Article Text

Download PDFPDF

Racial difference in the prevalence of epiretinal membrane between Caucasians and Asians
  1. R Kawasaki1,
  2. J J Wang1,2,
  3. P Mitchell2,
  4. T Aung3,4,5,
  5. S-M Saw3,6,
  6. T Y Wong1,3,4,5,6,
  7. for the Singapore Malay Eye Study Group
  1. 1
    Centre for Eye Research Australia, University of Melbourne, Australia
  2. 2
    Centre for Vision Research, University of Sydney, Australia
  3. 3
    Singapore Eye Research Institute, Singapore
  4. 4
    Singapore National Eye Centre, Singapore
  5. 5
    Department of Ophthalmology, Yong Loo Lin School of Medicine, National University of Singapore, Singapore
  6. 6
    Department of Community, Occupational and Family Medicine, Yong Loo Lin School of Medicine National University of Singapore, Singapore
  1. Professor T Y Wong, Centre for Eye Research Australia, The Royal Victorian Eye and Ear Hospital, 32 Gisborne Street, Melbourne 3002, Australia; twong{at}


Background/aims: To describe the prevalence and risk factors of epiretinal membranes (ERM) in Asian Malays and to compare this with a white population.

Methods: The Singapore Malay Eye Study examined 3280 persons (78.7% response rate) aged 40–80 years in Singapore during 2004–6. ERM were graded from retinal photographs using standardised procedures at the University of Sydney, and rates were compared with those from the Blue Mountains Eye Study (BMES).

Results: Of the 3,280 participants, 3,265 had sufficient quality photographs for grading. The age-standardised prevalence of ERM was 7.9% (95% CI 7.1 to 8.7%) in the Singapore Census population. The prevalence of ERM was higher in Malays than in Caucasians from the BMES (age-standardised prevalence: 15.8% (CI 14.2 to 17.2%) in Malays vs 6.8% (CI 5.9 to 7.6%) in Caucasians). Of the 384 persons with ERM, 124 (32.3%) had secondary ERM. Age, female gender, hyperopia and narrower retinal arteriolar diameter were associated with higher prevalence of ERM, after adjusting for age and/or gender.

Conclusions: The prevalence of ERM in Asian Malays was higher than that in the Caucasians. Risk factors for ERM were older age, female gender, hyperopic refraction and narrower retinal arteriolar diameter.

Statistics from

Epiretinal membranes (ERM) are frequent retinal conditions affecting older people. Few studies have described the prevalence of ERM in Asian ethnic groups.1 2 It has been suggested that Asian persons have a lower prevalence of ERM compared with that in Caucasians.3 The age-standardised ERM prevalence was lower in a study from Japan (2.8%) than in studies from Western countries (5.1 to 9.1%).3 The Beijing Eye study also reported a low prevalence of ERM in ethnic Chinese (2.2%).2 However, comparisons of these studies are limited by differences in the assessment of ERM. In the Funagata Study, ERM was defined from photographs graded at the same centre as the Blue Mountains Eye Study (BMES),4 and the prevalence of ERM among Japanese was similar to Caucasians in the BMES5 and Beaver Dam Eye Study (BDES)6 population.

Eyes with ocular diseases such as proliferative diabetic retinopathy or retinal detachment develop ERM.7 Most cases are “idiopathic” or primary ERM without any apparent ocular pathology than posterior vitreous detachment. Risk factors for primary ERM are not clearly understood; age has been the only consistent risk factor for ERM in different studies13 5 6 8 (table 1).

Table 1 Summary of previously reported risk factors associated with primary epiretinal membranes in population-based epidemiological studies

In this report, we aim to describe the age- and gender-specific prevalence of primary and secondary ERM in a Malay population in Singapore, and compare this with the prevalence reported from the BMES, after age-standardisation to the BMES population. We also examined risk factors associated with primary ERM, including ocular, systemic and socio-demographic characteristics.


Study populations

The Singapore Malay Eye Study (SiMES) is a population-based cross-sectional study of 3280 Malay adults aged 40–80 years living in Singapore.9 10 The study design and details of population sampling of the SiMES have been described elsewhere.9 In brief, based on the National Registration identity card number, 16 069 potential participants were selected using residential telephone numbers. From this list, we derived a final sampling frame of 5600 names using an age-stratified random sampling strategy, selecting 1400 from each decade (40–49, 50–59, 60–69 and 70–79 years) across the 15 residential districts.9 The residents were a fair representatives of Singapore population in regard to age, housing and socio-economic status based on the 2000 Singapore Census.9 As a result, 4168 persons were determined as eligible, and 3280 Malays (overall response rate of 78.7%) were examined in the clinic between August 2004 and June 2006. Written consent was obtained from all participants. Ethics approval was obtained from the Institutional Review Board of the Singapore Eye Research Institute, Singapore, and the study was conducted in accordance with the World Medical Association’s Declaration of Helsinki.

