Refractive error and glaucoma

Kirsti Grødum, Anders Heijl and Bo Bengtsson

Department of Ophthalmology, Malmö University Hospital, Malmö, Sweden

ABSTRACT.
Purpose: To study the association between refractive error, glaucoma damage ton & Tomlinson 1973; Perkins & Phelps
and IOP in a large population. 1982; Wilson et al. 1987; Charliat et al.
Methods: We examined 32,918 citizens of the city of Malmö, Sweden, 57–79 1994). Case-control studies are suscep-
years of age, searching for individuals with undetected glaucoma. Refraction tible to selection bias and need to be con-
was measured with autorefractors. Glaucoma damage was defined as reproduc- firmed in population-based studies. In
ible visual field defects with the Humphrey Full Threshold 24–2 program. the Blue Mountains Eye Study (Mitchell
Results: Glaucoma prevalence was clearly associated with refractive state, in- et al. 1999), myopic individuals had a
creasing gradually with increasing myopia. This was seen both in males and two- to three-fold increased risk of glau-
coma compared to that of nonmyopic in-
females and persisted over the full age range. Glaucoma was significantly more
dividuals.
common in myopic than in hyperopic eyes with low IOP readings (pΩ0.024).
We performed a very large population
The overrepresentation of glaucoma in myopic eyes declined with increasing
survey in order to identify patients with
IOP and no relationship was observed in eyes with IOP Ø31 mmHg. undetected glaucoma to recruit partici-
Conclusion: In this large population, the prevalence of glaucoma increased with pants for the Early Manifest Glaucoma
increasing myopia. The association between myopia and glaucoma was strong Trial (Leske et al. 1999). The resulting
at lower IOP levels, and weakened gradually with increasing IOP. Our findings material, consisting of more than 30,000
indicate that myopia is an important risk factor for glaucoma and particularly individuals in Malmö alone, offered an
for normal tension glaucoma. unusual opportunity to study the re-
lationships between glaucoma and several
Key words: glaucoma – refractive error – myopia – intraocular pressure. risk factors.
The purpose of the present investiga-
Acta Ophthalmol. Scand. 2001: 79: 560–566 tion was to study the relationship be-
Copyright c Acta Ophthalmol Scand 2001. ISSN 1395-3907 tween refractive error and glaucoma,
more specifically to see whether myopia
could be confirmed as an important risk
factor for glaucoma, and if so whether
the association between myopia and glau-

O ver the last decades it has become

increasingly clear that the relation-
ship between glaucoma and intraocular
ersson 1989; Odberg 1993). Lately, glau-
coma has often been viewed primarily as
a neuropathy, and intraocular pressure is
coma depends on IOP.

pressure (IOP) is more complex than pre-
viously presumed. Several epidemio-
logical studies have shown that both nor-
mal tension glaucoma and ocular hyper-
tension are common entities (Hollows &
Graham 1966; Perkins 1973; Bengtsson
1981; Shiose et al. 1991). Longitudinal
studies have demonstrated that a ma-
jority of patients with ocular hyperten-
sion do not develop signs of glaucoma-
tous damage even after very long follow-
up without IOP-reducing therapy (Linnér
1976; Kitazawa et al. 1977; Lundberg et
al. 1987; Schulzer et al. 1991). As a result,
optic nerve susceptibility to IOP is now
believed to vary among individuals, and
risk factors for glaucoma, other than el-
evated intraocular pressure, are thought
to be of great importance as predictors
for the development of the disease (And-
currently considered a risk factor rather
than a causative agent (Gupta & Weinre-
ib 1997).
Myopia is one of the risk factors for
glaucoma that is most commonly men-
tioned in textbooks. There are several
theories as to why glaucoma should be
more frequent in myopic individuals than
others. A number of studies have found
higher IOP in myopic individuals com-
pared to non-myopic individuals (Abdal-
la & Hamdi 1970; Tomlinson & Phillips
1970; David et al. 1985, 1987), suggesting
that the relationship between glaucoma
and myopia might be pressure mediated.
