Characteristics and Related Parameters of Quick Contrast Sensitivity Function in Chinese Ametropia Children

Abstract

Purpose: 

To investigate the characteristics of quick contrast sensitivity function (qCSF) and its related parameters in Chinese ametropia children.

Methods: 

This case series study enrolled 106 eyes of 53 children (male/female=29/24, age: 9.04±2.06 years). Examinations included manifest refraction, axial length, corneal curvature, and monocular and binocular qCSF readings without refractive correction (area under log CSF [AULCSF], CSF acuity, and contrast sensitivity [CS] at 1.0 to 18.0 cpd). The subjects were divided into groups according to age and refractive parameters for analysis.

Results: 

The mean spherical equivalent (SE), AULCSF, and CSF acuity of the test eyes were −0.94±1.53 D, 0.44±0.33, and 8.50±5.97 cpd, respectively. In the monocular qCSF comparison, the refraction sphere (RS) was the major factor correlated with qCSF readings (B=0.186, P=0.009 for AULCSF; B=0.543, P=0.019 for CSF acuity; generalized linear model). The three groups stratified by RS/SE (<−1.00D, −1.00D to 0D, and >0D) showed significant differences in CS at medium spatial frequencies (3.0 and 6.0 cpd; all P<0.05). In the low RS/SE group (within −1 to 0 D), the CS at 12.0 cpd was significantly lower than that in the hyperopia group (all P<0.05). Binocular qCSF readings were significantly correlated with those of the eyes with lower RS (all P<0.05).

Conclusion: 

RS and SE are the major contributing factors of qCSF without refractive correction in children. The CS at medium spatial frequencies decrease significantly as the RS/SE increase. In low myopia children, the CS at medium and high spatial frequencies are significantly decreased, providing practical value in visual function screening in children.

In children who are in the critical period of visual development, amblyopia can result from disturbances in the normal development of visual pathways.¹ Approximately 19 million children worldwide have visual impairment caused by retinal disease, glaucoma, cataracts, or uncorrected refractive errors.² These diseases require early screening and diagnosis³; otherwise, they could lead to a decreased quality of life and increased socioeconomic burden.⁴ Currently, visual acuity testing is the most commonly used method for visual health screening in children. It assesses the ability to distinguish fine details at maximum contrast using the Snellen or Early Treatment of Diabetic Retinopathy Study (ETDRS) charts, ⁵ instead of those with different sizes at different contrasts. The contrast sensitivity test measures the contrast threshold at different spatial frequencies, describes the differences in patients with the same visual acuity, ⁶ and is considered a more comprehensive method for assessing visual function.⁷ A decrease in contrast sensitivity function (CSF) readings can be observed in some pathologic conditions, such as preterm labor, ⁸ amblyopia, ⁹ early glaucoma,⁶ and multiple sclerosis.¹⁰

The contrast sensitivity tests commonly used in clinical practice include alphabetic charts, Pelli–Pobson charts, and raster chart CSV-1000 tests, but these can only test at limited spatial frequencies.^11,12 In addition, the clinical use of laboratory CSF tests is limited by their duration of 30 to 60 min per test.¹³ The quick CSF (qCSF) test program has recently been demonstrated to be accurate and reliable.^14–16 Based on the Bayesian adaptive algorithm,¹⁷ qCSF tests acquire the CSF curve and the corresponding parameters using 10 digits within 25 stimuli,¹² which further improved the feasibility for clinical application. The qCSF test has broadly been used for assessment in amblyopia treatment,¹⁸ visual quality,^19,20 binocular visual function,²¹ peripheral vision,²² and retinal function disorder.^23,24

Previous studies have shown that age is the primary factor affecting CSF in adults with optimal refraction correction,^25,26 whereas there is a weak correlation between CSF and age in children with refractive correction.²⁷ However, the characteristics and correlated parameters of qCSF are still unclear in children with unaided acuity. Previous findings suggested that even a 0.25D undercorrection significantly affected the qCSF result.¹⁹ Therefore, assessment of the characteristics and correlated parameters of qCSF in ametropia children without refraction correction will help provide a reference for the promotion of qCSF testing. Thus, this study aimed to explore the distribution and related factors of qCSF in Chinese children to further assist optometrists in evaluating pediatric visual health. Toward this goal, qCSF readings in children under unaided acuity were evaluated according to age, sex, and refraction.