A digital retinal camera (Canon CR-DGi, EOS10D digital camera, Canon, Tokyo) was used to obtain colour photographs of Early Treatment for Diabetic Retinopathy Study (ETDRS) standard field 1 (centred on the optic disc) and standard field 2 (centred on the fovea) of each eye after pupil dilation. The fundus photographs were assessed in the grading centre in the Centre for Vision Research, University of Sydney, Australia, for the presence of ERM and other retinal diseases, following the definitions used in the BMES,5 which was originally adopted from the BDES.6 Fundus photographs were assessed initially by trained graders, and all pathologies were adjudicated by either or both a senior researcher (JJW) and a retinal specialist (PM).

Of the 3280 participants, 3265 (99.5%) had fundus photographs with sufficient quality for grading of ERM. Two stages of ERM were identified: a less severe form termed “cellophane macular reflex” (CMR) was defined as having glinting, water-silk, and shifting light reflex without visible retinal folds; and a more severe form termed “preretinal macular fibrosis” (PMF) was defined as having retinal folds with more opaque, greyish appearance on the inner retinal surface.5 6 ERM outside the 3000 μm radius grid on macula were not assessed. When both CMR and PMF are found within the same eye, it was categorised as having PMF. The term “any ERM” was defined to include subjects with either CMR or PMF.

The graders also assessed retinal diseases other than ERM using similar standardised grading protocols. Diabetic retinopathy were graded with the modified Airlie House Classification, as used in the Early Treatment Diabetic Retinopathy study.11 12 Age-related macular degeneration were defined following the modification of the Wisconsin Age-Related Maculopathy Grading System13 as used in the BMES.14 Branch or central retinal vein occlusion were assessed as present or absent. If ERM and those retinal diseases coexisted within the same eye, they were classified as secondary ERM.

The best-corrected visual acuity was measured using a logarithm of the minimum angle of resolution (logMAR) number chart. “Low vision” was defined as having logMAR >0.30 to <1.00 and “blindness” as having logMAR ⩾1.00.10 Lens status was assessed for nuclear, cortical, and posterior subcapsular cataract following the Lens Opacities Classification System (LOCS III).15 Central retinal arteriolar equivalent (CRAE) and central retinal venular equivalent (CRVE) were calculated using standardised method described elsewhere.16

A comprehensive physical examination, laboratory tests, and interview were performed following standardised protocols as described elsewhere.9 Blood pressure was measured in standard manner after 5 minutes’ rest with a digital automatic blood pressure monitor. Hypertension was defined as systolic blood pressure ⩾40 mm Hg, diastolic blood pressure ⩾90 mm Hg or physician’s diagnosis. Diabetes mellitus was defined as having a random glucose of ⩾11.1 mmol/l, or self-reported history of diabetes and use of diabetic medication.12 A detailed interviewer-administrated questionnaire was used to collect information about medical history, cigarette smoking (defined as current, past and never), alcohol consumption, current medication use and socio-economic status.17

Statistical analysis

All analyses were carried out using Stata (Version 10.0, StataCorp, College Station, TX) and Statistical Package for Social Science (SPSS 13.0, SPSS, Chicago). Age- and gender-specific prevalence rates were estimated, and then age-standardised to the Malay population from the 2000 Singapore Census. Logistic regression analyses were used to estimate odds ratios and 95% CI, adjusting for age and gender. When eye-specific characteristics and systemic characteristics were analysed together, a marginal model with generalised estimating equation (GEE) was used to account for correlation between right and left eyes. Age-standardised prevalence of ERM was compared between this current study and BMES, using the direct standardisation method with BMES population as a standard population.


In our study subjects, CMR and PMF were found in 5.8% (190/3265) and 5.9% (194/3265), respectively. The age-standardised prevalence of CMR, PMF and any ERM in the 2000 Singapore Census population aged 40–80 years was 4.1% (CI 3.4 to 4.7%), 3.8% (CI 3.3 to 4.4%) and 7.9% (CI 7.1 to 8.7%), respectively (table 2).