Investigations studying the relationship
between myopia and glaucoma, defined
as glaucoma damage, are mainly case-
control studies, and the results partially
conflicting (Drance et al. 1973; Leigh-
Material and Methods
Recruitment
The Early Manifest Glaucoma Trial
(EMGT) is a randomised prospective
trial involving patients with newly de-
tected, previously untreated glaucoma
(Leske et al. 1999). In order to recruit pa-
tients for the EMGT, we conducted a
large eye survey of citizens of Malmö
(total population: 250,000). Between
1992 and 1997 we invited all persons in
certain age cohorts registered as residents
in the city of Malmö, who had not visited
the Eye Department in Malmö within the
preceding year. Both men and women
born between 1918 and 1932 were invited.
The survey was extended to women born

560
1933–1939. One letter of invitation was
sent to each person, offering a free eye
health examination and an appointment.
The invitation was accompanied by a
short questionnaire regarding eye history
and ongoing medications. The study was
approved by the Ethics Committee of the
University of Lund.
Screening methods
The self-reported ophthalmic history was
updated at the time of examination. Re-
fraction and corrected visual acuity were
measured using computerised autorefrac-
tors having a built-in function for deter-
mination of visual acuity (Humphrey Au-
torefractor 595, Humphrey Systems, Du-
blin, CA). If visual acuity was less than
0.8, subjective refraction was performed
in the autorefractor. In cases of non-com-
pliance, the patient was tested with his or
her own eyeglasses or with the correction
obtained through manual subjective re-
fraction. IOP was measured with Gold-
mann applanation tonometry. Monosco-
pic fundus colour photographs were ob-
tained, usually with dilated pupils, using
Topcon non-mydriatic TRC-NW3 fundus
cameras (Tokyo Optical Company,
Tokyo, Japan) and Kodachrome 64 ASA
slide film. The picture angle was 50 æ with
the optic disc placed centrally in the im-
age. All images were inspected at tenfold
magnification and searched for signs of
glaucomatous disc or RNFL damage by
at least one of the authors (BB). A person
was considered screening positive if one
or more of the following characteristics
were present
1. localised narrowing of the optic disc
rim
2. vertically elongated cupping
3. localised RNFL defects
4. IOP ⬎25 mmHg.
Subjects who screened positive were
called back for a post-screening examina-
tion.
Post-screening examination
At the post-screening examination all pa-
tients were subjected to a full eye exami-
nation performed by an experienced oph-
thalmologist, including determination of
refraction and visual acuity, tonometry,
slit lamp examination, ophthalmoscopy
and computerised static threshold per-
imetry using the Humphrey Full Thresh-
old 24–2 program. Visual field test results
were classified with the Glaucoma Hemi-
field Test (GHT) (Åsman & Heijl 1992a,
Fig. 1. Screening and study flowchart.
1992b). Patients with GHT results ‘‘Out-
side Normal Limits’’ or ‘‘Borderline’’
were re-examined whenever possible at a
second post-screening visit one or two
weeks later.
The definition of glaucoma used in this
study was in principle based on the pres-
ence of repeatable visual field defects in
Humphrey 24–2 Full Threshold tests
compatible with glaucoma and not ex-
plained by other ocular or neurological
causes. Only subjects having previously
undiagnosed primary open-angle glau-
coma, normal tension glaucoma or
pseudoexfoliation glaucoma were in-
cluded.
Small visual field defects would have to
be outside normal limits by the Glau-
coma Hemifield Test in the same area on
two consecutive tests on different days.
Sectors 1 and 2 were considered as the
same area. The eyes were also classified
as glaucomatous when the Glaucoma
Hemifield Test was ‘‘Borderline’’ in one (7
eyes) or two (1 eye) of the consecutive
tests if the optic disc showed obvious,
corresponding glaucomatous changes.
Our criteria were thus the same as the
EMGT visual field eligibility criteria (Le-
ske et al. 1999).
A total of 540 eyes were classified as
glaucomatous in the present study.
EMGT criteria had been used in 475 eyes
(88%). In 40 (7.4%) of the 540 glaucoma-
tous eyes analysed, the diagnosis was
based on typical glaucomatous visual
field defects at a single available test with
corresponding, obvious changes of the
optic nerve head. Declining health or
death was the main reason for complet-
ing only one visual field test. In 25 eyes
(4.6%), we were unable to obtain any vis-
ual field data. The glaucoma diagnosis
was then based on optic nerve head
photographs, only including patients
with advanced glaucomatous changes.
End-stage optic nerve head cupping and
very little remaining visual function was
typically what prevented these patients
from completing the visual field test.
