Corneal Thickness Determination and Correlates in Singaporean Schoolchildren

Abstract

purpose. To determine the central cornea thickness (CCT) in Singaporean children and to examine the possible relationship between intraocular pressure (IOP) and other biometric factors and CCT.

methods. This was a cross-sectional study. The subjects (N = 652) were obtained from the Singapore Cohort Study of the Risk Factors for Myopia (SCORM). The subjects’ ages ranged from 9 to 11 years. There were 485 Chinese, 92 Malayan, and 75 Asian Indian children. Measurement procedures included air-puff tonometry, noncontact slit lamp optical pachymetry, cycloplegic autorefraction, and autokeratometry.

results. The mean CCT was 543.6 ± 32.0 μm. Chinese children had thicker corneas than Malayan or Indian children (P = 0.002). The boys had thicker corneas than girls (P = 0.011), but the mean difference was only 6.4 μm. There was high correlation of CCT (r = 0.98) and IOP (r = 0.88) between right and left eyes. IOP was correlated with CCT (r = 0.45, P < 0.001). In a multiple linear regression model, each millimeter of mercury of IOP was associated with a CCT difference of 5.90 μm (95% confidence interval [CI], 4.98–6.82). The radius of corneal curvature correlated with CCT (r = 0.19, P < 0.001). The following parameters were not significantly (P > 0.05) associated with CCT: age, family income, father’s education, axial length, and spherical equivalent.

conclusions. The mean CCT in Singaporean children aged 9 to 11 years was 543.6 μm and showed ethnic and gender variation. CCT affected measured IOP and correlated weakly with corneal curvature. Compared with data in adults, a change in CCT was associated with a greater difference in measured IOP.

Measurement of normal central corneal thickness (CCT) in children is important because it affects interpretation of measured intraocular pressure (IOP) ¹ and allows for modeling of one aspect of human eye growth that cannot be derived from other biometric components. ^{2 3}  

In Western subjects, CCTs in newborns were above 580 μm, ⁴ which decreased to adult values (approximately 520 μm) by 3 years of age. ⁵ In another study, ⁶ white children 5 to 15 years of age were found to have similar mean CCT compared with adults >32 years of age. In contrast, a study in Chinese Hong Kong children ⁶ showed a marked reduction of CCT between 10 and 25 years of age. The normal CCT in Chinese children before 10 years of age would have to be determined to investigate factors responsible for the decrease of CCT after the age of 10 years. In fact, modeling of the pressure volume and ocular rigidity of the corneoscleral shell performed previously ⁷ with various assumptions may benefit from the incorporation of actual measured CCTs in the growing eye. 

In adults, a thicker cornea had been shown to be associated with a higher measured IOP, in some ^{8 9 10 11} but not other ^{12 13} studies. However, the statistical strength and magnitude of this association have not been studied in children. As far as we are aware, CCT and its relationship with demographic and socioeconomic factors also have not been reported in children. 

Given the importance of CCT and the unresolved questions just mentioned, a study was designed to evaluate the normal CCT and its possible correlations in schoolchildren, using a highly reliable, noninvasive, objective pachymeter. 

Methods

Study Subjects

The cross-sectional results of an ongoing longitudinal observational study, ¹⁴ the Singapore Cohort Study of the Risk Factors for Myopia (SCORM) in Singapore have been reported previously. ¹⁴ The study commenced in 1999, and schoolchildren in grades 1 to 3, attending three Singaporean schools (located in the Eastern, Northern, and Western part of Singapore) were recruited. Exclusion criteria included serious medical conditions such as leukemia or heart disorders and allergies to eye drops. For the present study, the third-year data from one (Western) school were used. The subjects (N = 652) were systematically sampled 1 in 2. Every alternate schoolchild examined in the study center had CCT measurement. The recruited subjects’ ages ranged from 9 to 11 years. The group comprised 333 boys and 319 girls. There were 485 Chinese, 92 Malayan, and 75 Asian Indian children. 