Table 2 Prevalence of primary and secondary epiretinal membranes by age and gender: the Singapore Malay Eye Study, 2004–6

The age-specific prevalence of primary ERM in this Malay population was higher than that in the BMES in all age groups (fig 1). After age standardisation to the BMES population aged 49 to 80 years, the prevalence of CMR, PMF and ERM in this study was 8.3% (CI 7.2 to 9.5%), 7.4% (CI 6.4 to 8.5%) and 15.7% (CI 14.3 to 17.2%), respectively. These prevalences were higher than rates reported for Caucasians aged 49 to 80 years in the BMES (CMR: 4.6% (CI 3.9 to 5.4%); PMF: 2.1% (CI 1.6 to 2.6%) and ERM: 6.8% (CI 5.9 to 7.6%)).

Figure 1 Comparison of age-specific prevalence of primary epiretinal membranes between the Singapore Malay Eye Study and the Blue Mountains Eye Study.

Of the 384 persons with epiretinal membrane, 124 (32.3%) were considered to be secondary epiretinal membrane. The most frequent cause of secondary epiretinal membrane was a history of cataract surgery (77% (95/124)). The prevalence of epiretinal membrane in subjects with diabetic retinopathy, retinal vein occlusion and history of cataract surgery was 11.3% (36/320), 13.6% (3/22) and 28.0% (130/465), respectively (fig 2).

Figure 2 Comparison of prevalence of secondary epiretinal membranes by cause between the Singapore Malay Eye Study and the Blue Mountains Eye Study.

Increasing age and female gender were associated with any epiretinal membrane (table 3). After adjustment with age and gender, hyperopia (OR 1.68, CI 1.24 to 2.27) and narrower retinal arteriolar diameter (OR per 10 μm decrease in CRAE 1.22, CI 1.04 to 1.42) were associated with ERM (table 3). The retinal arteriolar diameter-ERM prevalence association remained significant after further adjusting for hypertension (OR 1.20, CI 1.02 to 1.40) or systolic blood pressure (OR 1.21, CI 1.03 to 1.42). On the other hand, current smoking was inversely associated with any ERM (age–gender-adjusted OR 0.60, CI 0.38 to 0.94). Myopia and presence of posterior subcapsular cataract were also inversely associated with ERM (table 3).

Table 3 Associations of epiretinal membranes adjusted for age and gender: the Singapore Malay Eye Study, 2004–6

There were no significant associations of ERM with history of cardiovascular diseases, serum lipids, haemoglobin A1c, BMI, plasma glucose, alcohol intake, final education level, income level, outdoor job, intraocular pressure and axial length (data not shown).

Compared with eyes without any ERM, the mean best-corrected visual acuity in eyes with primary PMF was significantly lower by a logMAR score of 0.07 (CI 0.03 to 0.10). After adjustment for age and lens status, this difference was not significant (logMAR score difference −0.02, CI −0.05 to 0.01). Although eyes with PMF had a higher prevalence of visual impairment (low vision or blindness) before adjustment (OR 1.64, CI 1.01 to 2.67), this was not statistically significant after adjusting for age and gender (OR 0.72, CI 0.42 to 1.22).


In this study, we described an overall ERM prevalence of 7.9% in the Singapore Malay population aged 40–80 years. This includes 3.8% for CMR and 4.1% for PMF. There were significant age trends in the prevalence of any ERM (p<0.001). The age-specific prevalence of primary ERM in this Malay population was higher than that reported from the BMES in all age groups (fig 1); age-standardised prevalence rates of primary ERM were higher in the Malay population than in the BMES for persons aged 49–80 years (15.7% and 6.8%, respectively). Of all ERM, 32.3% were secondary ERM; the most frequent cause of secondary ERM was a history of cataract surgery (77.0%). In addition to age, we found that female gender, hyperopia and narrower retinal arteriolar diameter were positively associated with ERM. In contrast, current smoking, myopia and posterior subcapsular cataract were negatively associated with ERM.

In previous population-based studies in Asia, the ERM prevalence was lower than that in mainly white populations.1 2 However, when we compared our age-specific prevalence to that in the BMES, the Malay population seems to have a comparable or even higher ERM prevalence than in the white BMES population5 (fig 1), and a substantially higher prevalence compared with that reported by the Hisayama Study1 and the Beijing Eye Study.2 Also, the prevalence of secondary ERM in participants with diabetic retinopathy, retinal vein occlusion and history of cataract surgery in our study sample were comparable with those in the BMES (fig 2).5 A strength of our study is its high comparability to the BMES, because retinal photographs of both studies were evaluated in the same grading centre and by the same grader, using identical protocols. Interestingly, the ERM prevalence in the Funagata Study, which also used the same grading centre as the BMES, showed a similar prevalence of ERM to that in the BMES.4 Therefore, we speculate that the discrepancy in the prevalence of ERM between previous Asian populations and Caucasian populations could have also been due to systematic differences in the grading and detection of this retinal condition.

