Abstract
Purpose.:
Ethnic differences in childhood prevalence of myopia have not been well characterized in the United Kingdom. In this study, ethnic differences in refractive status and ocular biometry were examined in a multiethnic sample of British children.
Methods.:
This was a cross-sectional study of 10- and 11-year-old school children of South Asian, black African Caribbean, and white European ethnic origin. Vision, open-field autorefraction (without cycloplegia), and ocular biometry were measured in each eye. Myopia was defined as spherical equivalent refraction of −0.50 D with unaided vision of 20/30 or worse (in one or both eyes). Ethnic differences in the prevalence of myopia were examined by using logistic regression, and multiple linear regression was used for ethnic differences in ocular biometry. All models were adjusted for age, sex, and clustering within school.
Results.:
Data were available for 1179 children. The prevalence of myopia was 25.2%, 10.0%, and 3.4%, respectively, in the South Asian, black African Caribbean, and white European children. Adjusted odds ratios (ORs) of myopia compared with the white European children were 8.9 (95% confidence interval [CI] 4.0 to 19.4) in the South Asian and 3.2 (95% CI, 1.4 to 7.2) in black African Caribbean children. Ethnic differences in the prevalence of myopia were largely accounted for by ethnic differences in axial length. The South Asian and black African Caribbean children had longer axial lengths (0.44 mm; 95% CI, 0.30 to 0.57 mm and 0.30 mm; 95% CI, 0.16 to 0.44 mm, respectively).
Conclusions.:
Among British children exposed to the same schooling environment, the South Asians had the highest prevalence of myopia, followed by the black African Caribbeans compared with the white Europeans. A quarter of British South Asian children were myopic, which is strongly related to increased axial length.
Myopia is the most common cause of correctable visual impairment in the developed world and a leading cause of preventable blindness in developing countries.
1 Reduced vision in childhood is predominantly due to myopia,
2 –5 with a shift toward higher levels of myopia with increasing age.
6 Poor vision as a consequence of uncorrected refractive error has been identified as a priority area by the World Health Organization's global initiative to eliminate avoidable blindness by the year 2020.
7 Geographic variations in myopia prevalence are marked in both child and adult populations with the highest levels of myopia in East Asia, where approximately 80% of young adults are myopic.
8,9 Compared to East Asian children the prevalence of myopia is lower in children from South Asia.
10,11 The lowest prevalence appears to be in white children,
12 –14 with similarly low levels of myopia in children of African Caribbean origin.
3,15,16 These variations together with the recent rapid increases in the prevalence of myopia (especially among children in Asia
17 and in higher income countries,
18 ) suggest that environmental factors are important determinants of myopia
17 and hence of reduced unaided distance vision.
14 Axial length is a key determinant of the degree of myopia.
19,20 Thus, it is likely that ethnic differences in myopia prevalence would reflect ethnic differences in ocular biometry. However, few studies have examined these differences in children.
3,21,22
The population prevalence of childhood myopia is not well characterized in the United Kingdom, and little is known about levels of myopia in ethnic minority groups. We therefore examined ethnic differences in the prevalence of myopia and ocular biometry among children of white European, South Asian, and African Caribbean origin, who shared similar geography (resident in London or the Midlands) and schooling. Ocular and potential sociodemographic determinants of myopia in these children were explored.
Axial length (AL), anterior chamber depth (ACD), and front surface keratometry (FSK) were measured with noncontact optical coherence biometry (IOL Master; Carl Zeiss Meditec, Ltd. Welwyn Garden City, UK). Calibration was performed with the manufacturer's test eye before each session. A minimum of five measures of AL (differing by <0.2 mm, with a signal-to-noise ratio >2), three measures of keratometry (differing by <0.1 mm), and one measure of ACD (with depths less than <3 mm being repeated) were obtained per eye.
