence of A_subj in myopic subjects and a 16.6% prevalence in hyperopic subjects. Clinical characteristics and descriptive statistics of parameters for the myopic and hyperopic candidates are shown in Table 1. 

Table 1.

 

Distribution of Anisometropia, Sex, Age, and Refractive State of the Study Population

Factor	Anisometropia
	<1		≥1		Total
	Number	%	Number	%	Number	%
Patients	11,031	81.5	2,504	18.5	13,535	100.0
Male	4,748	43.0	942	37.6	5,690	42.0
Female	6,283	57.0	1,562	62.4	7,845	58.0
Myopia	9,832	81.3	2,265	18.7	12,097	81.7
Hyperopia	1,199	83.4	239	16.6	1,438	18.3

	Range	Mean (±SD)	Range	Mean (±SD)	Min/Max	Mean (±SD)
Age, y	18–74	35.7 (±10.0)	20–68	37.0 (±9.7)	18–74	35.9 (±10.0)
Myopia
SE	−12.9/−0.13	−3.7 (1.9)	−11.5/−0.13	−0.39 (1.9)	−12.9/−0.13	−3.7 (1.9)
SubjectiveSph	−12.0/2.8	−3.2 (1.9)	−11.0/2.3	−3.4 (1.9)	−12.0/2.8	−3.3 (2.0)
SubjectiveCyl	−6.5/0.0	−0.9 (0.8)	−6.5/−0.13	−1.1 (0.96)	−6.5/0.0	−0.9 (0.9)
Hyperopia
SE	0.0/7.13	2.3 (1.2)	0.0/−0.13	1.9 (1.1)	0.0/7.1	2.2 (1.2)
SubjectiveSph	0.5/8.25	2.9 (1.2)	0.75/−0.13	2.5 (1.1)	0.5/8.25	2.8 (1.2)
SubjectiveCyl	−6.8/0.0	−1.25 (1.3)	−5.8/−0.13	−1.2 (1.2)	−6.8/0.0	−1.2 (1.2)

Anisometropia and Spherical Ametropia

The prevalence and severity of anisometropia in myopic and hyperopic subjects is shown in Figures 1A–F. Myopic and hyperopic subjects were (separately) grouped according to the spherical power in the less ametropic eye, in 1.0-D interval categories, as described by Qin et al. ¹¹  

Figure 1.

 

(A–F) Prevalence and severity of A_subj (A, B, D, E) and A_cycl (C, F) with the extent of spherical ametropia (mean spherical equivalent [MSE] power). Prevalence of anisometropia for myopes (A) and hyperopes (B, C), and severity of anisometropia for myopes (D) and hyperopes (E, F). The level of anisometropia severity is shown on a grayscale. Subjects are grouped in 1.00-D intervals of MSE power in the less ametropic eye.

To examine whether the degrees of anisometropia and ametropia were significantly associated, we compared the prevalence of anisometropia, and its severity, with the manifest and cycloplegic SE power in the subjects' less ametropic eye. For myopes, there was a roughly linear trend of increasing anisometropia prevalence and severity with increasing myopia up to a maximum of approximately −7.00 D. For very high myopia (< −8.00 D), we observed a decreasing tendency of anisometropia with increasing myopia (Fig. 1A). 

In hyperopes (> +1 D), we observed a decreasing prevalence of both A_subj (Fig. 1B) and A_cycl (Fig. 1C) with increasing ametropia. Of note, there were no subjects with hyperopia > +4 D showing moderate or severe A_subj (Fig. 1B). To investigate the relationship between anisometropia and spherical ametropia more quantitatively, we examined the median level and mean rank of anisometropia in each of the refractive error categories. A positive association was evident in myopes (Fig. 1D; Fisher's Exact Test, P < 0.001 and Kruskal–Wallis chi-square test: χ² = 86.371, degrees of freedom [df] = 9 with P < 0.001). In hyperopes, the lowest prevalence and corresponding lowest mean rank of anisometropia were observed in the > +4 D hyperopia subgroup. The correlation was statistically borderline significant (Fig. 1E; Pearson's chi-square test: χ² = 25.564, P = 0.017; Kruskal–Wallis: χ² = 12.378, df = 6, P = 0.02. Note that 30% of the cells had expected counts of <5). 