General disease or disablement was the
other main reason. Four persons who
screened positive refused to participate in
the post-screening examination (Fig. 1).
Thus, the definition of early glaucoma
was independent of IOP.
Analyses
We calculated the frequencies of newly
detected glaucoma in four refractive
groups; moderate to high myopia (ƪ3
D), low myopia (⬎ª3 D to ƪ1 D), em-
metropia (⬎ª1 D to ⬍π1 D) and hyper-
opia (⭓π1 D), where all refractive errors
were expressed as equivalent spherical
values.
The refractive groups are identical to
the Blue Mountain Eye Study in order to
facilitate comparisons between the two
studies. The material was then stratified
into different age groups, as well as into
eyes with or without significant cataract
(defined as presence of cataract deter-
mined by an ophthalmologist, plus visual
acuity below 0.7). Pseudophakic eyes
were excluded. We also stratified the ma-
561
terial into five groups according to IOP:
Æ15 mmHg, 16–20 mmHg, 21–25
mmHg, 26–30 mmHg and Ø31 mmHg.
We chose to analyse eyes rather than
individuals since glaucomatous changes,
refractive error and IOP often are asym-
metrically distributed. Logistic regression
was used. In the model glaucoma (as the
dependent variable) was explained by
IOP, IOP2, spherical equivalent (SEq),
SEq2, interaction IOP*SEq, age, gender
and interaction age*gender, thus ad-
justing for age and gender. We also added
pseudoexfoliations classifications to the
model, in order to test their statistical sig-
nificance. We used the general estimating
equation (GEE) method by Liang and
Zeger (1986) to correct for intraindividu-
al correlation between eyes. ANOVA tests
were used to compare mean IOP in differ-
ent refractive groups.
Results
Publicly available records indicated that
the total number of Malmö citizens born
between 1918 and 1939 was 46,614. After
excluding 4,117 patients who had at-
tended the Eye Department of Malmö
University Hospital in the previous year,
42,497 citizens were invited to the screen-
Table 1. Prevalence of glaucoma in screened subjects.
Patients with newly detected glaucoma
Patients with previously diagnosed glaucoma
Patients with secondary or angle closure glaucoma
Normal
Total
Table 2. Number of eyes excluded, reasons for exclusion and total number of normal and glau-
comatous eyes included in the study.
Total number screened
Missing IOP or refractive values
Pseudophakic/aphakic
Non-eligible glaucoma
Eligible
Eligible glaucoma eyes, total
Eligible glaucoma eyes with pseudoexfoliation
Eligible glaucoma eyes with IOP Æ20 mmHg
Eligible glaucoma eyes with IOP ±20 mmHg
562
ing. The attendance rate was 77.5%
(32,918 of 42,497, Fig. 1). At the time of
the screening examination, all partici-
pants were between 57 and 79 years of
age. We identified 402 individuals with
previously undiagnosed primary open-
angle glaucoma, normal tension glau-
coma or pseudoexfoliation glaucoma
(Table 1), corresponding to 1.22% of
those screened. We found an additional
83 individuals with glaucoma who had
been previously diagnosed and 8 subjects
with other types of glaucoma (angle-clo-
sure glaucoma, secondary glaucoma).
The total number of eyes with newly
detected open-angle glaucoma was 540
(0.83%), 258 right eyes and 282 left eyes.
A total of 63,745 eyes were included in
the current analyses, among them 524 of
the 540 glaucoma eyes. We excluded eyes
in which we had been unable to obtain
visual acuity or IOP measurements, those
with previously diagnosed or other types
of glaucoma and eyes that had previously
undergone cataract surgery (Table 2).
The prevalence of myopia ƪ1 D was
13.7%. The prevalence of newly detected
glaucoma increased with increasing my-
opia (p⬍0.0001, Table 3). Thus the preva-
lence was 0.6% in eyes with hyperme-
tropia, compared to 0.9% in emmetropic
eyes and 1.5% in the group with moder-
ate to high myopia (Fig. 2). This strong
n %
402 1.22
83 0.25
8 0.02
32,425 98.50
32,918 100.00
n Discussion
(eyes) % The results from this very large survey
65,836
1,089 1.65
847 1.31
138 0.21
63,745 96.82
521 0.82
90 0.14
282 0.44
149 0.23
Table 3. Estimated coefficients and signifi-
cances in the logistic regression model with
glaucoma as the dependent variable. SEq:
spherical equivalent. Note that the gender co-
efficient (1 for males, 0 for females) is compen-
sated for by the interaction gender*age coef-
ficient.