PATIENTS AND METHODS

This case series study was approved by the Ethics Committee of the Eye and ENT Hospital of Fudan University (2,020,107, date: July 1, 2021) and was conducted according to the tenets of the Declaration of Helsinki. Informed consent was obtained from all the participants and their legal guardians.

This study enrolled children aged 6 to 14 years old who visited the Eye and ENT Hospital of Fudan University between November 2021 and January 2022. The exclusion criteria were (1) history of orthokeratology lenses or low-concentration atropine and other drug use and (2) history of ophthalmic disease, surgery or trauma; history of systemic diseases; and severe psychological or psychiatric diseases.

Examinations

For axial length and corneal curvature measurements, a Humphrey IOL Master700 (Carl Zeiss Meditec, Germany) was used. For cycloplegia, the tropicamide phenylephrine eye drops (Mydrin-P ophthalmic solution; Santen, Osaka, Japan) were applied five times at 5-min intervals. The RT-5100 phoropter (Nidek Technologies, Japan) was used to assess manifest refraction and corrected distance visual acuity evaluation under full cycloplegia (absence of light reflex), 30 min after the last application of the eye drops.

Quick Contrast Sensitivity Function Test

An NEC P403 monitor (Gension & Waltai Digital Video System Co, Ltd., China) was used to display numbers (from 0 to nine in sloan fonts) as visual stimulus,²⁸ with a resolution of 1920×1,080 pixels, a display area of 116.84×77.89 cm, a maximum and standard brightness of 700 and 550 cd/m², respectively, and a contrast ratio of 4,000:1 (Fig. 1). Using a Bayesian adaptive procedure, the qCSF test displayed 25 stimuli at different spatial frequencies on the screen, each with three numbers at a different contrast ratio. The test was conducted under full cycloplegia without refractive correction. Patients view the visual stimuli horizontally at 3 m in a mesopic environment and two eyes were examined separately with the contralateral one covered. They were taught to report any numbers they saw on the screen, and the tester input their responses in a pad immediately: correct answers, incorrect answers, or no numbers observed (Fig. 1B). The CSF curve was depicted directly by the computer, and the results include area under log CSF (AULCSF), CSF acuity, and contrast sensitivity (log units) at six different spatial frequencies (1.0, 1.5, 3.0, 6.0, 12.0, 18.0 cycle per degree [cpd]) (Fig. 1C). The general spatial visual representation is described by the AULCSF, whereas CSF acuity reflects the cut-off spatial frequencies of CSF curves.^29,30

FIG. 1.: Testing interface of qCSF, (A) Three filtered digits are displayed on the screen as visual stimuli. (B) The technician's terminal is displayed on a tablet. (C) The qCSF results.

Statistical Analysis

Continuous variables were presented as the mean±SD and range, whereas categorical variables were presented as frequency. The Kolmogorov–Smirnov test was used for normality analysis. Levene test was used for the homogeneity of the variance test. The Student t test was used to evaluate the differences in normally distributed continuous variables between two groups. One-way analysis of variance was used to evaluate differences in normally distributed continuous variables between three or more groups, and the corrected Bonferroni method was used for multiple post-hoc comparisons. The Wilcoxon signed-rank test was used to verify differences in non-normally distributed data. Pearson correlation coefficient was used to evaluate the correlation between the continuous variables. A generalized linear model was used as a multifactor parameter test method, excluding the influence of various factors (age, sex, visual parameters, and ocular biology parameters) and binocular inclusion.

In the subgroup analysis, the subjects were stratified according to age (6–8 years and 9–14 years), refraction sphere (RS) (high RS [−4.0 to −1.25 D], low RS [−1.0 to 0 D], and hyperopia [0.125–2.0 D]), refraction cylinder (RC) (high RC [<−1.0 D], low RC [−0.75 to −0.25 D], and no astigmatism [0 D]), and spherical equivalent (high SE [−5.0D to −1.125 D], low SE [−1.0 to 0 D], and hyperopia [0.125–1.75 D]). The −1.00 D was selected to distinguish between groups, because it was one of the indications for spectacle prescription. A better eye or a lower RS/RC/SE eye was defined as an eye with a refraction error closer to hyperopia or 0 than that of the other eye. All statistical analyses were performed using the Statistical Package for the Social Sciences (version 25.0; SPSS, Inc., Chicago, IL). Statistical significance was set at P<0.05.