The prevalence of myopia and mean values for ocular biometric measures (SER, AL, ACD, FSK) are presented by sex and ethnicity, with 95% confidence intervals (CIs) adjusted for clustered sampling (Stata/SE software; Stata/SE 10.1 for Windows; StataCorp LP, College Station, TX). Odds ratios (ORs) of myopia (using logistic regression models) comparing the white Europeans with other ethnic groups and girls with boys were adjusted for age, sex (for ethnic comparisons), and ethnicity (for comparisons between the sexes), with random effect for school. For biometric comparisons, linear regression used data from both right and left eyes with additional adjustment for ocular asymmetry as a fixed effect and child and school as random effects; adjustment for height was explored. Residuals from regression models with ocular biometric measures as outcome variables appeared normally distributed; hence, ocular biometric measures were not transformed before the analyses were performed. To examine the contribution of ocular biometric measures explaining ethnic differences in the prevalence of myopia, we also adjusted the ORs for myopia for AL, FSK, and ACD.
Measures of ocular biometry, including AL, ACD, FSK, SER, and level of astigmatism (in negative form), by sex and ethnic group are summarized (means and 95% CI) in
Table 3. Data by age are not presented, because there was no evidence of any changes with age. Corresponding mean differences comparing boys with girls and white Europeans with other ethnic groups are also shown in
Table 3. Although there was no evidence of differences between the sexes in SER and level of astigmatism (in agreement with the observations of no difference in the prevalence of myopia), the girls had shorter ALs (by nearly 0.5 mm), shallower ACD (by 0.1 mm), and steeper FSK (by 0.1 mm).
Table 3. Ocular Biometric Measures by Sex and Ethnic Group, Together with Adjusted Differences by Sex and Ethnic Groups Compared with White Europeans
Table 3. Ocular Biometric Measures by Sex and Ethnic Group, Together with Adjusted Differences by Sex and Ethnic Groups Compared with White Europeans
| n | Axial Length (mm) | Front Surface Keratometry (mm) | Anterior Chamber Depth (mm) | Spherical Equivalent Refraction (D) | Astigmatism (D) |
Mean (95% CI)* | | | | | | |
Sex | | | | | | |
Males | 561 | 23.47 (23.40 to 23.55) | 7.87 (7.85 to 7.90) | 3.47 (3.44 to 3.49) | 0.31 (0.16 to 0.47) | −0.58 (−0.63 to −0.53) |
Females | 618 | 23.01 (22.93 to 23.08) | 7.77 (7.74 to 7.79) | 3.36 (3.34 to 3.38) | 0.33 (0.18 to 0.48) | −0.61 (−0.66 to −0.56) |
Ethnic group | | | | | | |
White Europeans | 268 | 23.01 (22.90 to 23.11) | 7.80 (7.77 to 7.84) | 3.42 (3.38 to 3.45) | 0.75 (0.56 to 0.93) | −0.52 (−0.59 to −0.46) |
Black African Caribbean (All) | 280 | 23.25 (23.15 to 23.35) | 7.85 (7.82 to 7.88) | 3.39 (3.36 to 3.42) | 0.31 (0.13 to 0.48) | −0.68 (−0.74 to −0.62) |
Black Caribbean | 103 | 23.27 (23.10 to 23.43) | 7.82 (7.77 to 7.88) | 3.49 (3.44 to 3.53) | 0.38 (0.11 to 0.64) | −0.62 (−0.71 to −0.53) |
Black African | 154 | 23.26 (23.13 to 23.40) | 7.84 (7.79 to 7.88) | 3.43 (3.39 to 3.47) | 0.25 (0.04 to 0.47) | −0.59 (−0.66 to −0.51) |
