Prevalence of Refractive Errors and Associated

Risk Factors in Subjects with Type 2 Diabetes

Mellitus
SN-DREAMS, Report 18
Padmaja Kumari Rani, MS, FNB (Retina),1 Rajiv Raman, MS, DNB,1 Sudhir R. Rachapalli, MS, MPH,2
Vaitheeswaran Kulothungan, MSc, MPhil,2 Govindasamy Kumaramanickavel, MD,3
Tarun Sharma, MD, FRCSEd, MBA1

Purpose: To report the prevalence of refractive errors and the associated risk factors in subjects with type
2 diabetes mellitus from an urban Indian population.
Design: Population-based, cross-sectional study.
Participants: One thousand eighty participants selected from a pool of 1414 subjects with diabetes.
Methods: A population-based sample of 1414 persons (age ⬎40 years) with diabetes (identified as per the
World Health Organization criteria) underwent a comprehensive eye examination, including objective and
subjective refractions.
Main Outcome Measures: One thousand eighty subjects who were phakic in the right eye with best
corrected visual acuity of ⱖ20/40 were included in the analysis for prevalence of refractive errors. Univariate and
multivariate analyses were done to find out the independent risk factors associated with the refractive errors.
Results: The mean refraction was ⫹0.20⫾1.72, and the Median, ⫹0.25 diopters. The prevalence of em-
metropia (spherical equivalent [SE], ⫺0.50 to ⫹0.50 diopter sphere [DS]) was 39.26%. The prevalence of myopia
(SE ⬍⫺0.50 DS), high myopia (SE ⬍⫺5.00 DS), hyperopia (SE ⬎⫹0.50 DS), and astigmatism (SE ⬍⫺0.50 cyl)
was 19.4%, 1.6%, 39.7%, and 47.4%, respectively. The advancing age was an important risk factor for the three
refractive errors: for myopia, odds ratio (OR; 95% confidence interval [CI] 4.06 [1.74 –9.50]; for hyperopia, OR
[95% CI] 5.85 [2.56 –13.39]; and for astigmatism, OR [95% CI] 2.51 [1.34 – 4.71]). Poor glycemic control was
associated with myopia (OR [95% CI] 4.15 [1.44 –11.92]) and astigmatism (OR [95% CI] 2.01 [1.04 –3.88]). Female
gender was associated with hyperopia alone) OR [95% CI] 2.00 [1.42–2.82].
Conclusions: The present population-based study from urban India noted a high prevalence of refractive
errors (60%) among diabetic subjects ⬎40 years old; the prevalence of astigmatism (47%) was higher than
hyperopia (40%) or myopia (20%).
Financial Disclosure(s): The authors have no proprietary or commercial interest in any of the materials
discussed in this article. Ophthalmology 2010;117:1155–1162 © 2010 by the American Academy of Ophthalmology.

Refractive errors, especially uncorrected ones, are among
the most common causes of visual impairment in the world;
the World Health Organization estimates this population to
be around 153 million.1 All refractive errors, however, need
not be corrected for a person to have normal function and
this is especially true if the fellow eye vision is 20/20.
The US Centers for Disease Control and Prevention
reported that among individuals aged ⱖ20 years with
diabetes, 11% had visual impairment (i.e., presenting
visual acuity ⬍20/40 in their better-seeing eye while
wearing glasses or contact lenses, if applicable), and
about 65.5% of these cases were correctable.2 In individ-
uals with younger onset diabetes, diabetic retinopathy
was responsible for visual acuity of ⬍20/200 in around
80% of the eyes, whereas in individuals with older onset
diabetes, this was the case in one third.3 Important causes
of visual impairment, present in nearly half of the indi-
viduals with older onset diabetes, included cataract, glau-
coma, and macular degeneration.3
Although there are several reports on prevalence of re-
fractive errors in the general population, only limited data
exist about this in population with diabetes.4 –10 Therefore,
the present, population-based study is aimed at estimating
the prevalence and associated clinical and biochemical risk
factors for refractive errors in populations with type 2
diabetes.
Materials and Methods
Study subjects were recruited from the Sankara Nethralaya Dia-
betic Retinopathy Epidemiology and Molecular Genetic Study.
The study design and research methodology is described in detail
elsewhere.11 The study area was the Chennai metropolis with a
population of 4.3 million distributed in 155 divisions of 10 zones.