Anisometropia and Cylindrical Power

To further analyze astigmatic refractive error, both the vectorial (MSE, J₀, and J₄₅) and nonvectorial (spherical and cylindrical powers) approaches were used. In vectorial analyses, both the prevalence and severity of anisometropia varied with the level of astigmatism, but the distribution pattern differed for J₀ and J₄₅. For J₀, we observed an increasing prevalence and severity of anisometropia with increasing positivity of J₀ (Fig. 2A; Kruskal–Wallis test χ² = 72.684, df = 4, P < 0.001). For J₄₅, we found a symmetrical V-shaped prevalence of anisometropia with a minimum for J₄₅ between −1 D and +1 D (Fig. 2B; Kruskal–Wallis test χ² = 12.236, df = 3, P = 0.007). A relationship between anisometropia and cylindrical power was evident in nonvectorial analyses. In the eyes of myopes, the prevalence of anisometropia increased significantly as cylindrical power increased (Fig. 2C), as did the severity of anisometropia (Fig. 2G; Kruskal–Wallis test χ² = 154.121, df = 4, P < 0.001). In contrast, in the eyes of hyperopes, the prevalence of anisometropia did not increase significantly as cylindrical power increased (Fig. 2D). Furthermore, there was no significant increase in the severity of anisometropia with increasing cylindrical power in hyperopes. (Fig. 2H; Kruskal–Wallis test χ² = 4.700, df = 4, P = 0.319). 

Figure 2.

 

(A–H) Prevalence and severity of anisometropia (A_subj) in relation to the level of astigmatism. The prevalence and severity of anisometropia are represented in vectorial notation (A, B, E, F) and spherocylinder notation (C, D, G, H). Prevalence of anisometropia in all subjects with regard to their level of J₀ (A) and J₄₅ (B). Prevalence of anisometropia for myopes (C) and hyperopes (D). Anisometropia severity level is shown on a grayscale (grading: none, mild, moderate, and severe). (E–H) Level of anisometropia in the refractive population. Subjects are grouped in 1.00-D intervals of cylinder, J₀, and J₄₅ power in the subject‘s less ametropic eye.

Associations between Anisometropia and Age

The prevalence (Figs. 3A–C) and severity (Figs. 3D–F) of anisometropia increased significantly with increasing age with a peak in subjects aged 50 years, and then decreased with advancing age in this clinically selected group of refractive surgery candidates (Fig. 3A; Kruskal–Wallis test χ² = 58.426, df = 5, P < 0.001). 

Figure 3.

 

(A–F) Variations in the prevalence and severity of anisometropia (A_subj) in relation to age. Prevalence of anisometropia in the whole study population (A), in myopic (B), and hyperopic (C) subgroups. Anisometropia severity level is shown on a grayscale. Severity of anisometropia in the whole study population (D), in myopic (E), and hyperopic (F) subgroups.

In myopic subjects the prevalence of anisometropia was steadily increasing with advancing age with a peak in the sixth decade (Fig. 3B; Fisher's Exact Test, P < 0.001; Kruskal–Wallis test χ² = 75.405, df = 5, P < 0.001) and decreased in the seventh decade. In hyperopes the prevalence of anisometropia tended to decrease with increasing age up to the fifth decade. In the sixth and seventh decades this trend of negative correlation became more prominent (Fig. 3C; Pearson's test χ² = 33.313, df = 15, P = 0.004; Kruskal–Wallis test χ² = 22.678, df = 5, P < 0.001. Note that 37% of the cells had an expected count of <5). 