Variable Coefficient p
SEq ª0.373 0.000
SEq2 ª0.019 0.000
IOP 0.372 0.000
IOP2 ª0.002 0.000
SEq*IOP 0.008 0.024
Gender ª2.670 0.145
Age 0.064 0.000
Gender*age 0.038 0.146
constant ª14.689 0.000
relationship between glaucoma and re-
fractive error could be observed in both
men and women (Fig. 1), and remained
the same after excluding subjects with
significant cataract (VA⬍0.7); it also per-
sisted across all age groups (Fig. 3).
The interaction between IOP and re-
fraction was statistically significant (pΩ
0.024, Table 3). In eyes with IOP Æ15
mmHg, glaucoma was four times more
common in myopic than in hyperopic
eyes. This overrepresentation of glau-
coma in myopic eyes declined gradually
with increasing IOP, and no relationship
between refraction and glaucoma damage
was found in eyes with IOP Ø31 mmHg
(Fig. 3B–F). The addition of pseudoex-
foliations to the logistic regression model
was not statistically significant (pΩ0.13).
Mean IOP was 15.7 mmHg in normal
hyperopic eyes and increased gradually
with increasing myopia to 16.2 mmHg in
eyes with moderate to high myopia
(p⬍0.0001). An opposite but not signifi-
cant tendency was seen in the glaucoma
group (Table 4).
confirm and add important detail to the
association between myopia and glau-
coma found in the Blue Mountains Eye
Study (Mitchell et al. 1999) and in several
case-control studies (Leighton & Tomlin-
son 1973; Perkins & Phelps 1982; Wilson
et al. 1987). The Malmö Eye Survey pro-
vided us with a population-based sample,
which was nine times as large as that used
in any earlier study.
 We found a strong association be-
tween glaucoma and myopia but also
that the relationship depended strongly
upon IOP. The overrepresentation of
glaucoma in the myopic group declined
gradually with increasing IOP. There
was a striking, almost linear relationship
between the prevalence of glaucoma and
myopia in the group with IOP Æ15
mmHg (Fig. 4B) that flattened out alto-
gether in eyes with IOP Ø31 mmHg
(Fig. 4F) and this difference illustrates
the significant interaction between IOP
and refraction. The fact that the re-
lationship was so strong at lower IOP

Fig. 2. Prevalence of glaucoma in eyes with different refractive errors, Fig. 3. Prevalence of glaucoma in eyes with different refractive errors in
men and women. different age groups.

Fig. 4. A–F. Prevalence of glaucoma in eyes with different refractive errors at different IOP-levels.

563
Table 4. Mean IOP in normal and glaucomatous eyes related to refractive error.
Normal group Glaucoma group
Mean IOP Mean IOP
(mmHg) n (mmHg) n overlooked in the screening process and
Hyperopia (Øπ1 D) 15.7 31,404 22.7 175
Emmetropia (⬎ª1 D to ⬍π1 D) 15.8 23,238 21.3 219
Low myopia (⬎ª3 D to ƪ1 D) 16.0 5,322 21.8 80
Moderate to high myopia (ƪ3 D) 16.2 3,257 21.2 50
Total 15.8 63,221 21.9 524
levels and disappeared altogether at high partment were excluded from this study.
IOP levels suggests that myopia is a par- We also chose to exclude 83 patients
ticularly important risk factor for nor- with previously diagnosed glaucoma
mal tension glaucoma. who attended the screening, but had
The definition of glaucoma is of great been cared for by ophthalmologists out-
importance when examining potential side the Department of Ophthalmology
risk factors for the disease. Few large in Malmö. Most of them probably had
studies have addressed the association elevated IOP when detected. We can,
between refraction and glaucoma using therefore, presume that the IOP among
a glaucoma definition based on manifest individuals with newly detected glau-
damage. Modern definitions of glau- coma in our material were slightly lower
coma usually stress damage and de-em- than if no such exclusions had been
phasise IOP (Andersson 1989; Odberg made. It is often presumed that myopic
1993; Gupta & Weinreib 1997). We used individuals are more likely to see an
a damage-based definition independent ophthalmologist than others, and, there-
of IOP, documented with modern diag- fore, have a greater likelihood of having
nostic techniques according to current glaucomatous changes detected. If this
standards. Much of the existing data re- was the case in the Malmö population,
garding the relationship between refrac- a greater proportion of myopic persons
tion and glaucoma has been derived with glaucoma would have been ex-
from case-control studies of clinical cluded from the screening than non-my-
populations (Drance et al. 1973; Leigh- opic ones. Thus, excluding patients with
ton & Tomlinson 1973; Perkins & Phelps previously diagnosed glaucoma is not
1982; Wilson et al. 1987; Charliat et al. likely to have weakened the association
1994). Such studies may be influenced between myopia and glaucoma and can-
by selection bias. It is, therefore, reassur- not, in our opinion, alter the main con-
ing that population-based materials sup- clusions from the current study.