Black other | 23 | 23.11 (22.76 to 23.46) | 7.91 (7.79 to 8.02) | 3.42 (3.32 to 3.52) | 0.86 (0.32 to 1.40) | −0.73 (−0.91 to −0.54) |
South Asian (All) | 368 | 23.43 (23.34 to 23.53) | 7.81 (7.78 to 7.84) | 3.42 (3.38 to 3.45) | −0.26 (−0.45 to −0.07) | −0.61 (−0.67 to −0.54) |
Indian | 90 | 23.47 (23.29 to 23.65) | 7.77 (7.72 to 7.83) | 3.48 (3.43 to 3.54) | −0.35 (−0.65 to −0.05) | −0.65 (−0.75 to −0.55) |
Pakistani | 137 | 23.43 (23.29 to 23.58) | 7.82 (7.77 to 7.86) | 3.49 (3.44 to 3.53) | −0.06 (−0.32 to 0.20) | −0.59 (−0.67 to −0.50) |
Bangladeshi | 78 | 23.26 (23.07 to 23.45) | 7.80 (7.74 to 7.86) | 3.45 (3.39 to 3.50) | 0.22 (−0.16 to 0.61) | −0.43 (−0.56 to −0.30) |
South Asian other | 33 | 23.77 (23.48 to 24.06) | 7.77 (7.68 to 7.87) | 3.47 (3.38 to 3.56) | −0.93 (−1.43 to −0.44) | −0.59 (−0.74 to −0.43) |
Asian other | 81 | 23.29 (23.10 to 23.48) | 7.78 (7.72 to 7.84) | 3.44 (3.38 to 3.50) | 0.14 (−0.15 to 0.44) | −0.56 (−0.66 to −0.46) |
Other | 212 | 23.16 (23.05 to 23.28) | 7.82 (7.79 to 7.86) | 3.42 (3.38 to 3.45) | 0.48 (0.29 to 0.68) | −0.58 (−0.65 to −0.51) |
Adjusted differences (95% CI)† | | | | | | |
Sex | | | | | | |
Females vs males | | −0.48 (−0.58 to −0.39) | −0.11 (−0.14 to −0.08) | −0.11 (−0.13 to −0.08) | 0.02 (−0.14 to 0.17) | −0.02 (−0.07 to 0.03) |
Ethnic group | | | | | | |
Black African Caribbean (all) | | 0.30 (0.16 to 0.44) | 0.05 (0.01 to 0.10) | −0.01 (−0.05 to 0.03) | −0.45 (−0.68 to −0.22) | −0.15 (−0.22 to −0.08) |
Black Caribbean | | 0.29 (0.10 to 0.48) | 0.03 (−0.03 to 0.10) | 0.02 (−0.03 to 0.08) | −0.47 (−0.77 to −0.17) | −0.16 (−0.25 to −0.06) |
Black African | | 0.32 (0.15 to 0.48) | 0.06 (0.00 to 0.11) | −0.03 (−0.08 to 0.01) | −0.50 (−0.77 to −0.23) | −0.13 (−0.22 to −0.05) |
Black other | | 0.17 (−0.18 to 0.53) | 0.12 (0.01 to 0.24) | −0.03 (−0.13 to 0.07) | −0.04 (−0.59 to 0.52) | −0.25 (−0.43 to −0.07) |
South Asian (all) | | 0.44 (0.30 to 0.57) | 0.01 (−0.04 to 0.05) | 0.00 (−0.04 to 0.04) | −1.00 (−1.23 to −0.76) | −0.08 (−0.15 to 0.00) |
Indian | | 0.45 (0.25 to 0.64) | −0.03 (−0.09 to 0.03) | 0.01 (−0.05 to 0.06) | −1.15 (−1.49 to −0.82) | −0.11 (−0.22 to 0.00) |
Pakistani | | 0.43 (0.26 to 0.60) | 0.03 (−0.03 to 0.08) | 0.02 (−0.04 to 0.07) | −0.85 (−1.15 to −0.55) | −0.11 (−0.21 to −0.02) |
Bangladeshi | | 0.31 (0.10 to 0.52) | 0.02 (−0.04 to 0.09) | −0.03 (−0.09 to 0.03) | −0.63 (−1.03 to −0.22) | 0.05 (−0.08 to 0.19) |
South Asian other | | 0.76 (0.46 to 1.06) | −0.01 (−0.11 to 0.09) | −0.01 (−0.10 to 0.08) | −1.78 (−2.28 to −1.28) | −0.09 (−0.25 to 0.07) |
Asian other | | 0.30 (0.10 to 0.51) | −0.02 (−0.09 to 0.05) | 0.03 (−0.03 to 0.09) | −0.60 (−0.93 to −0.27) | −0.03 (−0.14 to 0.07) |
Other | | 0.17 (0.02 to 0.32) | 0.02 (−0.03 to 0.07) | 0.00 (−0.04 to 0.05) | −0.26 (−0.51 to −0.02) | −0.06 (−0.13 to 0.02) |
The black African Caribbean children were the tallest (mean height, 148.3 cm) followed by the white European (144.5 cm) and South Asian (144.2 cm) children. Adjustment for height made little difference in the results in