© 2010 by the American Academy of Ophthalmology ISSN 0161-6420/10/$–see front matter 1155
Published by Elsevier Inc. doi:10.1016/j.ophtha.2009.10.025
 Ophthalmology Volume 117, Number 6, June 2010
As a sample, a total of 5999 subjects selected from the general
population aged ⱖ40 years was enumerated; multistage, random
sampling was stratified on the basis of economic criteria. The
socioeconomic score was assessed using a multiindex question-
naire. The sample was stratified based on socioeconomic scoring:
low (0 –14), middle (15–28), or high (29 – 42).12 The data were
compared between responders (1563 who visited the base hospital)
and nonresponders (253 who did not visit the base hospital) with
regard to mean age, gender, diabetes status, and mean fasting
blood sugar levels. No differences were observed.
Of the 5999 subjects enumerated, 1414 persons identified with
diabetes, as per World Health Organization criteria13 (both known
and newly diagnosed), were analyzed for the study (96.20% re-
sponse rate for first fasting blood sugar estimation, 85.60% re-
sponse rate for base hospital examination, 8.7% turned out as
nondiabetic after second blood sugar and 0.78% of retinal images
were nongradable). The study was approved by the Institutional
Review Board, and an informed consent was obtained from sub-
jects as per the Declaration of Helsinki.
Visual Acuity and Refractive Status Assessment
Visual acuity was estimated using the modified Early Treatment of
Diabetic Retinopathy Study chart (Light House Low Vision Prod-
ucts, New York, NY); for those who could not read the English
alphabet, Landolt’s ring test was used. Objective refraction (streak
retinoscope, Beta 200, Heine, Germany) was always followed by
a subjective one. If the subject was unable to read the 4/40
(logarithm of the minimum angle of resolution, 1.0) line, vision
was checked at one meter, and if the subject was still unable to
identify any of the largest optotypes, perception of hand move-
ments was checked; and if the vision was less than hand move-
ments, perception of light was tested. Only the right eye of each
subject was considered, and of these, phakic eyes with best cor-
rected visual acuity ⱖ20/40 (logarithm of the minimum angle of
resolution 0.3) were included for analysis. Thus, a total of 1080
subjects were included for estimation of prevalence of refractive
error. We considered 20/40 as a cutoff point to prevent the con-
founding effect of nuclear sclerosis on refractive error estimation
as well as to compare our study results with other epidemiologic
studies done in the general population.
In the overall study group (n ⫽ 1414), refraction data were
available for 1379 subjects; 35 subjects were excluded because
refraction was not done (owing to blind eye or media opacity).
Refractive errors were compared between the overall group (n ⫽
1379) and the study group (n ⫽ 1080).
Definitions of Clinical Variables
Emmetropia was defined as a spherical equivalent between ⫺0.50
and ⫹0.50 diopter sphere (DS).10 Myopia was defined as a spher-
ical equivalent of less than ⫺0.50 DS, and high myopia, a spher-
ical equivalent of less than ⫺5.00 DS.4,10 Hyperopia was defined
as a spherical equivalent of greater than ⫹0.50 DS, and high
hyperopia, a spherical equivalent of more than ⫹5.00 DS.4,10
Astigmatic correction was prescribed in the minus cylinder format,
and astigmatism was defined as a cylindrical error less than ⫺0.50
diopter cylinder in any axis.4,10 Astigmatism was defined as with-
the-rule if the axis lay between 15° on either side of the horizontal
meridian, against-the-rule if the axis lay between 15° on either side
of the vertical meridian, and oblique if the axis lay between 15°
and 75° or between 105° and 165°.4,10 Grading of lens opacities
was performed by using the Lens Opacities Classification System
(LOCS) III (LOCS chart III, Leo T. Chylack, Harvard Medical
School, Boston, MA); significant nuclear sclerosis was defined as
nuclear opalescence of ⱖN2.
1156
The visual impairment rate owing to refractive errors in the
study population was ascertained as per the definition of low vision
and blindness, recommended by the World Health Organization
(low vision, 6/18 –3/60 and blindness, ⬍3/60; visual impairment
included both low vision and blindness data).14
Definitions of Biochemical Variables
Patients were considered to be newly diagnosed diabetics if the
fasting blood glucose level was ⱖ110 mg/dL on 2 occasions.13
Patients were considered to be known diabetics if they were using
hypoglycemic drugs, either oral or insulin, or both.9 Glycemic
control was categorized as normal (glycosylated hemoglobin
[HbA1c] ⬍5.6), good (HbA1c 5.6 –7.0), fair (HbA1c 7.1– 8.0),
unsatisfactory (HbA1c 8.1–10.0), or poor (HbA1c ⬎10).13 High
fasting plasma glucose was considered if the value was ⬎126
mg/dL.15 All subjects underwent measurement of height and
weight, after which the body mass index was calculated using the
formula: weight (kg)/height (m2). Based on body mass index,
individuals were classified as lean (male, ⬍20; female, ⬍19),
normal (male, 20 –25; female, 19 –24), overweight (male, 25–30;
female, 24 –29) or obese (male, ⬎30; female, ⬎29).16 Educational
status was classified as illiterate, primary education (1–5), higher
secondary education (6 –12), or college degree or higher.17
Statistical Analysis
Statistical analyses were performed using the statistical software
(SPSS for Windows, ver. 13.0 SPSS Science, Chicago, IL). The
results were expressed as mean values ⫾ SD if the variables were
continuous, and as percentage, if categorical. The Student t-test for
comparing continuous variables, and the chi-square test, for com-
paring proportions amongst groups were used. The newly diag-
nosed diabetics were given a value of zero for the duration of
diabetes. Both univariate and multivariate logistic regression anal-
yses were performed to study the effect of various risk factors
using refractive error and its components—astigmatism, myopia,
and hyperopia—as dependent variables. From the univariate anal-
ysis, variables with P values ⱕ0.05 and those that were already
established as risk factors were included in the multivariate logistic
regression analysis to derive the parsimonious model. Pⱕ0.05 was
considered significant.
Results
Out of the 1414 subjects with diabetes, 1232 (87.1%) were phakic
in the right eye. The remaining 182 (12.9%) subjects had a history
of cataract surgery and were either pseudophakic or aphakic in the
right eye. Of the 1232 phakic subjects, 1080 (76.37%) were
eligible for assessment of refractive status with best corrected
vision of ⱖ20/40. The remaining 152 (12.34%) subjects had visual
acuity ⬍20/40: 136 owing to cataract (LOCS III grade ⱖ N2,
ⱖC3, ⱖP2), and 16 owing to other ocular diseases.
Overall, the mean objective refraction in study subjects
was ⫹0.34⫾1.79, and the mean subjective refraction was
⫹0.20⫾1.72. There was no significant difference between the
mean objective and subjective refraction (P⫽0.07); refraction data
were analyzed based on subjective refraction. The mean refraction
between the right and left eye was not significant (Pearson corre-
lation, 0.845). Hence, the right eye alone was considered for
assessment of refractive status.
The mean age of the included 1080 subjects was 53.7⫾8.9
years (range, 40 – 81). The mean age of men and women was
54.5⫾9.4 and 52.8⫾8.2 years, respectively (P ⫽ 0.001).
 Rani et al 䡠 Refractive Errors in Type 2 Diabetes Mellitus

Table 1. Age- and Gender-specific Distribution of Refractive Errors among Persons with Type 2 Diabetes Mellitus