Associations between Anisometropia and Sex

The overall prevalence of anisometropia was higher in female (19.9%) than that in male (16.6%) subjects. The difference between the sexes was statistically significant (Pearson's χ² = 25.261, df = 3; Mann–Whitney test z = −4.1; for both tests P < 0.001). Although the median level of 0.37 D and distribution parameters of quartiles (Figs. 4C, 4D; Q25–Q75, 0.13–0.75) of anisometropia between male and female subjects were equal, the difference between mean ranks of the sex groups reached statistical significance due to higher anisometropia prevalence in female subjects. 

Figure 4.

 

(A–D) Variations in the prevalence and severity of anisometropia (A_subj) in relation to sex. Prevalence of anisometropia in myopic subjects (A) and in hyperopic subjects (B). Anisometropia severity level is shown on a grayscale. Level of anisometropia in myopes (C) and hyperopes (D).

Subgroup analysis revealed that in myopic subjects anisometropia prevalence was higher in female (20.2%) than that in male (16.6%) subjects (Fig. 4A; Pearson's χ² = 25.285, df = 3; Mann–Whitney test z = −4.2, for both tests P < 0.001), whereas in hyperopic subjects the prevalence of anisometropia was not significantly different between both sexes (Fig. 4B; Pearson's χ² = 1.16, df = 3, P = 0.725; Mann–Whitney test z = −2.17, P = 0.828). Logistic regression analysis revealed that female sex was independently associated with anisometropia in myopic subjects (for males compared with females, OR, 0.802; 95% CI, 0.728–0.883; P < 0.001). These results are shown in Table 2. The only effect reaching statistical significance in hyperopes was in 20- to 40-year-old subjects (for males compared with females: OR, 0.526; 95% CI, 0.318–0.871; P = 0.01), where the significance level would be borderline when taking into account multiple testing. 

Table 2.

 

Binomial Logistic Regression Analysis for the Presence of Anisometropia (>1.0-D Difference in MSE of Both Eyes) with Age Modeled as a Continuous Variable

Model*	Variable† ‡	Regression Coefficient	SE of Coefficient	Significance§	OR	95% CI of OR		k ‖
						Lower	Upper
Hyperopes	Spherical power	−0.329	0.069	<0.001	0.720	0.629	0.824	2.25
All ages (n = 1438)	Cylindrical power	0.025	0.067	NSD	1.026	0.900	1.169	¶
Nagelkerke R = 0.039	Age, y	−0.024	0.007	<0.001	0.976	0.964	0.989	47
	Sex, male	−0.074	0.144	NSD	0.929	0.701	1.231	¶
	Constant	0.394	0.396	NSD	1.483			—
Myopes	Spherical power	−0.072	0.012	<0.001	0.931	0.910	0.953	¶
All ages (n = 12,094)	Cylindrical power	−0.289	0.026	<0.001	0.749	0.712	0.788	−2.5
Nagelkerke R = 0.032	Age at TT	0.021	0.003	<0.001	1.021	1.016	1.026	53
	Sex, male	−0.221	0.049	<0.001	0.802	0.728	0.883	¶
	Constant	−2.633	0.109	<0.001	0.072			—
Hyperopes	Spherical power	−0.480	0.112	<0.001	0.619	0.497	0.770	1.5
Age 20 to 40 years (n = 446)	Cylindrical power	0.154	0.106	NSD	1.167	0.948	1.436	¶
	Age, y	0.007	0.020	NSD	1.007	0.967	1.048	¶
Nagelkerke R = 0.108	Sex, male	−0.643	0.257	0.013	0.526	0.318	0.871	¶
	Constant	0.311	0.762	NSD	1.364			—
Myopes	Spherical power	−0.076	0.014	<0.001	0.927	0.901	0.824	−7.00
Age 20 to 40 years (n = 8394)	Cylindrical power	−0.314	0.032	<0.001	0.731	0.686	0.824	−2.25
	Age, y	0.022	0.005	<0.001	1.022	1.012	0.824	¶
Nagelkerke R = 0.027	Sex, male	−0.198	0.061	0.001	0.820	0.728	0.824	¶
	Constant	−2.723	0.177	<0.001	0.066			—

 Age, spherical power, and cylindrical power were included in each model as continuous explanatory variables, along with sex as a categorical explanatory variable. Hyperopes were classified as subjects with a positive refractive error in the least plus meridian of both eyes. Myopes were classified as subjects with a negative refractive error in the least minus meridian of both eyes.