Table 2 (data not presented). Compared with the white European children, the South Asian children had 0.4 mm longer ALs, 1 D more myopia, and slightly higher levels of negative astigmatism; FSK and ACDs were similar. Differences between the white European, Asian Other, and black African Caribbean children followed a pattern similar to that in the South Asian children, but the differences were less marked (except for FSK, where the black African Caribbeans had flatter corneas). The Other ethnic group showed ocular biometry similar to that of white Europeans, except for marginally longer ALs and slightly more myopia. Patterns were very similar in the black African and black Caribbean ethnic subgroups. The South Asian Others had the longest AL and most myopic average SER (the same patterns as observed in the prevalence of myopia). The Indian and Pakistani subgroups had similar biometric characteristics, except the Indian children, who had a more myopic SER than did the Pakistani children; less extreme differences were observed in the Bangladeshis and Asian Others.
There was a suggestion (P = 0.02) that the ethnic differences in AL and SER were marginally larger in the South Asian girls than in the South Asian boys. Ethnic differences in AL (95% CI) and SER were 0.28 mm (0.09 to 0.47 mm) and −0.72 D (−1.02 to −0.40 D) in the South Asian boys and 0.59 mm (0.40 to 0.78 mm) and −1.27 D (−1.60 to −0.94 D) in the South Asian girls, compared with the measurements in the white European children. For the other main ethnic groups, there was no evidence of an interaction between ethnicity and sex.
There was some evidence that the South Asian children came from a lower socioeconomic level than did the white European children. However, additional adjustment for parental socioeconomic position had marginal impact on sex and ethnic differences, and hence these results are not presented.
Effect of Adjusting Ethnic Differences in Prevalence of Myopia for Ocular Biometry
Ethnic differences in myopia prevalence (adjusted for age and sex) were weakened after further adjustment for AL. The OR (95% CI) comparing the white European with the black African Caribbean children was reduced from 3.2 (1.4–7.2) to 1.7 (0.7–4.0) after adjustment for AL. Further adjustment for keratometry and ACD weakened the OR further to 1.4 (0.4–4.1). In the comparison of the white Europeans with the South Asians, the OR was reduced by a smaller amount, from 8.8 (4.0–19.4) to 5.4 (2.4–12.1) after adjustment for AL and to 2.1 (0.7–6.2) after additional adjustment for keratometry and ACD. Ethnic differences in mean SER were also weakened after adjustment for AL, FSK, and ACD but remained statistically significant. Compared with the white Europeans, the black African Caribbeans, and South Asians had a more myopic SER of −0.11 D (−0.23 to +0.02) and −0.20 D (−0.33 to −0.07 D), respectively, after full adjustment, with the largest contribution being ethnic differences in AL. Adjustment for AL reduced the total residual variance in SER by 35%. There was insufficient power to provide clear evidence of ethnic subgroup differences in myopia prevalence and SER with the additional adjustment for biometric measures; hence, data are not presented.