Myopia Hyperopia High Myopia Astigmatism

With- Against-
P P P P Oblique P the-rule P the-rule P
Age Group n n (%) Value n (%) Value n (%) Value Overall Value n (%) Value n (%) Value n (%) Value
Overall
40–49 379 74 (19.5) 78 (20.6) 7 (1.8) 125 (33.0) 20 (24.7) 10 (50.0) 95 (23.1)
50–59 423 68 (16.1) 197 (46.5) 5 (1.2) 222 (52.5) 40 (49.4) 5 (25.0) 177 (43.1)
60–69 216 45 (20.8) 129 (59.7) 4 (1.9) 127 (58.8) 15 (18.5) 5 (25.0) 107 (26.0)
⬎69 62 23 (37.1) 25 (40.3) 1 (1.6) 38 (61.3) 6 (7.4) 0 (0.0) 32 (7.8)
Subtotal 1080 210 (19.4) 0.032 429 (39.7) ⬍0.0001 17 (1.6) 0.915 512 (47.4) ⬍0.0001 81 (15.8) 0.459 20 (3.9) 0.036 411 (80.3) 0.089
Men
40–49 208 45 (21.6) 37 (18.3) 4 (1.9) 72 (34.6) 11 (23.9) 7 (50.0) 54 (23.0)
50–59 216 44 (20.4) 81 (37.5) 3 (1.4) 117 (54.2) 18 (39.1) 2 (14.3) 97 (41.3)
60–69 131 29 (22.1) 70 (53.4) 2 (1.5) 80 (61.1) 11 (23.9) 5 (35.7) 64 (27.4)
⬎69 43 18 (41.9) 14 (32.6) 1 (2.3) 26 (60.5) 6 (13.0) 0 (0.0) 20 (8.5)
Subtotal 598 136 (22.7) 0.058 203 (34.0) ⬍0.0001 10 (1.7) 0.959 295 (49.3) ⬍0.0001 46 (15.6) 0.641 14 (4.7) 0.144 235 (79.7) 0.726
Women
40–49 171 29 (17.0) 40 (23.4) 3 (1.8) 53 (31.0) 9 (25.7) 3 (50.0) 41 (23.3)
50–59 207 24 (11.6) 115 (56.1) 2 (1.0) 105 (50.7) 22 (62.9) 3 (50.0) 80 (45.5)
60–69 85 16 (18.8) 59 (69.4) 2 (1.0) 47 (55.3) 4 (11.4) 0 (0.0) 43 (24.4)
⬎69 19 5 (26.3) 11 (57.9) 0 (0.0) 12 (63.2) 0 (0.0) 0 (0.0) 12 (6.8)
Subtotal 482 74 (15.4) 0.503 226 (46.9) ⬍0.0001 7 (1.5) 0.889 217 (45.0) ⬍0.0001 35 (16.1) 0.076 6 (2.8) 0.078 176 (81.1) 0.016

Myopia was defined as a spherical equivalent of less than ⫺0.50 diopter sphere (DS). Hyperopia was defined as a spherical equivalent of greater than ⫹0.50
DS. High myopia was defined as less than ⫺5.00 D. Astigmatism was defined as a cylindrical error less than ⫺0.50 diopter cylinder (DC) in any axis.
Astigmatism was defined as with-the-rule if the axis lay between 15° on either side of the horizontal meridian, against-the rule if the axis lay between 15°
on either side of the vertical meridian, and oblique if the axis lay between 15° and 75° or between 105° and 165°.

The prevalence of emmetropia was 39.3% (424/1080; 95% CI,
36.4 – 42.2); the prevalence, after adjusting for age and gender for
the urban Chennai population was 39.9% (95% CI, 39.9 – 40.1).
The prevalence of emmetropia decreased significantly with in-
creasing age (chi-square test for trend P⬍0.0001). There were 249
men (41.6% of all men) and 175 women (36.3% of all women) in
the emmetropia group.
Table 1 shows age- and gender-specific distribution of refractive
errors in subjects with type 2 diabetes. The overall prevalence of myopia
was 19.4% (95% CI, 17.1–21.8), and that of high myopia, 1.6% (95% CI,
0.8–2.3). The prevalence of myopia and high myopia, after adjusting for
age and gender for the urban Chennai population, was 20.4% (95% CI,
20.4–20.5) and 1.6% (95% CI, 1.6–1.6), respectively. The mean age of
subjects with myopia was 55.1⫾10.2 years, and that of the entire popu-
lation, 53.8⫾8.5 years (P⫽0.015). The mean age of subjects with high
myopia was 53.8⫾ 8.9 years, and that of the entire population, 53.8⫾8.5
years (P⫽0.666). The prevalence of myopia increased significantly with
age (chi-square for trend, P⫽0.032). With regard to gender, there was no
difference in the prevalence of myopia and high myopia.
The overall prevalence of hyperopia was 39.7% (95% CI,
36.4 – 42.6), and that of high hyperopia, 0.2% (95% CI, ⫺0.1 to
0.5). The prevalence of hyperopia, after the age and gender were
adjusted for the urban Chennai population, was 37.9% (95% CI,
37.9 –38.1). Hyperopia was more prevalent among women (46.9 vs
34.0; P⬍0.0001). The mean age of subjects with hyperopia was
56.5⫾7.9 years, and that of the entire population, 51.9⫾9.1 years
(P⬍0.0001). The prevalence of hyperopia increased significantly
with age (chi-square for trend P⬍0.0001) through the age of 69
years; thereafter, it showed little decline.
The prevalence of astigmatism was 47.4% (95% CI, 44.4 –
50.4). The prevalence of astigmatism, after the age and gender
adjustment for the Chennai population, was 46.2% (95% CI,
46.2– 46.3). Astigmatism was more prevalent among men (49.3 vs
45.0; P⬍0.0001). Oblique astigmatism was seen in 15.8% (81/
512); with-the-rule in 3.9% (20/512) and against-the-rule in 80.3%
(411/512). The prevalence of astigmatism increased significantly
with age (chi-square for trend, P⬍0.0001) in the overall group. In
subjects having oblique astigmatism, age and gender did not in-
fluence the prevalence; in subjects having with-the-rule astigma-
tism, the prevalence decreased with age in the overall group
(P⫽0.04) and no difference was noted regarding gender; in sub-
jects having against-the-rule, the prevalence decreased with age,
only in women (P⫽0.02), and no difference was noted in the
overall group and in men.
Table 2 shows the univariate analysis of the risk factors for
refractive errors in persons with type 2 diabetes. Clinical param-
eters that were significantly associated with myopia included in-
creasing age, primary education, and nuclear sclerosis (⬎2); sig-
nificant biochemical parameter was poor glycemic control. For
hyperopia, significant clinical variables were increasing age, fe-
male gender, known diabetes, increased duration of diabetes, nu-
clear sclerosis (⬎2); a significant biochemical variable was low
fasting plasma glucose levels. For astigmatism (all types) and in a
subgroup of individuals with against-the-rule astigmatism, signif-
icant clinical variables were advanced age, greater duration of
diabetes, significant nuclear sclerosis, and diabetic retinopathy; a
significant biochemical variable was poor glycemic control.
Table 3 shows significant risk factors found on multiple logistic
regression analyses. Increased age was significantly associated
with all types of refractive errors: myopia, hyperopia, and astig-
matism (all types) and against-the-rule astigmatism. For myopia,
presence of diabetes for ⬎15 years showed an inverse relation OR
(0.38; 95% CI, 0.18 – 0.81); increasing prevalence of myopia was
noted as the glycemic control worsened from good to poor. For
hyperopia, female gender showed an increased prevalence (OR,
2.00; 95% CI, 1.42–2.82).
Astigmatism (all types) showed an increased prevalence with
poor glycemic control (OR, 2.01; 95% CI, 1.04 –3.88). In subjects
with against-the-rule astigmatism, increased duration of diabetes
⬎15 years was a risk factor (OR, 1.67; 95% CI, 1.05–2.67).