This study was approved by the ethics committee of the Singapore Eye Research Institute, and all procedures adhered to the Declaration of Helsinki. Informed written consent was obtained from parents of the children. 

Measurements

CCT measurements were obtained with an automated, noncontact, optical low-coherence reflectometry (OLCR) pachymeter (Haag-Streit, Bern, Switzerland) that was mounted on a slit lamp microscope for support. The details of the design of an optical pachymeter based on OLCR have been described previously. ¹⁵ Briefly, the design is based on the Michelson interferometer. The physical thickness of the cornea can be calculated from its refractive index and the optical distance between interference signals corresponding to the tear–cornea and cornea–aqueous humor interfaces, as analyzed by computer. This commercially available instrument is able to determine the CCT in children within a short time and produces a printout of the average and SD of five readings. The pachymeter automatically rejects readings that are not consistent, ensuring high-quality data. The refractive index of the cornea used was 1.376. ¹⁶ Studies have reported the precision of OLCR pachymetry to be 0.3 to 1.6 μm. ^{17 18 19} A reproducibility study was undertaken in 20 normal children, who were sampled from the main cohort study. The test–retest repeatability of this pachymeter (in the right eyes for this study) was −0.04 μm (95% confidence interval [CI], −5.33 to +5.25). Children whose CCT exceeded 600 μm were referred for slit lamp biomicroscopy examination by an experienced ophthalmologist. None of these eyes showed any clinically evident corneal disease. 

IOP measurements were obtained using an autotonometer (NT-3000; Nidek Co. Ltd., Tokyo, Japan), a noncontact, air-puff tonometer. Care was taken to avoid exerting unnecessary force on the upper lid during tonometry. The mean of three readings was obtained for further analysis. All children with IOP above 21 mm Hg were referred for slit lamp anterior segment examination and biomicroscopic optic disc examination to exclude glaucomatous optic nerve head changes. 

The cycloplegic regimen consists of a drop of 1% cyclopentolate repeated twice at 5-minute intervals. All measurements were performed at least 30 minutes after instillation of the last eye drop. Cycloplegic autorefraction and autokeratometry in two perpendicular meridians were performed (Canon RK-5; Canon Ltd., Tochigiken, Japan). The two corneal radii were measured five times, and the mean was used for further analysis. 

A-mode ultrasonography was performed (US-800 EchoScan; Nidek Co. Ltd.) to ascertain the axial length of the eyes. A mean of six readings was used for further calculations. 

Questionnaire

A detailed parent-administered questionnaire was used to evaluate various sociodemographic factors. The father’s educational level and total family income were used as socioeconomic indicators. Educational level was divided into five categories: no formal education, primary or elementary school education, secondary school education, diploma education, and university or postgraduate education. Total family income per month was categorized into three levels: <S$2000, S$2001 to S$5000, and >S$5000. 

Data Analysis

The refractive error, in diopters (D), was calculated as the spherical equivalent (SE). The mean of five SE measurements in each eye was used for further analysis. Because there was high correlation of CCT (intraclass correlation coefficient [ICC] = 0.98; 95% CI, 0.97–0.99) and IOP (ICC = 0.88; 95% CI, 0.79–0.93) between right and left eyes, only the right eye data were used for further analyses. 

The Student’s t-test was used to evaluate the differences in CCT between groups such as boys and girls or Chinese and non-Chinese girls. The ANOVA procedure was used to compare CCT between different ages. Pearson correlation analysis was used to determine the relationship of two continuous variables (e.g., SE and CCT). Throughout this article, the correlation coefficient (r) refers to the Pearson correlation coefficient. To evaluate the possible contribution of various factors to the variance of CCT, we constructed a generalized linear regression model (analysis of covariance; [ANCOVA]), with CCT as the dependent variable. Each model was constructed manually in a backward stepwise fashion, comparing with competing models so that the most parsimonious and scientifically sound model was selected. Tolerances for the proposed model were examined for possible collinearity between variables. For all analyses, the level of statistical significance was set at α = 0.05. Analysis was conducted with commercially available software (SPSS 12.0 for Windows; SPSS Sciences, Chicago, IL). 