RESULT

Patient Characteristics

A total of 106 eyes of 53 subjects were included in the study. The patient characteristics are shown in Table 1. Age and spherical equivalent (SE) distributions are shown in Figure 2A. All examinations were completed successfully, with a data loss of less than 5%. The mean values of qCSF parameters under unaided refraction are shown in Figure 2B.

TABLE 1. - Patient Demographics

Characteristic	Mean±SD	Range
Age (years)	9.04±2.06	6, 14
Gender (male/female)	29/24
Axial length (mm)	23.98±0.83	20.25, 25.82
Refraction sphere (D)	−0.65±1.47	−4.00, 2.00
Refraction cylinder (D)	−0.59±0.68	−3.50, 0
Spherical equivalent (D)	−0.94±1.53	−5.00, 1.75
K-flat (D)	42.67±1.52	38.50, 49.00
K-steep (D)	43.97±1.53	39.00, 50.50
Km (D)	43.32±1.49	38.88, 49.75
CDVA (LogMAR)	0.01±0.02	0, 0.15

CDVA, corrected distance visual acuity; D, diopter; K-flat, flat keratometry; Km, mean keratometry; K-steep, steep keratometry.

FIG. 2.: Subject characteristics of age, refraction, and qCSF. (A) Distribution of age and spherical equivalents. (B) Mean contrast sensitivity (log units) at different spatial frequencies (left panel). Mean area under the line of the contrast sensitivity function (AULCSF) and contrast sensitivity function (CSF) acuity (right panel).

Correlation Analysis

Table 2 presents the Pearson correlations between qCSF parameters and other factors, including age and refraction. Age was negatively correlated with AULCSF and CS at 6.0 cpd (r=−0.202, −0.251; P<0.05). GLM analyses demonstrated that RS was the major factor affecting the change in qCSF parameters. Positive correlations were observed between the RS and AULCSF (r²=0.679; P<0.01) and between the RS and CSF acuity (r²=0.543; P<0.05) (Fig. 3).

TABLE 2. - Association Between qCSF (Quick Contrast Sensitivity Function) Parameters With Other Factors in Pearson Correlation Analysis

qCSF	AULCSF	CSF Acuity	CS (1.0 cpd)	CS (1.5 cpd)	CS (3.0 cpd)	CS (6.0 cpd)	CS (12.0 cpd)	CS (18.0 cpd)
Age
r	−0.202	−0.130	−0.021	−0.085	−0.182	−0.251	−0.169	−0.170
P	0.038	0.184	0.832	0.386	0.062	0.009	0.084	0.081
RS
r	0.824	0.737	0.707	0.807	0.850	0.749	0.427	0.241
P	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.013
RC
r	0.143	0.158	0.083	0.075	0.108	0.162	0.225	0.174
P	0.145	0.107	0.397	0.445	0.270	0.097	0.020	0.075
SE
r	0.821	0.741	0.696	0.790	0.839	0.754	0.459	0.270
P	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.005
AL
r	−0.430	−0.362	−0.379	−0.439	−0.479	−0.385	−0.097	−0.085
P	0.000	0.000	0.000	0.000	0.000	0.000	0.341	0.406
Km
R	−0.331	−0.339	−0.204	−0.239	−0.301	−0.324	−0.354	−0.215
P	0.001	0.001	0.051	0.022	0.004	0.002	0.001	0.040

Values with statistical significance are shown in bold.

AL, axial length; AULCSF, area under log CSF; cpd, cycle per degree; CS, contrast sensitivity; Km, mean keratometry; RC, refraction cylinder; RS, refraction sphere; SE, spherical equivalent.

FIG. 3.: Correlations between RS and qCSF parameters. Simple linear regressions between RS and (A) AULCSF, (B) CSF acuity, and CS (log units) at (C) 1.0 c/d, (D) 1.5 c/d, (E) 3.0 c/d, (F) 6.0 c/d, (G) 12.0 c/d, and (H) 18.0 c/d. The solid lines illustrate the line of linear regressions, and the dotted line illustrates the confidential intervals. Top left: equation of linear regression, coefficient of determination (r²), and P values of generalized linear models. The significance of r² is shown by asterisks (*** P<0.001; ** P<0.01; * P<0.05).