1157
 Ophthalmology Volume 117, Number 6, June 2010

Table 2. Univariate Analysis of Risk Factors for Refractive Errors in Persons with Type 2 Diabetes

Against-the-rule
Myopia Hyperopia Astigmatism Astigmatism
Variables n OR (95% CI) n OR (95% CI) n OR (95% CI) n OR (95% CI)
Clinical variables
Age groups (yrs)
40–49 81 1.00 78 1.00 125 1.00 95 1.00
50–59 73 1.29 (0.89–1.89) 197 3.63 (2.60–5.07)* 222 2.24 (1.68–2.99)* 177 2.15 (1.59–2.91)*
60–69 49 3.50 (2.14–5.74)* 129 9.58 (6.14–14.93)* 127 2.90 (2.05–4.09)* 107 2.93 (2.06–4.18)*
ⱖ70 24 5.01 (2.43–10.32)* 25 5.42 (2.64–11.12)* 38 3.22 (1.85–5.59)* 32 3.19 (1.84–5.52)*
Gender
Male 146 1.00 203 1.00 295 1.00 235 1.00
Female 81 0.79 (0.57–1.10) 226 1.58 (1.21–2.08)* 217 0.84 (0.66–1.07) 176 0.89 (0.69–1.14)
Educational status
Illiterate 3 1.00 21 1.00 16 1.00 15 1.00
Primary (1 to 5) 48 4.00 (1.09–14.62)* 64 0.76 (0.36–1.61) 89 1.54 (0.76–3.12) 75 1.24 (0.61–2.52)
Higher Secondary (6–12) 124 2.40 (0.68–8.45) 261 0.72 (0.36–1.43) 307 1.31 (0.68–2.53) 241 0.96 (0.49–1.86)
Degree and more 52 2.86 (0.79–10.33) 83 0.65 (0.31–1.35) 100 1.14 (0.57–2.27) 80 0.88 (0.43–1.77)
BMI (kg/m2)
Normal 92 1.00 149 1.00 191 1.00 157 1.00
Lean 13 1.14 (0.54–2.43) 25 1.36 (0.72–2.57) 30 1.19 (0.68–2.07) 24 1.10 (0.63–1.94)
Overweight 92 0.86 (0.60–1.24) 189 1.09 (0.81–1.49) 220 0.98 (0.75–1.29) 176 0.94 (0.72–1.24)
Obese 30 0.69 (0.42–1.13) 66 0.93 (0.62–1.39) 71 0.77 (0.54–1.11) 54 0.71 (0.49–1.04)
Diabetes status
Newly detected 57 1.00 62 1.00 90 1.00 341 1.00
Known diabetes 170 0.88 (0.61–1.29) 367 1.76 (1.23–2.49)* 422 1.34 (0.99–1.81) 70 1.36 (0.99–1.86)
Duration of diabetes (yrs)
⬍5 150 1.00 238 1.00 284 1.00 221 1.00
5–10 40 0.93 (0.60–1.43) 76 1.11 (0.78–1.59) 107 1.62 (1.18–2.24)* 87 1.61 (1.16–2.23)*
10–15 23 1.32 (0.74–2.33) 62 2.24 (1.41–3.55)* 54 1.14 (0.77–1.69) 46 1.29 (0.86–1.94)
ⱖ15 14 0.67 (0.35–1.29) 53 1.61 (1.03–2.53)* 67 2.35 (1.54–3.60)* 57 2.37 (1.57–3.60)*
Lens status
Significant nuclear sclerosis (ⱕ2) 99 1.00 188 1.00 221 1.00 175 1.00
Significant nuclear sclerosis (⬎2) 128 2.13 (1.53–2.96)* 241 2.11 (1.61–2.78)* 291 1.83 (1.43–2.32)* 236 1.73 (1.35–2.22)*
Diabetic retinopathy
No diabetic retinopathy 187 1.00 349 1.00 409 1.00 326 1.00
Diabetic retinopathy 40 1.18 (0.77–1.82) 80 1.27 (0.88–1.81) 103 1.49 (1.08–2.05)* 85 1.48 (1.08–2.04)*
Biochemical variables
HbA1c (g%)
⬍5.6 6 1.00 31 1.00 23 1.00 19 1.00
5.6–7.0 63 2.87 (1.14–7.24)* 140 1.23 (0.71–2.14) 144 1.64 (0.95–2.83) 116 1.49 (0.84–2.66)
7.1–8.0 41 2.88 (1.11–7.45)* 79 1.07 (0.59–1.93) 96 1.92 (1.08–3.41)* 77 1.70 (0.93–3.11)
8.1–10.0 69 2.85 (1.14–7.16)* 106 0.85 (0.47–1.48) 145 1.82 (1.05–3.16)* 117 1.65 (0.92–2.93)
⬎10.0 48 3.66 (1.42–9.42)* 73 1.08 (0.59–1.95) 104 2.39 (1.34–4.25)* 82 1.98 (1.08–3.61)*
Fasting plasma glucose (mg/dL)
⬍126 91 1.00 212 1.00 214 1.00 174 1.00
ⱖ126 136 0.98 (0.71–1.36) 217 0.67 (0.51–0.88)* 298 1.15 (0.90–1.46) 237 1.09 (0.85–1.39)

BMI ⫽ body mass index; CI ⫽ confidence interval; FPG ⫽ fasting plasma glucose; HbA1c ⫽ glycosylated hemoglobin; OR ⫽ odds ratio.
BMI: lean (male ⬍20, female ⬍19), normal (male 20 –25, female 19 –24), overweight (male 25–30, female 24 –29) or obese (male ⬎30, female ⬎29).
*P⬍0.05.