*  Goodness-of-model fit, given as Nagelkerke R.

†  Spherical and cylindrical power refer to the less ametropic of the two eyes.

‡  For myopes, both spherical and cylindrical powers are taken to be negative (thus giving increased ORs for increasing levels of myopia or astigmatism). For hyperopes cylindrical power was converted to minus cylinder format.

§  NSD, not significantly different.

‖  The value of the variable required to give a twofold increase/decrease in the likelihood of subjects being anisometropic.

¶  Note that not all variables exert sufficient influence to produce a twofold increase in the OR.

Logistic Regression Models

In hyperopic subjects only spherical refractive error and age were independently associated with anisometropia. Both parameters appeared to be inversely related to anisometropia; anisometropia decreased with increasing spherical ametropia (OR, 0.72/D) and advancing age (OR, 0.976/year; 95% CI, 0.964–0.989 in the full age range, but in the 20- to 40-year age group of hyperopes this trend was statistically not significant). A spherical refractive error of +2.25 D led to an approximately twofold decrease in the OR for anisometropia. Cylindrical power and sex did not significantly affect anisometropia in hyperopes. 

In myopic subjects all explanatory variables (spherical power, cylindrical power, age, and sex) were independently associated with anisometropia. Cylindrical power was most strongly associated with the severity of anisometropia; a cylindrical power of 2.5 D led to an approximately twofold increase in the OR for anisometropia. Advancing age (Table 2; OR, 1.021/year; 95% CI, 1.016–1.026, for the full age range of subjects; and OR, 1.022/year; 95% CI, 1.012–1.033, in the 20- to 40-year-old age group), increasing spherical (OR, 0.93/D; 95% CI, 0.91–0.953) and cylindrical power (OR, 0.75/D; 95% CI, 0.712–0.788) correlated positively with increasing anisometropia in myopic subjects. Table 2 lists the parameters of the logistic regression models, describing the associations between anisometropia and the explanatory variables spherical power, cylindrical power, age, and sex. 

The association between anisometropia and age appeared to be complex and to differ between myopic and hyperopic subjects. Between the ages of 30 and 49 years anisometropia in hyperopic subjects remained relatively stable (Fig. 3C), with a marginal tendency of decreasing A_subj. For hyperopes >50 years the probability of being anisometropic decreased significantly (OR, 0.502–0.336 per decade) with advancing age. 

In myopic refractive surgery candidates, in comparison, there was a steady increase in the OR with each decade of advancing age from the third to the fifth decades (OR, 1.394–1.607 per decade). This trend of increasing anisometropia with advancing age was also observed in younger (<30 years) and older (>60 years) myopic subjects, but logistic regression analysis did not reach the level of statistical significance for these decades (Table 3). 

Table 3.