port the associations found in these Strict diagnostic criteria were used in
studies. this study, as we wanted a high degree
Only phakic eyes were included in this of certainty that normal eyes were not
study. Eyes with intraocular lenses have falsely labelled as glaucomatous. Some
refractive powers that differ from orig- individuals with early stages of glau-
inal preoperative values, and these eyes coma have certainly been classified as
were therefore not included. Eyes with normal; we saw eyes with seemingly
cataract-induced myopia could also clearly glaucomatous optic disc changes
present a bias, but recalculations after that did not fulfil our criteria for visual
exclusion of patients with significant cat- field abnormality. We see no reason to
aract (VA ⬍0.7) did not alter our re- believe, however, why the existence of
sults. such patients with probable but uncer-
There are some potential sources of tain glaucoma should jeopardise the re-
bias in our study. Patients who had at- sults of this study.
tended the eye department at Malmö The initial screening was based on op-
University Hospital during the preced- tic nerve head photographs, and visual
ing year were not invited to the screen- field testing was not performed at the
ing. This means that all glaucoma pa- screening. Myopic optic nerve heads can
tients previously known at the eye de- sometimes have a configuration that
564
makes the assessment of glaucomatous
changes more difficult. It is possible that
some myopic individuals with IOP Æ25
and visual field defects may have been
classified as normal. This error, if pres-
ent, would have weakened any associ-
ation between myopia and glaucoma,
and should therefore not invalidate our
conclusions.
It is easier to detect glaucoma damage
in larger optic discs than in smaller ones
(Heijl & Molder 1993). If myopic eyes
have larger optic discs this would result
in a possible bias. The Baltimore Eye
Study did, however, not find any sig-
nificant association between refractive
error and optic disc size or topography
(Varma et al. 1994). Jonas and co-
workers found that the optic nerve head
morphometry does not differ signifi-
cantly between normal eyes and glau-
coma eyes for myopic refractive errors
smaller than 8 D (Jonas & Papastho-
poulos 1996). Only three of the glau-
comatous eyes included in this study
had a refractive error below ª8 D.
There was a clear overrepresentation of
glaucoma among eyes with low to mod-
erate myopia, as well as in eyes with
high myopia in our material. This indi-
cates that the relationship between glau-
coma and myopia is genuine and not a
result of bias due to differences in optic
nerve head configuration.
Myopia in itself may sometimes give
superotemporal defects, irregular defects
due to myopic choroidal dystrophy or
enlargement of the blind spot (Greve &
Furuno 1980; Nicolela & Drance 1996).
These defects are usually easily dis-
tinguished from defects specific to glau-
coma. If encountered visual field defects
were in themselves not typical and com-
pletely convincing, we classified an eye
as glaucomatous only in the presence of
corresponding glaucomatous optic nerve
head changes. Thus, it is very unlikely
that eyes with myopic visual field defects
would have been erroneously classified
as glaucomatous.
The IOP in the non-glaucomatous
population was 0.5 mm higher in the
group with moderate to high myopia
than in the group with hyperopia. This
is in line with the IOP difference be-
tween myopic and non-myopic individ-
uals found in the Blue Mountains Study
(Mitchell et al. 1999). Abdalla and
Hamdi (1970) found a slightly higher
difference, and Tomlinson and Philips
(1970) found higher IOP values in sub-
jects with longer axial lengths. Even
though no relationship was found in the
case-control study conducted by Daubs
and Crick (Daubs & Crick 1981), there
seems to be convincing evidence for a
slightly higher IOP in myopic normal
subjects than in nonmyopic ones. It
could be tempting to assume that the re-
lationship between myopia and glau-
coma is mediated by IOP, since mean
IOP generally is higher in myopic indi-
viduals than in nonmyopic individuals.