Table 4 shows comparative analysis of refractive errors be- Discussion

tween the overall group (n ⫽ 1379) and the study population (n ⫽
1080). The prevalence of myopia was higher in the overall group
(24.1% vs 19.4%; P⫽0.005) and in women (21.9% vs 15.4%;
P⫽0.006); no differences were observed with other refractive errors.
Table 5 shows visual impairment rate due to refractive errors in
the overall population. Visual impairment rate (defined as visual
acuity of ⬍6/18) was 6.2% in the right eye and 3.3% in both eyes.
In the right eye, contributing refractive errors to visual impairment
included astigmatism (33.7%), myopia (31.4%), hyperopia
(26.7%), and high myopia (8.1%); and in both eyes, contributing
refractive errors were myopia (35.6%), astigmatism (31.1%), hy-
peropia (24.4%), and high myopia (8.9%).
The present, population-based study found that the preva-
lence of all types of refractive errors in type 2 diabetes
mellitus was about 60%. The prevalence of myopia, hyper-
opia, and astigmatism in the study population with type 2
diabetes mellitus was 20%, 40%, and 47%, respectively. We
attempted to extrapolate these prevalence estimates to the
urban population of India. Based on the 2001 census, the
urban population was 286 million. Of these, about 114
million were over age 30.18 Assuming the current preva-

1158
 Rani et al 䡠 Refractive Errors in Type 2 Diabetes Mellitus

Table 3. Multivariate Analysis of Risk Factors for Refractive Errors in Persons with Type 2 Diabetes

Against-the-rule
Myopia Hyperopia Astigmatism Astigmatism
Variables OR (95% CI) OR (95% CI) OR (95% CI) OR (95% CI)
Clinical variables
Age group (yrs)
40–49 1.00 1.00 1.00 1.00
50–59 1.24 (0.81–1.89) 3.61 (2.51–5.19)* 2.03 (1.49–2.77)* 1.95 (1.41–2.69)*
60–69 3.16 (1.71–5.86)* 10.67 (6.31–18.07)* 2.36 (1.56–3.56)* 2.41 (1.58–3.66)*
ⱖ70 4.06 (1.74–9.50)* 5.85 (2.56–13.39)* 2.51 (1.34–4.71)* 2.46 (1.31–4.59)*
Gender
Male 1.00 1.00 1.00 1.00
Female 0.85 (0.57–1.28) 2.00 (1.42–2.82)* 0.84 (0.63–1.11) 0.91 (0.68–1.21)
Educational status
Illiterate 1.00 1.00 1.00 1.00
Primary (1–5) 3.46 (0.89–13.45) 0.71 (0.31–1.62) 1.65 (0.79–3.45) 1.34 (0.64–2.81)
Higher secondary (6–12) 2.37 (0.63–8.92) 0.91 (0.43–1.93) 1.41 (0.70–2.81) 1.03 (0.52–2.07)
Degree and more 3.17 (0.82–12.32) 0.89 (0.39–2.02) 1.24 (0.59–2.59) 0.96 (0.46–2.01)
BMI (kg/m2)
Normal 1.00 1.00 1.00 1.00
Lean 1.10 (0.48–2.50) 1.17 (0.58–2.34) 1.08 (0.60–1.95) 1.01 (0.56–1.82)
Overweight 0.94 (0.63–1.39) 1.02 (0.72–1.45) 1.14 (0.85–1.53) 1.07 (0.79–1.44)
Obese 0.90 (0.51–1.60) 0.85 (0.52–1.38) 1.03 (0.68–1.55) 0.89 (0.58–1.37)
Diabetes status
Newly detected 1.00 1.00 1.00 1.00
Known diabetes 0.91 (0.58–1.43) 1.46 (0.94–2.25) 1.07 (0.75–1.51) 1.03 (0.72–1.49)
Duration of diabetes (yrs)
⬍5 1.00 1.00 1.00 1.00
5–10 0.83 (0.51–1.35) 0.67 (0.44–1.04) 1.38 (0.97–1.97) 1.39 (0.97–1.98)
10–15 0.94 (0.48–1.83) 1.52 (0.88–2.63) 0.88 (0.57–1.37) 1.03 (0.66–1.61)
ⱖ15 0.38 (0.18–0.81)* 0.67 (0.38–1.18) 1.61 (0.99–2.59) 1.67 (1.05–2.67)*
Lens status
Significant nuclear 1.00 1.00 1.00 1.00
sclerosis (ⱕ2)
Significant nuclear 1.29 (0.85–1.98) 1.02 (0.72–1.44) 1.19 (0.89–1.59) 1.11 (0.82–1.49)
sclerosis (⬎2)
Diabetic retinopathy
No diabetic retinopathy 1.00 1.00 1.00 1.00
Diabetic retinopathy 1.25 (0.75–2.08) 1.19 (0.77–1.84) 1.17 (0.81–1.67) 1.19 (0.83–1.70)
Biochemical variables
HbA1c (g%)
⬍5.6 1.00 1.00 1.00 1.00
5.6–7.0 2.94 (1.11–7.80)* 1.39 (0.75–2.60) 1.70 (0.96–3.01) 1.57 (0.86–2.86)
7.1–8.0 2.89 (1.04–8.04)* 1.14 (0.57–2.28) 1.71 (0.92–3.18) 1.55 (0.81–2.96)
8.1–10.0 2.99 (1.08–8.31)* 1.05 (0.52–2.11) 1.66 (0.89–3.10) 1.52 (0.79–2.91)
⬎10.0 4.15 (1.44–11.92)* 1.21 (0.57–2.54) 2.01 (1.04–3.88)* 1.66 (0.84–2.29)
Fasting plasma glucose
(mg/dL)
⬍126 1.00 1.00 1.00 1.00
ⱖ126) 0.92 (0.59–1.41) 0.83 (0.56–1.22) 1.12 (0.82–1.54) 1.08 (0.78–1.49)