 

Binomial Logistic Regression Analysis for the Presence of Anisometropia (>1.0-D Difference in MSE of Both Eyes) with Age Modeled as a Categorical Variable

Model*	Variable	Regression Coefficient	SE of Coefficient	Significance	OR	95% CI of OR
						Lower	Upper
Hyperopes	Spherical power	−0.340	0.069	<0.001	0.712	0.621	0.815
Age range, 20–99 y	Cylindrical power	0.004	0.065	NSD	1.004	0.883	1.142
Nagelkerke R = 0.05
	Sex, male	−0.049	0.145	NSD	0.952	0.717	1.264
	Age: 20–29 y	−0.596	0.697	NSD	0.551	0.140	2.162
	Age: 30–39 y	0.017	0.392	NSD	1.017	0.472	2.195
	Age: 40–49 y	−0.087	0.271	NSD	0.917	0.539	1.559
	Age: 50–59 y	−0.688	0.245	0.005	0.502	0.311	0.812
	Age: ≥60 y	−1.090	0.384	0.005	0.336	0.158	0.714
Myopes	Spherical power	−0.071	0.012	<0.001	0.931	0.910	0.953
Age range, 20–99 y	Cylindrical power	−0.289	0.026	<0.001	0.749	0.712	0.788
Nagelkerke R = 0.032
	Sex, male	−0.220	0.049	<0.001	0.803	0.729	0.884
	Age: 20–29 y	0.285	0.262	NSD	1.330	0.796	2.225
	Age: 30–39 y	0.332	0.137	NSD	1.394	1.065	1.825
	Age: 40–49 y	0.448	0.099	<0.001	1.566	1.290	1.899
	Age: 50–59 y	0.474	0.109	<0.001	1.607	1.299	1.988
	Age: ≥60 y	0.444	0.302	NSD	1.559	0.862	2.820

 Spherical power and cylindrical power were included in each model as continuous explanatory variables, along with age and sex as categorical explanatory variables. NSD, not significantly different.

*  Goodness-of-model fit, given as Nagelkerke R.

Subjective Refraction versus Cycloplegic Refraction

The prevalence of anisometropia (A_cycl > 1 D) in the whole study population was 19.3% compared with 18.5% for A_subj. Median severity for both A_cycl and A_subj was 0.37 D (0.12–0.75). Mild A_cycl was observed in 14.5% (A_subj 13.9%), moderate A_cycl was observed in 3.2% (A_subj 3.2%), and severe A_cycl in 1.6% (A_subj 1.4%) of the subjects. The prevalence of anisometropia (A_cycl) in myopic subjects was 19.5% and in hyperopic subjects 18.0%, being slightly higher than the prevalence of A_subj in myopic (18.7%) and hyperopic (16.6%) subjects (Figs. 1A, 1B). In summary, the trends and correlations between anisometropia and the analyzed variables were very similar for subjective and cycloplegic refraction with only minor differences. 

First, the observed effect in 20- to 40-year-old hyperopes, with female sex being independently associated with A_subj, in logistic regression analysis did not reach statistical significance, when calculated for cycloplegic refraction (OR, 0.801; 95% CI' 0.516–1.244; P = 0.323). Second, the correlation between increasing hyperopia (spherical ametropia) and decreasing anisometropia in the 20- to 40-year-old hyperopes (subjective refraction; Table 2) was not verified by cycloplegic refraction recalculation (OR, 0.868,; 95% CI, 0.754–1.001; P = 0.051). Third, in hyperopes age correlated negatively with A_cycl for subjects older than 60 years (compared with an age range of 50–59 years; OR, 0.289; 95% CI, 0.125–0.670,; P = 0.004). For A_subj statistical significance was also evident in hyperopes ranging in age from 50 to 59 years (compared with the preceding decade). Fourth, in myopes the observed continuous increase in the OR with each decade of advancing age from the third to the fifth decades for A_subj was expanded to the sixth decade when calculated for cycloplegic refraction (>60 years compared with the preceding decade; OR, 1.903; 95% CI, 1.069–3.387; P = 0.029). 

Discussion

To our best knowledge, the present study comprises the largest refractive candidate-based data on the prevalence and the severity of anisometropia and its correlation with spherical and cylindrical ametropia, age, and sex. To minimize a potential bias and to deliver better comparability with published data, we first analyzed our data set based on subjective refraction and then repeated calculations based on cycloplegic refraction. 