Group Analysis

As shown in Table 3 and Figure 4, CS at medium spatial frequencies (3.0 and 6.0 c/d) were significantly higher in the 6 to 8-year-old group than in the 9 to 14-year-old group (0.71±0.48 vs. 0.57±0.42 at 3.0 cpd, 0.42±0.40 vs. 0.28±0.34 at 6.0 cpd, all P<0.05) (Fig. 4A). The differences in qCSF parameters among the three RS groups were similar to those among the three SE groups, in which CS at medium spatial frequencies (3.0 cpd and 6.0 cpd) changed in a gradient (Fig. 4B, D). In the low SE group (SE within −1 to 0 D), the CS at low spatial frequencies was not significantly different than that in the hyperopia group (1.09±0.16 vs. 1.13±0.16 at 1.0 cpd; 1.05±0.16 vs. 1.17±0.15 at 1.5 cpd; P>0.05), whereas the CS at high spatial frequencies were significantly lower than that in the hyperopia group (0.07±0.13 vs. 0.23±0.25, P<0.05).

TABLE 3. - Comparison of qCSF (Quick Contrast Sensitivity Function) Readings in Different Groups

Groups	n	AULCSF	CSF Acuity	CS (1.0 cpd)	CS (1.5 cpd)	CS (3.0 cpd)	CS (6.0 cpd)	CS (12.0 cpd)	CS (18.0 cpd)
Ages (years)
6∼8	48	0.50±0.36	9.15±6.36	0.92±0.38	0.87±0.43	0.71±0.48	0.42±0.40	0.11±0.20	0.02±0.05
9∼14	58	0.39±0.30	7.96±5.62	0.90±0.35	0.82±0.38	0.57±0.42	0.28±0.34	0.06±0.13	0.01±0.03
RS (D)
−4.0∼−1.25	38	0.11±0.14	3.20±2.75	0.57±0.37 a	0.43±0.34 a	0.16±0.23	0.02±0.10	0.00±0.03	0.00±0.00
−1.0∼0	37	0.49±0.23	9.39±4.89	1.08±0.18	1.01±0.20	0.75±0.31	0.34±0.30	0.06±0.13	0.01±0.03
0.125∼2.0	31	0.78±0.20	13.93±4.44	1.12±0.17	1.14±0.16	1.08±0.18	0.73±0.28	0.21±0.23 a	0.03±0.07
RC (D)
−4.0∼−1	27	0.36±0.28	6.99±4.70	0.83±0.36	0.77±0.37	0.54±0.43	0.25±0.31	0.03±0.06	0.00±0.00
−0.75∼−0.25	40	0.45±0.33	8.58±5.61	0.98±0.30	0.88±0.39	0.64±0.46	0.35±0.37	0.07±0.16	0.01±0.04
0	39	0.49±0.36	9.45±6.96	0.88±0.41	0.85±0.44	0.69±0.46	0.40±0.41	0.13±0.21	0.02±0.06
SE (D)
−5.0∼−1.125	43	0.12±0.13	3.30±2.63	0.61±0.38 a	0.46±0.35 a	0.17±0.23	0.02±0.09	0.00±0.03	0.00±0.00
−1∼0	37	0.55±0.20	10.36±4.57	1.09±0.16	1.05±0.16	0.84±0.24	0.41±0.29	0.07±0.13	0.01±0.03
0.125∼1.75	26	0.81±0.20	14.43±4.45	1.13±0.16	1.17±0.15	1.11±0.17	0.78±0.26	0.23±0.25 a	0.03±0.07

Values with statistical significance between all (two/three) groups are shown in bold.

^aThere is no statistical significance between two genders, vs other two groups, P<0.05.

RS: refraction sphere; RC: refraction cylinder; D: diopter; SE: spherical equivalent; AULCSF: Area Under Log CSF; cpd: cycle per degree.

FIG. 4.: Analyses of qCSF parameters by groups of age and refraction. (A) Differences in qCSF parameters among age groups. (B) Differences in qCSF parameters in the groups of refraction spheres. (C) Differences in qCSF parameters in groups of refraction cylinders. (D) Differences in qCSF parameters in groups with spherical equivalents.