BMI ⫽ body mass index; CI ⫽ confidence interval; FPG ⫽ fasting plasma glucose; HbA1c ⫽ glycosylated
hemoglobin; OR ⫽ odds ratio.
*P ⬍ 0.05.

lence of diabetes mellitus to be around 28% (in urban
Chennai, above the age of 40 years),19 there would be
approximately a 32 million-person cohort of diabetic indi-
viduals in urban India. Of these 32 million subjects with
diabetes, around 6.4 million will have myopia, about 12.8
million will have hyperopia, and approximately 15 million
will have astigmatism. However, we note that the present
study had a limitation of not collecting the data on the use
of spectacles and habitual visual acuity.
Transient refractive error shift, in association with
plasma glucose concentration, has been reported in clinical
studies; usually myopia occurs with hyperglycemia, and
hyperopia, with hypoglycemia.20,21 However, how common
the occurrence of refractive errors in population with dia-
betes is not well-reported. A community-based study from
Taiwan (n ⫽ 547; ⬎40 years of age) in self-reported sub-
jects with diabetes reported prevalence of myopia (44.1%),
hyperopia (24.1%), and astigmatism (87.8%).3 However,

1159
 Ophthalmology Volume 117, Number 6, June 2010

Table 4. Comparison of Prevalence of Refractive Errors

between the Overall Group* and the Study Group†

Overall Group Study Group

Variables (1379) % (1080) % P Value
Myopia 24.1 19.4 0.005
High myopia 2.3 1.6 0.189
Hypermetropia 36.2 39.7 0.073
Astigmatism 50.4 47.4 0.141
Men
Myopia 26.1 22.7 0.154
High myopia 3 1.7 0.117
Hypermetropia 31.7 34 0.385
Astigmatism 51.2 49.3 0.507
Women Figure 1. Refractive error status comparison among persons with type 2
Myopia 21.9 15.4 0.006
diabetes (present study vs Taiwan study). Same definition of Taiwan study
High myopia 1.6 1.5 0.891
applied to our study data. Refractive error was defined as myopia ⱕ0.50
Hypermetropia 41.3 46.9 0.062
diopters (D) and as hyperopia ⱖ0.50 D. Emmetropia between ⫺0.50 D and
Astigmatism 49.5 45 0.133
⫹0.50D and high myopia was defined as less than ⫺6.00 D. Astigmatism was
defined as a cylindrical error ⱕ0.50 diopter cylinder (DC) in any axis.
*Overall group (n ⫽ 1379) includes refractive error data from the right eye
with inclusion of all visual acuities (including ⱕ20/40) and any lens status
(phakics, aphakics, and pseudopahkics) in the total study population.
†
Study group (n ⫽ 1080) includes refractive error data from the right eye, the Taiwan study; however, hyperopia was more common in
phakic status and visual acuity ⱖ20/40. our study (45.6% vs 24.1%). Ethnic differences could also be
a possible reason for the higher prevalence of high myopia in
the Taiwan study.3,22,23
there were several differences between their study and the Table 6 compares the data of the present study (popula-
present one. Their population was from self-reported indi- tion with diabetes) with the prevalence data of refractive
viduals, whereas ours is a true epidemiologic study with errors in various reported series (general population). The
⬎1000 subjects being analyzed. Second, the definition of prevalence of emmetropia in the present study was noted
refractive errors was different; in our study, myopia and hy- to be around 40% compared with around 25% to 30%
peropia have been defined as a spherical equivalent of less than reported from Chennai Glaucoma Study, Beaver Dam
⫺0.50 DS and more than ⫹0.50 DS, respectively, whereas in Study, Blue Mountain Study, and Tanjog Pagar Singa-
their study, the definition included myopia ⫺0.50 DS or lower pore study.5,22,24,25 This finding suggests that diabetes
and hyperopia ⫹0.50 DS or greater. Thus, there is the possi- mellitus per se does not make an individual ametropic.
bility of overestimating the prevalence of refractive errors. The prevalence of myopia in the present study (around
When we applied the same definition of refractive errors as 20%) was similar to other studies (around 14%–26%), but
was used in the Taiwan study (Fig 1), we found that the for Singapore study (39%); this high prevalence of myopia
prevalence of astigmatism (87.8% vs 63.4%), myopia (44.1% in Singapore could be due to ethnic differences between the
vs 23.9%), and high myopia (13% vs 1.01%) were higher in 2 populations. The converse was true for hyperopia in

Table 5. Visual Impairment Rates* Owing to Refractive Errors in the Overall Study Population with
Diabetes (n-1379)

Overall Group Myopia High Myopia Hyperopia Astigmatism

(n ⴝ 1379) (n ⴝ 333) (n ⴝ 32) (n ⴝ 499) (n ⴝ 553)
Right eye
Low vision 80 ( 5.8%) 27 6 19 28
Blindness 6 (0.4%) 0 1 4 1
Visual impairment 86 (6.2%) 27 (31.3%) 7 (8.1%) 23 (26.7%) 29 (33.7%)
Both eyes (any 1 better eye)
Low vision 44 (3.2%) 16 3 11 14
Blindness 1(0.07%) 0 1 0 0
Visual impairment 45 (3.3%) 16 (35.5%) 4 (8.9%) 11 (24.4%) 14 (31.1%)

Myopia was defined as a spherical equivalent of less than ⫺0.50 diopter sphere (DS). Hyperopia was defined as a spherical
equivalent of greater than ⫹0.50 DS. High myopia was defined as less than ⫺5.00 D. Astigmatism was defined as a
cylindrical error less than ⫺0.50 diopter cylinder (DC) in any axis. Astigmatism was defined as with-the-rule if the axis
lay between 15° on either side of the horizontal meridian, against- the rule if the axis lay between 15° on either side of the
vertical meridian, and oblique if the axis lay between 15° and 75° or between 105° and 165°.
*Visual impairment rate owing to refractive errors in the study population was ascertained as per the definition of low
vision and blindness, recommended by the World Health Organization (Low vision, 6/18 –3/60; blindness, ⬍3/60;
and visual impairment included both low vision and blindness data).