Female gender was associated with ERM after adjusting for age, consistent with findings from the Los Angeles Latino Eye Study.8 Although not statistically significant, similar trends were reported in the BMES,5 the Hisayama Study1 and the Funagata Study.4 We found that a narrower retinal arteriolar diameter was associated with ERM. Although narrower retinal arteriolar diameter may represent arteriolosclerosis of the retinal vasculature,18 it could also be a consequence of PMF (PMF distorts the retinal surface and its vessels). The BDES has reported that arterio-venous nicking, a second microvascular sign of atherosclerosis, was associated with ERM,6 lending support to our first speculation.

Interestingly, current smoking was inversely associated with ERM after adjusting for age and gender. This was also observed in the Visual Impairment Project,3 although no adequate explanation was offered for this association. One possibility is a substantial difference in the proportion of male and female current smokers in our study participants. The current smoking rate was 40% among men and only 3% in women.19 As a result, 92.7% of current smokers were men. In addition, current smokers were significantly younger than non-smokers or past smokers (the mean age of current and non-current smokers was 54.7 years and 58.5 years, respectively; p<0.001). As female gender was associated with ERM in our study sample (age-adjusted OR 1.66), we adjusted for age and gender in the association between current smoking and ERM, and found that the inverse association between current smoking and ERM was attenuated after this adjustment (unadjusted OR 0.39 (CI 0.26 to 0.59); age–gender-adjusted OR 0.60 (CI 0.38 to 0.94)). Despite this adjustment, residual influence from the substantial gender difference in the proportion of current smokers cannot be excluded. An alternative explanation proposed by McCarty et al3 could be survival effect caused by higher mortality among male current smokers. However, we cannot verify this possibility in our study.

In our study, myopia and posterior subcapsular cataract were negatively associated with ERM. These inverse associations were both stronger for CMR than for PMF (data not shown). The reduced visibility of the fundus due to cataract could have led to an underdetection of mild ERM. A tessellated fundus appearance in myopic eye or an impaired ability to discriminate ERM because of posterior subcapsular cataract would make it difficult to detect ERM, particularly if these membranes were mild. We could not confirm previous associations of ERM reported with diabetes5, plasma glucose5 or serum cholesterol,1 or a negative association with early ARM.5

The strengths of our study include its large sample size with high response rate. We graded all the fundus signs using standardised protocols at the same grading centre used in the BMES.5 The limitations should also be stated. First, some risk factors such as those previously reported were not examined and we could not assess them. Therefore, residual confounding is possible. Second, some other possible cause of secondary ERM (ie, uveitis) might have been overlooked. This could lead to an overestimation of primary ERM and an underestimation of secondary ERM. Third, there might be systematic differences between the SiMES and the BMES in risk factors for ERM. For example, the prevalence of diabetes was higher in the SiMES as compared with the BMES (23% and 6%, respectively). Another possible difference is that the BMES used film-based photographs to define ERM, while SiMES used digital photography. Digital photography has been shown to be reliable in epidemiological studies when compared with 35 mm film in grading of age-related macular degeneration;20 we assume this may also be applicable to the grading of ERM. Finally, as our study is cross-sectional, we can only speculate on causal relationships from the associations we found.

In summary, we report the prevalence of ERM in a population-based sample of Malays aged 40–80 years. Comparison of age-specific and age-standardised prevalence revealed that the prevalence of ERM in this Malay population was higher than that in the white population of the BMES aged 49–80 years. We could not confirm racial variation between Asians and Caucasians, and speculate that systematic differences in the detection of this condition may be an alternative contributing factor to the variations in the prevalence reported by previous population-based studies.



  • Funding: This study was supported by the National Medical Research Council Grants No. 0796/2003, 0863/2004 and CSI/0002/2005, and Biomedical Research Council Grant No. 501/1/25-5.

  • Competing interests: None.

  • Ethics approval: Ethics approval was obtained from the Institutional Review Board of the Singapore Eye Research Institute, Singapore.

  • Patient consent: Obtained.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.