Binocular Quick Contrast Sensitivity Function Readings

Table 4 displays the correlations between binocular qCSF parameters and other factors as analyzed using GLM. Figure 5 depicts the distribution of binocular qCSF parameters and associated factors. The RS of the better eye was positively correlated with binocular AULCSF, CSF acuity, and CS at medium spatial frequencies (3.0 and 6.0 c/d) (B=0.200, 4.015, 0.261, 0.262; P<0.05). Furthermore, the SE of the better eye showed similar correlations with the corresponding qCSF parameters (B=0.181, 3.826, 0.224, 0.241; P<0.05). At low spatial frequencies, age was positively correlated with CS at 1.0 cpd (B=0.046, P<0.05), whereas the RS of the better eye was positively correlated with CS at 1.5 cpd (B=0.133, P<0.05).

TABLE 4. - Correlation Between Binocular qCSF Readings With Other Factors Analyzed With Generalized Linear Model

Parameters	AULCSF		CSF Acuity		CS (1.0 cpd)		CS (1.5 cpd)		CS (3.0 cpd)		CS (6.0 cpd)		CS (12.0 cpd)		CS (18.0 cpd)
	B	P	B	P	B	P	B	P	B	P	B	P	B	P	B	P
Ages (years)	0.005	0.759	0.363	0.293	0.046	0.010	0.033	0.053	0.006	0.737	−0.009	0.639	0.005	0.739	−0.001	0.840
RS
Higher	0.028	0.567	−0.095	0.931	0.085	0.133	0.066	0.225	0.021	0.725	0.003	0.953	0.032	0.488	0.019	0.304
Lower	0.200	<0.001	4.015	0.001	0.062	0.314	0.133	0.025	0.261	<0.001	0.262	<0.001	0.073	0.150	0.004	0.832
RC
Higher	−0.264	0.233	−5.927	0.150	−0.295	0.121	−0.269	0.211	−0.232	0.402	−0.338	0.180	−0.146	0.315	−0.048	0.347
Lower	0.393	0.123	9.283	0.051	0.311	0.156	0.297	0.231	0.327	0.306	0.531	0.068	0.269	0.108	0.082	0.162
SE
Higher	0.033	0.507	−0.045	0.966	0.083	0.148	0.068	0.235	0.034	0.587	0.012	0.835	0.029	0.508	0.010	0.580
Lower	0.181	0.001	3.826	0.001	0.044	0.482	0.109	0.081	0.224	0.001	0.241	<0.001	0.075	0.123	0.014	0.461
AL
Higher	−0.112	0.556	−1.427	0.693	−0.168	0.322	−0.142	0.437	−0.097	0.674	−0.075	0.736	−0.116	0.382	−0.063	0.186
Lower	−0.086	0.660	−2.522	0.495	0.040	0.816	−0.031	0.868	−0.159	0.501	−0.158	0.490	0.042	0.761	0.046	0.344
Km
Higher	−0.081	0.155	−1.055	0.307	−0.088	0.092	−0.095	0.091	−0.118	0.092	−0.077	0.247	−0.025	0.518	−0.012	0.353
Lower	−0.021	0.744	−0.582	0.620	0.057	0.341	0.030	0.638	−0.004	0.964	−0.045	0.551	−0.034	0.441	−0.004	0.800

Values with statistical significance are shown in bold. Higher/Lower: Eyes with Higher/Lower RS, RC, SE, AL or Km of each patient.

AL, axial length; AULCSF, area under log CSF; B, Partial Regression coefficients; CS, contrast sensitivity; cpd, cycle per degree; RS, refraction sphere; RC, refraction cylinder; SE, spherical equivalent; qCSF, quick contrast sensitivity function.

FIG. 5.: Correlations between binocular qCSF parameters and age or refractions. Simple linear regressions show the relationships between (a) binocular AULCSF and RS (lower) or SE (lower); (B) between binocular CSF acuity and RS (lower) or SE (lower); (C) between binocular CS (1.0 c/d) and age; (D) between binocular CS (1.5 c/d) and RS (lower); (E) between binocular CS (3.0 c/d) and RS (lower) or SE (lower); and (F) between binocular CS (6.0 c/d) and RS (lower) or SE (lower). Lower: Eyes with lower RS or SE in each patient. The regression equations are shown at the top left and bottom right. The thick lines show the lines of linear regression, and the thin lines show the lines of confidence intervals.