This was, however, not the case in this
glaucoma material. On the contrary,
mean IOP was lower in the myopic glau-
coma eyes than in the hyperopic glau-
coma eyes. Furthermore, the logistic re-
gression model showed that IOP and re-
fraction are both independent risk
factors for glaucoma in addition to
having an interacting effect (Table 3).
One might speculate as to the reason
why glaucoma prevalence was so highly
dependent on refractive state. Some
authors have suggested that myopic eyes
are more susceptible to slightly raised
intraocular pressure (Anderson 1989;
Leighton & Tomlinson 1973; Perkins &
Phelps 1982). Our results could support
this hypothesis since we found that glau-
coma was clearly overrepresented in my-
opic individuals with low IOP, but here,
again, the logistic regression model
clearly demonstrates that myopia in it-
self is a significant risk factor for glau-
coma. Glaucoma damage in myopic in-
dividuals could also be totally unrelated
to IOP. There have been suggestions that
myopic individuals have abnormal con-
nective tissue that could predispose for
glaucoma (Curtin et al. 1979; Fong et al.
1990).
In congenital glaucoma the eye
stretches in response to raised IOP, re-
sulting in myopia (Vidic & Lerchner
1990). This mechanism seems an un-
likely explanation for the association be-
tween myopia and glaucoma in adults,
especially considering that our results
indicate that the relationship is strongest
at lower IOP levels.
Thus, analysis of this large popula-
tion-based study showed that the preva-
lence of glaucoma was strikingly de-
pendent on refraction and markedly
over-represented in myopic eyes. These
findings are in agreement with previous
studies and thus confirm myopia as an
important risk factor for glaucoma. A
new and interesting finding was that this
association depended strongly on IOP.
The relationship between glaucoma and
myopia was stronger in eyes with low
IOP, weaker in the group with inter-
mediate pressure readings, and disap-
peared in eyes with high pressures.
Acknowledgements
Jonny Olsson, Ph. D. and Boel Bengtsson,
Ph.D. provided statistical advice.
This study was supported by grants from the
Järnhardt foundation, Malmö University Hos-
pital and by the Alcon Research Institute,
Forth Worth, Texas.
References
Abdalla MI & Hamdi M (1970): Applanation
ocular tension in myopia and emmetropia.
Br J Ophthalmol 54: 122–125.
Anderson DR (1989): Glaucoma: The damage
caused by pressure. XLVI Edward Jackson
memorial lecture. Am J Ophthalmol 108:
485–495.
Bengtsson B (1981): Manifest glaucoma in the
aged: I Occurrence nine years after a popu-
lation survey. Acta Ophthalmol 59: 321–331.
Charliat G, Jolly D & Blanchard F (1994):
Genetic risk factor in primary open-angle
glaucoma: a case-control study. Ophthalmic
Epidemiol 1: 131–138.
Curtin BJ, Iwamoto T & Renaldo DP (1979):
Normal and staphylomatous sclera of high
myopia. An electron microscopic study.
Arch Ophthalmol 97: 912–915.
Daubs JG & Crick RP (1981): Effect of refrac-
tive error on the risk of ocular hypertension
and open angle glaucoma. Trans Ophthal-
mol Soc U K 101: 121–126.
David R, Zangwill L, Stone D & Yassur Y
(1987): Epidemiology of intraocular pressure
in a population screened for glaucoma. Br J
Ophthalmol 71: 766–771.
David R, Zangwill L, Tessler Z & Yassur Y
(1985): The correlation between intraocular
pressure and refractive status. Arch
Ophthalmol 103: 1812–1815.
Drance S, Sweeney V, Morgan R & Feldman
F (1973): Studies of factors involved in the
production of low tension glaucoma. Arch
Ophthalmol 89: 457–465.
Fong DS, Epstein DL & Allingham RR (1990):
Glaucoma and myopia: are they related? Int
Ophthalmol Clin 30: 215–218.
Greve EL & Furuno F (1980): Myopia and
glaucoma. Graefe’s Arch Clin Exp Ophthal-
mol 213: 33–41.