The major difference between our study and other population-based cross-sectional studies is the method of recruitment of subjects. We exclusively included refractive surgery candidates who had decided to attend a laser clinic to correct refractive errors. The clinical selection of subjects is thus biased toward persons with refractive errors, mainly at the expense of emmetropes. Furthermore, the number of pre- and teenage subjects (<18 years old) as well as older individuals (>70 years old) is underrepresented. The strength of our study includes a large sample size, homogeneity of the method of refraction (subjective and cycloplegic refraction), and a strict exclusion of ocular pathologies as a result of a detailed ophthalmological examination of the subjects. 

It is widely accepted that the prevalence of anisometropia is lower in school children than that in adults and differs between ethnic groups (for review see Weale ¹ ). Yamashita et al. ¹⁴ reported an anisometropia (SE difference > 1 D) prevalence of 3.1% in school children, whereas Tong et al. ⁷ reported an anisometropia (SE difference > 1.5 D) rate of 1.57% in school children. Another study by Wu et al. ¹⁵ revealed an anisometropia prevalence of 14.4% in Singaporean males ranging in age from 16 to 25 years, significantly higher than that observed in children. In the Blue Mountains Eye Study, Guzowski et al. ⁵ reported that anisometropia was present in 14.7% of the participants (3402 Australians older than 50+ years), with moderate to high anisometropia present in 2.1% of the subjects. The authors were able to further restrict their analysis to phakic subjects, excluding potential effects of intraocular lens implant power mismatches. Thus, the 18.5% prevalence of A_subj (mild 13.9%; moderate 3.2%; severe 1.4%) and 19.3% prevalence of A_cycl in our cohort of refractive surgery candidates in central Europe is higher than that reported for the general population of developed countries, ^5,8 but comparable to the approximately 17% prevalence of anisometropia reported by Qin et al. ¹¹  

Although Guzowski et al. ⁵ and Tong et al. ⁷ did not observe a sex difference with regard to anisometropia, Qin et al. ¹¹ found that the prevalence of anisometropia was higher in females compared with that in males, with the difference just reaching statistical significance (P = 0.02). We found a statistically significant higher prevalence of anisometropia in female subjects, in particular female myopic candidates (Fig. 4A). In addition our analysis suggests that the female sex is independently associated with anisometropia in myopic subjects (P < 0.001), but since the median level of anisometropia between the two sexes was equal (0.37 D) this effect of higher prevalence in female subjects is not clinically relevant. 

Differences in study subjects (general cross-sectional, population-based ⁵ vs. selected subjects ¹¹ and our study), differences in definition of anisometropia (SE difference > 1.5 D ⁷ vs. SE difference > 1 D ¹¹ and our study) and differences in recruiting refractive data (spectacle prescriptions, ¹² cycloplegic autorefraction, ⁷ subjective refraction ⁹ ) must be considered when comparing our results with other published data. 

Weale ¹⁰ has summarized data on the age-related prevalence of anisometropia from a comprehensive meta-analysis of the literature, and noted a positive linear relationship with the prevalence increasing by approximately 1.4% for each decade increase in age. Qin et al. ¹¹ further explored the association between anisometropia and spherical ametropia, astigmatism, age, and sex in a group of 87,759 subjects attending optometry practices in the United Kingdom. One drawback of the Qin study is the inability to exclude subjects who present anisometropia as a result of unilateral cataract extraction or untreated cataracts, because detailed ophthalmological examination was not available. In contrast, information related to ocular history, subjective and cycloplegic refraction, and detailed slit-lamp examination were available for the 13,535 subjects included in the present study. Because the data set of Qin et al. ¹¹ comprised only spectacle prescriptions, it is probable that the full extent of myopia might have been overestimated in some younger subjects, due to incomplete relaxation of accommodation, or that some younger hyperopes were prescribed only a partial correction. Both of these events would be less likely to occur in adults, due to the decline in the amplitude of accommodation with age. 