1160
 Rani et al 䡠 Refractive Errors in Type 2 Diabetes Mellitus

Table 6. Comparison of Prevalence Data of Refractive Errors between the Present Study (Population
with Diabetes) and Various Published Studies (General Population)

Age Emmetropia Myopia Hyperopia Astigmatism

Study n (yrs) Population (%) (%) (%) (%)
Present study* 1080 ⱖ40 Type 2 Diabetes 39.2 21.1 39.7 47.4
Chennai Glaucoma 3143 ⱖ40 Urban General 29.4 18.4 52.3 53.0
Study20* population
Beaver Dam study5* 4275 ⱖ40 General population 24.8 26.2 49.0 NR
Blue mountain 3174 ⱖ50 General population 28.6 14.4 57.0 NR
study25*
Tanjog Pagar 1232 ⱖ40 General population 32.9 38.7 28.4 37.8
Singapore22*

NR ⫽ Not reported.
*Myopia was defined as a spherical equivalent of less than ⫺0.50 diopter sphere (DS). Hyperopia was defined as a
spherical equivalent of greater than ⫹0.50 DS. Astigmatism was defined as a cylindrical error less than ⫺0.50 diopter
cylinder (DC) in any axis. Astigmatism was defined as with-the-rule if the axis lay between 15° on either side of the
horizontal meridian, against- the rule if the axis lay between 15° on either side of the vertical meridian, and oblique
if the axis lay between 15° and 75° or between 105° and 165°.

Singapore (around 28%); the prevalence of hyperopia in the
present study was somewhat less compared with other series
(around 40% vs 52%, 49%, 57%). The prevalence of astig-
matism was higher in the general population of Chennai
compared with the present study (53% vs 47%); however,
the prevalence was lower in Singapore (around 38%). The
exact reason for these differences in the prevalence esti-
mates is unknown. It could be because of the influence of
biochemical variables such as blood glucose and glycosy-
lated hemoglobin on the refractive status in population with
diabetes;20,21 however, to prove such a cause and effect
relationship, longitudinal studies are warranted.
Logistic regression analysis revealed advancing age as
the most significant risk factor common to all the refractive
errors—myopia, hyperopia, astigmatism (all types), and
against-the-rule astigmatism. Similar findings were ob-
served in studies conducted worldwide in the general pop-
ulation as well as in the Taiwan study involving population
with diabetes.3-8 Age, as a risk factor, could be due to the
biometric changes in the optical system of lens. To prevent
a confounding effect of nuclear sclerosis on the prevalence
of refractive errors, we excluded subjects with visual acuity
ⱕ20/40. That is why, even though significant nuclear scle-
rosis was found as a common risk factor for the 3 refractive
errors on univariate analysis, the significance disappeared
on multiple logistic regression analysis.
Worsening of glycemic control resulted in the risk of
myopia increasing by nearly 4-fold; hyperglycemia causing
myopic shift is a known phenomenon reported earlier.21
Hyperopia showed an increased prevalence through age
69 (10-fold); thereafter, there was a decline (5-fold). A
similar pattern was found in various studies conducted in
the general population.9,10 This could be because of the
effect of nuclear sclerosis inducing a myopic shift with
advancing age. Apart from age, female gender was a sig-
nificant risk factor for hyperopia, with odds of 1.89 times.
Female gender as a risk factor for hyperopia was also
observed earlier.4,10
Women were found to have shorter axial lengths and
hence more risk for hyperopia.26 Low fasting glucose and
known diabetes status were associated with hyperopia on
univariate analysis. The influence of low plasma glucose
(both acute and chronic) inducing hyperopic shift is a
known phenomenon.20 We found an inverse relationship
between myopia prevalence and increased duration of dia-
betes (⬎15 years). A similar finding was observed in a study
conducted in persons with type 1 diabetes.27
Astigmatism, especially against-the-rule (nearly 80%),
was found in the present study, and this could be due to the
reduced lid laxity (age-related) resulting in flattening of
vertical meridian of the cornea.28,29 Risk of astigmatism
increased nearly 2-fold in individuals present with poor
glycemic control. Apart from age influence, astigmatism in
diabetes may be because of alterations in corneal topogra-
phy from blood sugar.21 The presence of diabetic retinopa-
thy and increased duration of diabetes were associated with
astigmatism only on univariate analysis; significance disap-
peared on multivariate analysis.
One of the important causes of visual impairment in the
general population worldwide is uncorrected refractive er-
rors. The present study observed that around 6% have visual
impairment in 1 eye, and around 3% in both the eyes owing
to refractive errors, in subjects with diabetes, above the age
of 40 years. Extrapolating these estimates to approximately
32 million urban subjects with diabetes mellitus, around 2
million will have visual impairment in 1 eye, and around 1
million in both the eyes. Therefore, creating awareness and
conducting screening programs for the detection of correct-
able refractive errors and treating them with suitable glasses
will have tremendous impact in reducing the magnitude of
visual impairment in subjects with diabetes mellitus.
References
1. World Health Organization Fact file. Blindness and Visual
Impairment. Available at: http://www.who.int/features/factfiles/
vision/02_en.html. Accessed October 11, 2009.
2. CDC. National diabetes fact sheet: United States, 2005. Avail-
able at: http://www.cdc.gov/diabetes/pubs/pdf/ndfs_2005.pdf.
Accessed March 24, 2009.