Gupta N & Weinreib R (1997): New definitions
of glaucoma. Curr Opin Ophthalmol 8: 38–
41.
Heijl A & Molder H (1993): Optic disc diam-
eter influences the ability to detect glau-
comatous disc damage. Acta Ophthalmol
71: 122–129.
Hollows FC & Graham PA (1966): Intraocular
pressure, glaucoma and glaucoma suspects
in a defined population. Br J Ophthalmol
50: 570–586.
Jonas J & Papastathopoulos K (1996): Optic
disc shape in glaucoma. Graefe’s Arch Clin
Exp Ophthalmol 234: 167–173.
Kitazawa Y, Horie T, Aoki S, Suzuki M & Ni-
shioka K (1977): Untreated ocular hyperten-
sion. Arch Ophthalmol 95: 1180–1184.
Leighton DA & Tomlinson A (1973): Ocular
tension and axial length of the eyeball in
open-angle glaucoma and low tension glau-
coma. Br J Ophthalmol 57: 499–502.
Leske C, Heijl A, Hyman L, Bengtsson B &
The Early Manifest Glaucoma Trial Group
(1999): Early Manifest Glaucoma Trial: de-
sign and baseline data. Ophthalmol 106 (11):
2144–2153.
Liang KY & Zeger SL (1986): Longitudinal
dataanalysis using generalized linear
models. Biometrics 73: 13–22.
Linnér E (1976): Ocular hypertension: I. The
clinical course during 10 years without ther-
apy. Aqueous humour dynamics. Acta
Ophthalmol 54: 707–720.
Lundberg L, Wettrell K & Linnér E (1987):
Ocular hypertension – A prospective 20-year
follow-up study. Acta Ophthalmol 65: 705–
708.
Mitchell P, Hourihan F, Sandbach J & Wang
JJ (1999): The relationship between glau-
coma and myopia. Ophthalmol 106 (10):
2011–2015.
Nicolela MT & Drance SM (1996): Various
glaucomatous optic nerve appearances.
Ophthalmol 103: 640–649.
Odberg T (1993): Visual field prognosis in
early glaucoma – a long term clinical follow-
up. Acta Ophthalmol 71: 721–726.
Perkins ES & Phelps CD (1982): Open angle
glaucoma, ocular hypertension, low-tension
glaucoma and refraction. Arch Ophthalmol
100: 1464–1467.
Perkins ES (1973): The Bedford glaucoma sur-
vey. I. Long-term follow-up of borderline
cases. Br J Ophthalmol 57: 179–185.
Schulzer M, Drance SM & Douglas GR
(1991): A comparison of treated and un-
treated glaucoma suspects. Ophthalmol 98:
301–307.
Shiose Y, Kitazawa Y, Tsukahara, Akamatsu
T, Mizokami K, Futa R, Katsushima H &
Kosaki H (1991): Epidemiology of glau-
coma in Japan – A nationwide study. Jpn J
Ophthalmol 35: 35–155.
Tomlinson A & Phillips CI (1970): Applan-
ation tension and axial length of the eyeball.
Br J Ophthalmol 54: 548–553.
Varma R, Tielsch JM, Quigley HA, Hilton SC,
Katz J, Spaeth GL & Sommer A (1994):
Race-, age-, gender-, and refractive error-re-
lated differences in the normal optic disc.
Arch Ophthalmol 112 (8): 1068–1076.
Vidic B & Lerchner H (1990): Long-term bio-
metric results in glaucoma in children.
Fortsch Ophthalmol 87: 25.
Wilson R, Hertzmark E, Walker A, Childs-
565
 Shaw K & Epstein D (1987): A case-control
study of risk factors in open angle glau-
coma. Arch Ophthalmol 105: 1066–1071.
Åsman P & Heijl A (1992a): Evaluation of
methods for automated hemifield analysis in
perimetry. Arch Ophthalmol 10: 820–826.
566
Åsman P & Heijl A (1992b): Glaucoma Hemi- Correspondence:
field Test; automated visual field evaluation. Kirsti Grødum, MD
Arch Ophthalmol 10: 812–819. Dept. of Ophthalmology
Malmö University Hospital
Received on April 5th, 2001. S-205 02 Malmö
Accepted on June 7th, 2001. Sweden
Tel: π46 40 33 31 03
Fax: π46 40 33 62 12
e-mail: kirsti.grodum/oftal.mas.lu.se