Results

CCT was normally distributed (Fig. 1) . Table 1 shows the mean CCT in children of the different ages and genders. 

The boys had significantly thicker corneas than did the girls (P = 0.011), but the mean difference was only 6.4 μm. The CCT was not significantly associated with the age of the children (P = 0.82). The mean CCT in 9-, 10- and 11-year-old children were 544.6 ± 32.8, 542.7 ± 29.1, and 543.6 ± 34.1 μm, respectively. The Chinese children had thicker corneas than did the non-Chinese children (P = 0.002). 

Table 2 shows the relationship between refractive error, IOP, and biometric components and CCT. CCT increased from a mean of 527.2 μm in the first IOP quartile (10.7–15.3 mm Hg) to 563.7 μm in the highest IOP quartile (19.1–24.7 mm Hg). IOP correlated significantly with CCT (r = 0.45, P < 0.001; Fig. 2 ) The measured IOP in these children ranged from 10.7 to 24.7 mm Hg. The scatterplot shows that there were no segments of IOP that exhibited a linear trend with a different gradient from the entire group of children. After adjustment for gender, ethnicity, and age, the relationship between increasing IOP and CCT remained statistically significant. 

The radius of corneal curvature (Fig. 3) correlated with CCT (r = 0.194, P < 0.001). The CCT increased from the steepest corneas (lowest quartile for corneal radius of curvature) of 535.5 to 552.2 μm in the flattest corneas (highest quartile). This statistical significance remained after adjustment for gender, ethnic group, and age (see rightmost column in Table 2 ). 

The following parameters were not significantly (P > 0.05) associated with CCT: age (P = 0.09), total family income (P = 0.91), father’s education (P = 0.51), axial length (P = 0.26), and SE (P = 0.58). Thinner corneas were not significantly associated with more severe myopia (Table 2 ; P = 0.100). 

The results of the final ANCOVA model with CCT as the dependent variable and age, gender, ethnic group, IOP, and corneal curvature radius as the covariates are shown in Table 3 . Each millimeter of mercury increase in IOP was associated with a CCT difference of 5.90 μm (95% CI, 4.98–6.82), controlling for age, gender, ethnic group, and corneal curvature radius. For every 1-mm increase in corneal curvature radius, the CCT increased by 18.05 μm (95% CI, 9.19–26.91), after controlling for age, gender, ethnicity, and IOP. 

Discussion

The mean CCT in Singaporean schoolchildren aged 9 to 11 years was 543.6 ± 32 μm. This was highly symmetrical between right and left eyes, with Chinese children having greater CCT than non-Chinese children, and boys having greater CCT than girls. The CCT correlated with increased IOP and flatter corneas. 

The children’s CCTs was only marginally (3.5 μm) smaller than those in Singaporean adults (the Tanjong Pagar Survey), who had a mean CCT of 547.1 μm. ¹⁰ The presently reported CCT was greater than those in Mongolians ⁸ aged 10 to 19 years (513 μm) and white ⁶ children aged 5 to 15 years (529 μm). Hong Kong Chinese (10–15 years of age) ²⁰ were found to have a mean CCT of 598 ± 27 μm (SD). ²⁰  

In the present study, each millimeter of mercury change in IOP was associated with a CCT difference of approximately 6 μm. The direction of the association was the same as in an adult study, ¹⁰ also conducted in Singapore, but the latter reported a difference of 0.12 mm Hg with each 10-μm difference in CCT, with a rather wide 95% CI of 0.05 to 0.189 mm Hg. ¹⁰ This implies that as much as a 250-μm difference in CCT in adults but only a 18-μm difference in children would be needed to account for a change of 3 mm Hg of measured IOP. In a separate study in an elderly population, ¹¹ IOP was found to increase by 0.19 mm Hg (95% CI, 0.09–0.28) with each 10-μm increase in central corneal thickness. 