1161
 Ophthalmology Volume 117, Number 6, June 2010
3. Chen SJ, Tung TH, Liu JH, et al. Prevalence and associated
factors of refractive errors among type 2 diabetics in Kinmen,
Taiwan. Ophthalmic Epidemiol 2008;15:2–9.
4. Dandona L, Dandona R, Naduvilath TJ, et al. Refractive errors
in an urban population in southern India: the Andhra Pradesh
Eye Disease Study. Invest Ophthalmol Vis Sci 1999;40:
2810 – 8.
5. Wang Q, Klein BE, Klein R, Moss SE. Refractive status in the
Beaver Dam Eye Study. Invest Ophthalmol Vis Sci 1994;35:
4344 –7.
6. Katz J, Tielsch JM, Sommer A. Prevalence and risk factors for
refractive errors in an adult inner city population. Invest
Ophthalmol Vis Sci 1997;38:334 – 40.
7. Eye Diseases Prevalence Research Group. The prevalence of
refractive errors among adults in the United States, Western
Europe, and Australia. Arch Ophthalmol 2004;122:495–505.
8. Wensor MD, McCarty CA, Taylor HR. The prevalence and
risk factors of myopia in Victoria, Australia. Arch Ophthalmol
1999;117:658 – 63.
9. Wu SY, Nemesure B, Leske MC, Barbados Eye Study Group.
Refractive errors in a black adult population: the Barbados
Eye Study. Invest Ophthalmol Vis Sci 1999;40:2179 – 84.
10. Raju P, Ramesh SV, Arvind H, et al. Prevalence of refractive
errors in a rural South Indian population. Invest Ophthalmol
Vis Sci 2004;45:4268 –72.
11. Agarwal S, Raman R, Paul PG, et al. Sankara Nethralaya-
Diabetic Retinopathy Epidemiology and Molecular Genetic
Study (SN-DREAMS 1): study design and research method-
ology. Ophthalmic Epidemiol 2005;12:143–53.
12. Oakes JM, Rossi PH. The measurement of SES in health
research: current practice and steps towards a new approach.
Soc Sci Med 2003;56:769 – 84.
13. Expert Committee on the Diagnosis and Classification of
Diabetes Mellitus. Report of the expert committee on the
diagnosis and classification of diabetes mellitus. Diabetes
Care 2003;26(suppl):5S–20.
14. ICD-10: International Statistical Classification of Diseases and
Related Health Problems. Tenth rev. Geneva: World Health
Organization; 1993. Code-H54.
15. American Diabetes Association. Tests of glycemia in diabetes.
Diabetes Care 2000;23(suppl):S80 –2.
16. Mohan V, Vijayaprabha R, Rema M, et al. Clinical profile of
lean NIDDM in South India. Diabetes Res Clin Pract 1997;
38:101– 8.
17. Rani PK, Raman R, Subramani S, et al. Knowledge of diabetes
and diabetic retinopathy among rural populations in India, and
Footnotes and Financial Disclosures
Originally received: March 24, 2009.
Final revision: October 11, 2009.
Accepted: October 13, 2009.
Available online: February 16, 2010. Manuscript no. 2009-417.
1
Shri Bhagwan Mahavir Vitreoretinal Services, Sankara Nethralaya,
Chennai, Tamil Nadu, India.
2
Department of Preventive Ophthalmology, Sankara Nethralaya, Chennai,
Tamil Nadu, India.
3
Department of Molecular Genetics, Sankara Nethralaya, Chennai, Tamil
Nadu, India.
1162
the influence of knowledge of diabetic retinopathy on attitude
and practice. Rural Remote Health [serial online] 2008;8:838.
Available at: http://www.rrh.org.au/publishedarticles/article_
print_838.pdf. Accessed September 26, 2009.
18. Age-wise distribution of population, growth of population,
India. Available at: http://www.infobanc.com/india_view/key_
statictics/stat_india_17.htm. Accessed October 11, 2009.
19. Raman R, Rani PK, Reddi Rachepalle S, et al. Prevalence of
diabetic retinopathy in India: Sankara Nethralaya Diabetic
Retinopathy Epidemiology and Molecular Genetics Study re-
port 2. Ophthalmology 2009;116:311– 8.
20. Tai MC, Lin SY, Chen JT, et al. Sweet hyperopia: refractive
changes in acute hyperglycemia. Eur J Ophthalmol 2006;16:
663– 6.
21. Sonmez B, Bozkurt B, Atmaca A, et al. Effect of glycemic
control on refractive changes in diabetic patients with hyper-
glycemia. Cornea 2005;24:531–7.
22. Wong TY, Foster PJ, Hee J, et al. Prevalence and risk factors
of refractive errors in adult Chinese in Singapore. Invest
Ophthalmol Vis Sci 2000;41:2486 –94.
23. Saw SM, Gazzard G, Koh D, et al. Prevalence rates of refrac-
tive errors in Sumatra, Indonesia. Invest Ophthalmol Vis Sci
2002;43:3174 – 80.
24. Prema R, George R, Sathyamangalam Ve R, et al. Comparison
of refractive errors and factors associated with spectacle use in
a rural and urban South Indian population. Indian J Ophthal-
mol 2008;56:139 – 44.
25. Attebo K, Ivers RQ, Mitchell P. Refractive errors in an older
population: the Blue Mountains Eye Study. Ophthalmology
1999;106:1066 –72.
26. George R, Paul PG, Baskaran M, et al. Ocular biometry in
occludable angles and angle closure glaucoma: a population
based survey. Br J Ophthalmol 2003;87:399 – 402.
27. Pensyl CD, Harrison RA, Simpson P, Waterbor JW. Distribu-
tion of astigmatism among Sioux Indians in South Dakota.
J Am Optom Assoc 1997;68:425–31.
28. Jacobsen N, Jensen H, Lund-Andersen H, Goldschmidt E. Is
poor glycaemic control in diabetic patients a risk factor of
myopia? Acta Ophthalmol 2008;86:510 – 4.
29. Gudmundsdottir E, Jonasson F, Jonsson V, et al, Iceland-
Japan Co-Working Study Groups. “With the rule” astigmatism
is not the rule in the elderly. Reykjavik Eye Study: A popu-
lation based study of refraction and visual acuity in citizens of
Reykjavik 50 years and older. Acta Ophthalmol Scand 2000;
78:642– 6.
Financial Disclosure(s):
The authors have no proprietary or commercial interest in any of the
materials discussed in this article.
Funded by the RD Tata Trust, Mumbai.
Correspondence:
Dr. Tarun Sharma, MD, FRCSEd, MBA, Principal Investigator & Director,
Shri Bhagwan Mahavir Vitreoretinal Services, Sankara Nethralaya, 18
College Road, Chennai, 600006, Tamil Nadu, India. E-mail: drtaruns@
gmail.com.