In this study, provides strong evidence that British South Asian children have a higher prevalence of myopia than do white European children. The black African Caribbean children also showed a higher prevalence of myopia compared with the white children, but not to the same degree as the South Asian children. Differences in myopia prevalence and SER corresponded with ethnic differences in ocular biometry, with black African Caribbean children and especially South Asians having longer ALs.
The prevalence of myopia observed in white Europeans (OR, 3.4%; 95% CI, 1.7%–6.6%) is similar to that reported in studies in which different methods were used to define myopia in white British children of a similar age, including definitions based on vision cutoffs in three large birth cohorts
14 and lower than in a recent U.K. study of older children,
26 in which Refractive Error Study in Children definitions were used.
24 The mean SER (+0.75 D) in the present study lay between levels reported in younger and older white U.K. children (Logan N, et al.
IOVS 2008;49:ARVO E-Abstract 2602). Collectively, these studies provide little evidence that there has been a sizeable change in the recent prevalence of myopia or degree of myopia among British white children. Similar estimates have been reported in white children in the United States
3 and Australia,
13 with the prevalence of myopia among white populations in Australia remaining stable over time.
27
Studies in nonwhite populations have suggested rapid increases in the prevalence of myopia, especially among children from Asia (particularly in children living in East Asia).
9,28 –30 Moreover, a high prevalence of myopia in childhood has also been observed among Asians resident outside their country of origin, both in the United States
3 and Australia.
13 Preliminary findings in other U.K.-based studies have suggested a higher prevalence of myopia among Asians compared with non-Asians at 10 years of age,
31 and the emergence of higher levels of myopia among South Asians in adolescence (Logan N, et al.
IOVS 2008;49:ARVO E-Abstract 2602). However, in most studies to date, the sampling frames used were not specifically designed to examine ethnic differences (mostly white populations studied).
3,13,31 The present study provides strong evidence that the burden of myopia is higher among British South Asians than among white Europeans (especially among Pakistani and Indian children). Our estimate of myopia's prevalence in British Indian children (27%) agrees with a study of migrant
13 and indigenous
32 Indian children. Others have reported much lower estimates of between 4% and 10% in similar aged children living in India.
10,33 –35 The prevalence of myopia was similar between children of black African and Carib bean origin, and our estimates agree with studies in black Americans,
3 but lower estimates have been reported in black Africans living in Africa.
15,16,36 Most of the black African Caribbean children in the present study were born in the United Kingdom. The frequency of spectacle use was similar across the ethnic groups and suggests that cultural/ethnic differences in health-seeking behavior are unlikely to explain these findings.
Differences between the sexes in the prevalence of myopia and SER were absent in the present study (except perhaps for South Asians), in agreement with findings in studies in similar aged children.
13,14,37 Sex-related differences have been reported later in childhood and adolescence, with higher levels of myopia in girls.
13,14,37,38 The reason for sex-related differences remains unclear but it may reflect differences between the sexes in response to environmentally determined factors, such as educational demand. The observation that girls have shorter ALs, steeper corneas, and shallower anterior chambers than boys at this age is supported by others
13,22,38,39 and was unaffected by adjustment for height.
An important feature of the present study design was to ensure similar numbers of children of white European, South Asian, and black African Caribbean origin. The cluster sampling design limits the influence of confounding factors, such as area of residence, degree of urbanization, educational status, and other sociodemographic factors. The high response rate (71%) and the similar response rates of the ethnic groups, limits the role of selection bias, wherein those with different visual characteristics do not participate.
One potential weakness of the present study is the use of habitual levels of ametropia as opposed to using levels ascertained after cycloplegia. There has been considerable debate about using cycloplegia in child surveys.
28,40 The current consensus recommends the use of cycloplegia to allow latent hyperopia and pseudoaccommodative myopia to be identified.