Radius of corneal curvature was not found to be associated with CCT in Hong Kong Chinese individuals, ²⁰ whereas this association has been found in adult studies. ^{21 22}  

In the present study, CCT was not associated with refractive error, axial length, or socioeconomic status. A study in Eskimos found a higher CCT in town dwellers than in villagers, and a higher CCT in families with an indoor occupation. ²³  

Most of the discrepancies in all these findings could be due to methodological differences. The Tanjong Pagar study ¹⁰ involved an ethnic profile similar to that in the present study but involved adults and used an earlier, nonautomated optical pachymeter. In the Mongolian study ⁸ and the study in Hong Kong children ²⁰ mentioned earlier, CCTs were assessed by ultrasonic pachymetry. 

The discrepancy between true manometric IOP and measured IOP could not be entirely attributed to CCT differences. ²⁴ One should not make any presumptions concerning the relationship of CCT and actual IOP. 

The strengths of this study include a reasonable sample size and homogeneity of the method of refraction and biometry. The study was performed in an age group with ethnic backgrounds that had not been adequately studied in the past. In particular, the method of pachymetry used is highly repeatable and is likely to be an improvement over previously published methods. This pachymetric method’s advantage is that subjective location of corneal layers is not required. Subjective evaluation of the position of the anterior surface of the Bowman’s layer (“overlap” method) or the anterior epithelial surface (“touch” method) may be erroneous. ⁸ The current OLCR is ideal for epidemiologic studies in its lack of contact with the cornea. By using a highly repeatable, comfortable, and quick method of pachymetry, the participation rate would not be compromised by the introduction of this additional procedure. 

In the present study, the schools were not randomly selected, and this may limit the ability to generalize the findings to all schoolchildren. This study did not allow direct comparison with adult CCTs and was cross-sectional in nature. The slightly higher CCTs in Singaporean adults ¹⁰ should be confirmed using the same pachymeter. The narrow age range of the children studied presently would not enable detection of age-specific differences. A longitudinal study of CCT is needed to discover relationships with longitudinal refractive change. Such a study would also refine our understanding of ocular growth in terms of biometric changes of the cornea, especially if the same eyes were to be followed up to adulthood. The relationship between corneal curvature and CCT could also be better evaluated in a longitudinal study. Further studies can also be conducted specifically to examine ethnic variations of Malayans and Asian Indians by increasing the number of Malayans and Indians investigated. 

In conclusion, mean CCT in Singaporean children was 543.6 ± 32 μm. This study also showed an ethnic variation of CCT in a multiethnic population, in which Chinese male children had a thicker cornea than did non-Chinese female children. The study showed a weak correlation between radius of corneal curvature and CCT. This study showed a definite relationship between CCT and measured IOP. In children, an increase in CCT was associated with a greater increase in IOP than in adults, and the relationship between the two parameters was linear across the normal IOP range. 

 Supported by Grant SERI/MG/97-04/005 from the Singapore Eye Research Institute and Grant NMRC/0695/2003 from the National Medical Research Council.

 Submitted for publication February 6, 2004; revised April 9, May 18, and July 20, 2004; accepted July 29, 2004.

 Disclosure: L. Tong, None; S.-M. Saw, None; J.-K. Siak, None; G. Gazzard, None; D. Tan, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Normal distribution of CCT in the study population (N = 652).

Table 1.