As shown in Table 5, the binocular qCSF parameters were significantly different from the corresponding monocular parameters. There were positive correlations between all binocular qCSF parameters and the corresponding parameters of eyes with lower RS. Although other qCSF parameters were positively correlated, binocular CS at 1.0 cpd was not correlated, and CS at 18.0 cpd was negatively correlated with that of eyes with higher RS (B=−2.149, P<0.05). The ratio of binocular AULCSF and CSF acuity to that of eyes with lower RS was 1.34±0.47 and 1.18±0.28, and the corresponding ratio was 2.54±3.7 and 1.23±0.33 to that of eyes with higher RS.

TABLE 5. - Difference and Correlation Between Binocular qCSF Readings With qCSF Readings of Stratified RS Groups

Parameters	AULCSF	CSF Acuity	CS (1.0 cpd)	CS (1.5 cpd)	CS (3.0 cpd)	CS (6.0 cpd)	CS (12.0 cpd)	CS (18.0 cpd)
Binocular	0.65±0.37	11.78±7.04	1.12±0.32	1.08±0.36	0.88±0.46	0.56±0.44	0.19±0.25	0.04±0.09 a
RS
Higher	0.38±0.34	7.25±5.54	0.85±0.39	0.76±0.43	0.55±0.46	0.28±0.37	0.07±0.16	0.01±0.04
Δ b	0.26±0.19	4.42±4.9	0.26±0.26	0.3±0.23	0.31±0.24	0.27±0.28	0.12±0.18	0.03±0.08
B	0.491	0.966	0.656	0.657	0.761	0.714	0.890	−2.149
P	0.038	0.004	0.127	0.013	0.044	< 0.001	0.014	0.043
Lower	0.5±0.32	9.75±6.17	0.97±0.33	0.92±0.36	0.71±0.44	0.4±0.37	0.1±0.17	0.01±0.04
Δ c	0.14±0.1	1.89±3.12	0.14±0.23	0.14±0.16	0.16±0.15	0.15±0.16	0.09±0.14	0.02±0.07
B	0.685	0.796	0.537	0.305	0.427	0.865	0.885	0.986
P	<0.001	< 0.001	0.031	<0.001	0.027	<0.001	<0.001	<0.001

Higher/Lower: Eyes with Higher/Lower RS of each patient.

Values with statistical significance are shown in bold.

^aBinocular CS (18.0c/d) vs. that of higher/lower RS group, P=0.011 and 0.012 respectively, other binocular qCSF readings vs that of higher/lower RS group, P<0.001.

^bBinocular qCSF readings minus that of higher RS group.

^cBinocular qCSF readings minus that of lower RS group.

AULCSF, area under log CSF; cpd, cycle per degree; CS, contrast sensitivity; RS, refraction sphere; RC, refraction cylinder; SE, spherical equivalent; B, partial regression coefficients; qCSF: quick contrast sensitivity function.

DISCUSSION

Normal visual experience is of great significance to the visual development of children.¹ The contrast sensitivity test presents sensitivity thresholds at a series of spatial frequencies and provides more information regarding visual functions.^6,7 The recently developed qCSF test is highly efficient in measuring a complete CSF curve with reliable accuracy^14–16 and has great potential to be applied as a screening method for visual health in children.

This study found a lower AULCSF value in children aged 6 to 14 years than in previous studies.¹⁹ In a study of individuals aged 17 to 34 years, the median AULCSF was 1.33 with 0.25 D undercorrection, and the median CSF acuity was 22.38 cpd.¹⁹ In adults with optimal refraction distance correction, the AULCSF and CSF acuity were 1.11 to 1.46 and 16.99 to 26.37 cpd, respectively.^19,25 In healthy fellow eyes of amblyopia children with optimal refraction correction (median age: 7.7 years), the median AULCSF ranged from 1.50 to 1.57.¹⁸ Discrepancies in the study subjects and testing methods may have contributed to the differences in results among studies. The current study included more myopia children, and the qCSF test was performed without refractive correction, but uncorrected refractive errors significantly impaired qCSF parameters.¹⁹ In addition, the pupil diameter increased because of cycloplegia. Therefore, an increase in intraocular higher-order aberrations would further decrease contrast sensitivity readings.^31,32