24 It was not feasible to include cycloplegia in the present study because of the possibility that it would influence the systemic vascular measures (carotid–femoral pulse wave velocity) that were performed at the same time. There were also concerns about short-term effects on the child's schooling and mobility, which could have led to lower participation rates. Hence, noncycloplegic levels of refraction were used to define myopia, which may overestimate the burden of myopia.
27 However, a recent study showed the open-field autorefraction without cycloplegia in 7- to 12-year-old children agreed well with cycloplegic refraction for the diagnosis of myopia of ≥0.5 D,
41 and this approach was validated in a similar age group showing a 90% detection rate for myopia (Logan N, personal communication, 2010; Aston Eye Study, Aston University, Birmingham, UK; information available at
http://www1.aston.ac.uk/lhs/research/health/org/eye-study/). The use of an open- rather than closed-field autorefractor with a nonaccommodative distant target (along with using the most positive of five consecutive readings with an appropriate visual cutoff) limits the extent of myopia misclassification. This approach is likely to limit difficulties encountered in earlier studies in which closed-field autorefraction without cycloplegia has been used to define myopia
40,42 –44 and is unlikely to invalidate the ethnic differences in myopia observed, as any potential measurement error would exert a similar effect across ethnic groups.
High levels of educational demand and urbanization, rather than genetic predisposition, have been proposed as factors that contribute to the increased prevalence of myopia in childhood and adolescence,
18 especially among children resident in East Asia.
45 –48 Our method of sampling resulted in such factors being balanced across the ethnic groups as the children in this study were of similar background with shared schooling; additional adjustment for parental socioeconomic status had little impact on the findings. Hence, the results of this study suggest that, as well as environmental factors and cultural patterning of visual activities,
13 certain ethnic groups may be more susceptible to myopia in response to their visual environment. This conclusion is contrary to the perceived view that populations are equally susceptible to environmental risk factors for myopia.
18 Variations in susceptibility, either as a result of genetic factors or early life exposure, such as birthweight,
14,49,50 may be responsible for the myopic shift in East Asians and potentially emerging myopic shift in South Asian and black African children, although further studies quantifying the global variation in prevalence of myopia are needed. Heterogeneity between South Asian boys and girls in comparison to white children for AL and SER may be indicative of the difference between the sexes in refractive error and myopia prevalence, emerging earlier within South Asians. Differences between the sexes have been observed in other studies.
13,14,37,38 It remains to be established whether these myopigenic environmental influences extenuate ethnic differences in later childhood; longitudinal studies will help in investigating this question.
Prevalence of myopia among British South Asian children is higher than among white European and black African Caribbean children attending the same schools. Understanding the reasons for these ethnic differences (the degree to which these are environmental and/or genetic) may help in formulating strategies to reduce the burden of myopia in later life, especially as myopia is a major cause of visual impairment in lower income countries with less scope for visual correction.
This study was performed during a cardiovascular survey funded by the British Heart Foundation project grant (PG/07/033/22770), and the study of myopia and ocular biometry was supported by the Bupa Foundation. The Child Heart and Health Study in England (CHASE) was supported by a grant from the Wellcome Trust (068362/Z/02/Z).
Disclosure:
A.R. Rudnicka, None;
C.G. Owen, None;
C.M. Nightingale, None;
D.G. Cook, None;
P.H. Whincup, None
The authors thank the members of the CHASE study team (in particular, Alicja Kiedzik, Devina Jo-Anne Joysurry, and Melanie Prescott, who made the ocular measures); the participating schools, pupils, and parents; the Birmingham Optical Group (Birmingham, UK) who provided a handheld fundus camera (NM-200D; Nidek, Gamagori, Japan); and Nicola Logan (Aston University, Birmingham, UK) who provided an autorefractor (Shin-Nippon Commerce Inc., Tokyo, Japan) for the duration of the study.