An association between anisometropia and astigmatism has been reported previously. ^5,9 However, this association has to be interpreted with caution due to a potential bias as a consequence of the use of spherical equivalent anisometropia. ¹⁶ Therefore, we used logistic regression analysis to control for the individual effects of spherical ametropia and astigmatism (explanatory variables) on anisometropia and to reveal an independent association of each contributing factor. Qin et al. ¹¹ were the first to show an independent association between anisometropia and both spherical ametropia and astigmatism. Our study confirmed the independent association between anisometropia and both spherical ametropia and astigmatism in myopes. Logistic regression analysis was used to control for the effects of all explanatory variables. In contrast to the study by Qin et al., ¹¹ the prevalence and level of anisometropia in hyperopic subjects of our study did not increase as spherical (Fig. 1B) and cylindrical power (Fig. 2D) increased and we did not find a positive linear association between increasing anisometropia and increasing ametropia. In turn our data revealed a decreasing prevalence (Figs. 1B, 1C) and severity (Figs. 1E, 1F) of anisometropia with increasing hyperopic ametropia. 

In summary, unlike other studies, ^5,11 we did not observe a remarkable symmetry in the prevalence of anisometropia as refractive errors diverged from zero. 

Exclusive selection of candidates who are willing to undergo refractive surgery and the limited number of hyperopic subjects (n = 1438) resulting in sampling variation for small hyperopic subgroups might limit the validity of our findings. Young hyperopic subjects with good UDVA and accommodative capacity do not usually visit refractive clinics to correct refractive error. Anisometropic low hyperopes may decide to visit a clinic for laser vision correction, whereas nonanisometropic low hyperopes with good accommodative capacity may not. This resulting selection bias could partly explain the relative abundance of anisometropic hyperopic subjects (<40 years) in our data set and the decreasing prevalence and severity of anisometropia with increasing hyperopia. The possible overrepresentation of anisometropic young hyperopes could, in turn, potentially mask a relationship between increasing anisometropia and increasing age in hyperopic subjects. 

Although age-related accommodative failure is largely attributed to peripheral lenticular changes, neurosenescence may explain the increased prevalence of anisometropia with advancing age in myopic subjects. ^10,17 This in turn could also affect the outcome of laser refractive surgery at advanced ages (>45 years). Since it is widely accepted that the prevalence of myopia is increasing over the years ^18,19 and that anisometropia is positively correlated with myopia, ¹¹ future longitudinal studies covering all age groups and focusing on the impact of anisometropia on progression of refractive errors will be of great interest. If future longitudinal studies will confirm the observed association between anisometropia/astigmatism and age, refractive surgery for myopia in prepresbyopic subjects may achieve emmetropia for only a limited period. 

Conclusion

Our retrospective analysis in a cohort of laser refractive candidates provides evidence for an independent association between anisometropia and spherical ametropia and age. We and others ¹¹ found a strong association between anisometropia and age, although our detailed evaluation suggested that the relationship differed in myopic and hyperopic subjects. 

In contrast to previous reports, ¹¹ we observed an inverse relationship between anisometropia and spherical ametropia and age in hyperopic subjects: anisometropia decreased with increasing spherical ametropia and advancing age. Cylindrical power and sex did not significantly affect anisometropia in hyperopes. 

In myopic subjects all explanatory variables were independently associated with anisometropia. Advancing age and increasing spherical/cylindrical power correlated positively with increasing anisometropia. Cylindrical power was most strongly associated with anisometropia in myopic subjects. Female sex was more closely associated with anisometropia, but to a clinically insignificant extent. 

Footnotes

 Disclosure: S.J. Linke, Care Vision Germany GmbH (E); G. Richard, None; T. Katz, Care Vision Germany GmbH (E)

The authors thank the staff and patients of Care Vision for their support in establishing the anonymized refractive data collection (Hamburg Refractive Data Base) and Vasyl Druchkiv for supporting the database and expert statistical analysis.