The current study showed that RS was the primary factor influencing the qCSF parameters without refractive correction. The SE was similar to the RS, both of which showed positive correlations with the qCSF parameters (Table 2). The CS at medium spatial frequencies was significantly different between the three RS and SE subgroups (Table 3). Previous studies have suggested that the ability of the human eye to distinguish objects is highest at medium spatial frequencies, and it is a reflection of daily visual ability.^33,34 The CS at medium spatial frequencies is critical for evaluating children's visual health. At low spatial frequencies, the CS of myopia children with RS within −1.00 D was not significantly different from that of nonmyopia children. At high spatial frequencies, CS significantly decreased with myopia (Table 3). Therefore, the current study findings indicate that the impact of refractive errors on CS at different spatial frequencies varies. The CS at medium spatial frequencies changes in gradient as refraction changes, providing an obvious distinction for assessing visual status using qCSF parameters in children.

As subjects were grouped by the refraction of −1.00 D (usually an indication for prescribing glasses in clinical practice), the current study findings suggest that a decrease in CS at high, medium, and low spatial frequencies indicates that there is myopia above −1.00 D, and refractive correction is needed. Although a decrease in CS only at high and medium spatial frequencies indicates that although myopia is still within −1.00 D, the visual function of the subjects is already altered, and the children should seek medical advice as necessary to enhance the visual function at medium and high spatial frequencies, especially in a finer learning environment. Meanwhile, although daily activity was probably not affected, there was a statistically significant decrease in CS values at medium and high spatial frequencies. This information will help promote the application of qCSF testing in pediatric eye health screening and visual function assessment. However, the relationship between contrast sensitivity and progression of myopia still requires further investigation.

The results of GLM analyses in this study suggested that RS was not significantly correlated with CS at high spatial frequencies (Fig. 3). A previous study found that CS at high spatial frequencies would show significant variability, even in subjects with normal visual acuity,²⁶ which indicates that it is susceptible to various factors. In this study, RC was positively correlated with CS only at 12.0 cpd, suggesting that astigmatism was more likely to affect CS at high spatial frequencies. Meanwhile, no significant correlation was found between RC and CS at 18.0 cpd, which was associated with the poor performance of the CSF at high spatial frequencies without refractive correction. Previous studies have revealed the negative effects of uncorrected astigmatism on contrast sensitivity and low-contrast visual acuity,^35,36 but the specific effects of CS at different spatial frequencies have not been evaluated. The results of this study suggest that astigmatism may have a more significant effect on CS at high spatial frequencies. Uncorrected astigmatism may affect neural processing and the development of the retina at different axes, and it may lead to decreased CS at high spatial frequencies. This indicates the necessity of correcting high astigmatism during visual development in children.

Our findings also demonstrate that age is negatively correlated with monocular CS at medium spatial frequencies (Table 2). Previous studies have shown that age is an important factor of CS in adults,²⁵ and elderly people have lower CS.^25,26 However, the effect of age on CS in children has not been confirmed. Studies have suggested that it is significantly lower³⁷ than or comparable³⁸ to that in adults. A recent study by Dekker et al.²⁷ stipulated that CS in children improved by approximately 0.3 log units from age 4 to 18 years. However, age explained only 16% of the variability of CS, and approximately 90% of children demonstrated a comparable CS with adults. In the current study, only CS at medium spatial frequencies, which were sensitive to uncorrected refractive errors, was significantly different between the two age groups, and the myopic shift with age may be the major reason for CS changes. In the binocular test, age was positively correlated with CS at low spatial frequencies, consistent with the findings by Dekker et al.²⁷ It was speculated that the correlation may be a joint effect of developmental factors (e.g., lateral inhibition or changes in the magnocellular pathway) and nonvisual factors (e.g., boredom or inattention).²⁷ Another reason may be that CS at low spatial frequencies is less affected by refraction, and therefore, correlations between CS and age can be manifested. However, linear regression analyses showed that age only explained approximately 2.3% of the variation in CS, and the correlation was weak (Fig. 5C). The finding that age was correlated with binocular rather than monocular CS at low spatial frequencies suggests that in the monocular test, CS is mainly affected by refractive errors. In addition, age-related changes (e.g., developmental changes in neuroprocessing) have less significant effects in the monocular test than in the binocular test. Therefore, this study shows that qCSF is less affected by age, and the qCSF test without refractive correction can be used for longitudinal follow-up in children. In addition, it provides a convenient method for continuous monitoring of visual health through a comparison of qCSF parameters in different age groups. Consequently, the qCSF test has more advantages for visual health screening in children compared with the visual acuity test.