 

CCT in the Study Population

	All			Chinese			Non-Chinese
		n	Mean CCT (SD)	n	Mean CCT (SD)	n	Mean CCT (SD)
Total all ages	652	543.6 (32.0)	485	546.0 (31.8)	167	536.6 (31.5)
9 years	209	544.6 (32.8)	155	546.7 (34.3)	54	538.7 (27.5)
10 years	225	542.7 (29.1)	165	546.3 (26.9)	60	532.6 (32.4)
11 years	218	543.6 (34.1)	165	545.0 (33.9)	53	539.0 (34.4)
Males	333	546.7 (33.0)	257	550.3 (32.9)	76	534.6 (30.7)
9 years	103	547.2 (35.1)	82	550.7 (36.8)	21	533.4 (23.4)
10 years	116	545.4 (31.4)	82	550.0 (29.6)	34	534.3 (33.2)
11 years	114	547.6 (33.0)	93	550.1 (32.4)	21	536.3 (34.0)
Females	319	540.3 (30.5)	228	541.2 (29.8)	91	538.3 (32.3)
9 years	106	542.1 (30.4)	73	542.1 (30.9)	33	542.1 (29.6)
10 years	109	539.8 (26.2)	83	542.7 (23.7)	26	530.3 (31.9)
11 years	104	539.2 (34.8)	72	538.5 (34.9)	32	540.8 (35.1)

 Data are expressed as mean micrometers (SD).

Table 2.

 

The Factors that Affect CCT

		Central Corneal Thickness (μm)
				Mean (SD)	Age-Gender-Race Adjusted Mean (SD)*
Intraocular pressure (mm Hg)
First quartile (10.7–15.3)	171	527.2 (27.4)	526.2 (47.7)
Second quartile (15.4–17.0)	167	537.1 (31.0)	538.4 (30.9)
Third quartile (17.1–19.0)	177	550.0 (28.1)	550.5 (32.5)
Fourth quartile (19.1–24.7)	137	563.7 (30.0)	562.7 (50.7)
P		<0.001	<0.001
Corneal curvature radius (mm)
First quartile (7.10–7.61)	169	535.5 (30.8)	536.8 (52.9)
Second quartile (7.62–7.75)	158	542.8 (29.5)	541.4 (34.2)
Third quartile (7.76–7.92)	162	544.2 (33.4)	545.9 (34.5)
Fourth quartile (7.93–8.68)	163	552.2 (32.0)	550.5 (53.4)
P		<0.001	<0.001
Axial length (mm)
First quartile (20.59–23.19)	163	540.4 (28.7)	542.8 (56.3)
Second quartile (23.20–23.93)	163	542.1 (29.4)	543.3 (35.2)
Third quartile (23.94–24.62)	164	547.0 (33.5)	543.8 (35.3)
Fourth quartile (24.63–27.46)	161	544.8 (35.7)	544.3 (56.4)
P		0.26	0.69
Refractive error status
Hypermetropia (SE > 1.0 D)	30	540.9 (31.0)	547.9 (74.6)
Emmetropia (−0.5 D < SE ≤ 1.0 D)	238	545.4 (30.3)	545.5 (43.3)
Low myopia (−3.0 D < SE ≤−0.5 D)	230	543.8 (32.6)	543.1 (32.6)
High myopia (SE ≤−3.0 D)	154	541.0 (33.7)	540.6 (56.1)
P		0.580	0.100

 Data show CCT (in micrometers) in the presence of each factor.

*  Adjusted for age, gender, and ethnic group.

Figure 2.

 

Scatterplot of IOP against CCT. Solid line: regression line; dotted lines: 95% CI of the slopes of the regression line.

Figure 3.

 

Scatterplot of corneal radius of curvature against CCT. Solid line: regression line; dotted lines: 95% CI of the slopes of the regression line.

Table 3.

 

Final Generalized Linear Model of CCT and Associated Factors

	Regression Coefficient* (95% CI)	P
Age (y)	−2.32 (−4.99–0.35)	0.090
Gender (females vs. males)	−4.38 (−8.83–0.08)	0.050
Ethnic group (non-Chinese vs. Chinese)	−3.66 (−6.86–−0.47)	0.030
Intraocular pressure (mm Hg)	5.90 (4.98–6.82)	<0.001
Corneal curvature radius (mm)	18.05 (9.19–26.91)	<0.001

*  All explanatory variables are adjusted for each other.

The authors thank research assistant Angela Cheng, Mandarin Optomedics Singapore, and Stephen Gee, our IOVS volunteer editor.