In the present study, binocular CS was mainly affected by RS or SE of the better eye, especially at medium spatial frequencies, similar to the monocular test. Our findings showed that binocular qCSF parameters were significantly larger than were corresponding monocular parameters, and they were positively correlated except for CS at 18.0 c/d. This indicated the contrast summation of both eyes, that is, binocular CS is better than monocular CS. The binocular contrast summation ratio (binocular reading/that of the better eye) was higher than that in previous studies (1.34±0.47 vs. 1.15–1.19).^21,39 The subjects enrolled in this study were children rather than adults, and the contrast summation in children may differ from that in adults, which may explain the differences. A higher level of contrast summation may be related to the sensitive period of visual development, and this may be reflected by the finding that age was correlated with binocular CS at low spatial frequencies, but not with monocular CS (Table 5).

In addition, this study showed that binocular CS at 18.0 cpd was negatively correlated with that of eyes with higher RS (Table 5). This may be explained by the inhibition caused by the increased asymmetry of images received by both eyes, that is, the bad eye interferes with the good one.²¹ Although the CS of the bad eye would decrease more at high spatial frequencies, the increased interocular asymmetry of the CS may negatively affect the binocular CS.⁴⁰ Our findings revealed the necessity of performing monocular and binocular qCSF tests simultaneously for the following reasons. First, the evaluation of binocular contrast summation would help to demonstrate the function of binocular contrast summation in children. Second, it may impair the ability to distinguish fine objects when the CS of the two eyes is significantly different, indicating the need for immediate clinical intervention. Further studies are needed to evaluate the characteristics of qCSF under unaided acuity and binocular contrast summation ratios in strabismus and amblyopia. In addition, the CS at 18.0 cpd was mostly 0 in this study, and the effects of intraocular CS differences on binocular CS require further investigation.

This study describes the characteristics of qCSF parameters without refractive correction in children with ametropia. Previous studies tested CS in subjects with the best correction and found that qCSF parameters were mainly correlated with age.^18,19,25 The current study showed that the qCSF parameters significantly differed according to various degrees of refractive error in children, and the effect of age may be obscured. The qCSF test without refractive correction can be performed conveniently and rapidly, facilitating the implementation of large-scale screening. In addition, the qCSF test provides more information about visual health than the visual acuity test,^6,7 and its combination with other examinations would be helpful for discovering and diagnosing ophthalmic diseases. Therefore, the qCSF test is a practical method for screening visual health in children with moderate-to-low myopia and hyperopia. However, children with high myopia were not included in the present study; thus, further studies are needed to confirm the applicability of this method in high myopia children, because refractive errors would significantly affect qCSF parameters.

This study had some limitations. First, the sample size was relatively small. Instead of children of all ages, only children aged 6 to 14 years were included. Therefore, the general characteristics of the uncorrected CSF parameters in children may not be fully reflected. Further studies are warranted to include more subjects with different characteristics such as amblyopic patients whose visual acuity values are fine, but who complain of reduced visual function. These would deepen understanding regarding contrast sensitivity and visual diseases. Second, the range of refractive errors in the study population was relatively limited. Future studies are needed to assess the characteristics of the qCSF in high myopia children. Third, peripheral contrast sensitivity was not measured in this study. Evaluating the peripheral qCSF parameters may help assess the visual functions of the peripheral retina in myopic children and therefore provide a new direction for evaluating the clinical value of this method for myopia control and prevention. Fourth, the traditional contrast sensitivity test such as CSV-1000 was not conducted in the study, mainly because the traditional test method would take a long time and the measurement parameters are limited. Future studies concerning the comparison between the qCSF and traditional methods would be helpful for the clinical application of the qCSF test.

In conclusion, RS and SE are the major contributing factors of qCSF without refractive correction in Chinese ametropia children. CS values are decreased at medium and high spatial frequencies in children with low myopia. The combination of monocular and binocular qCSF tests is helpful in evaluating binocular visual health in children and has practical value for screening visual function.