Association Between Ocular

Pressure and Certain

Health Parameters
RAFAEL S. CAREL, MD, MPH,· AMOS D. KORCZYN, MD, MSc,tt MEIR ROCK, MD,t
I. GOYA, MDt

Abstract: Intraocular pressure (lOP) was measured in a mixed population of

12,803 apparently healthy employed people. Mean lOP was 13.5 ± 3.3 mmHg,
without sex difference. Frequency distribution demonstrated skewness towards
high values. lOP weakly correlated with age (R = 0.06), and older subgroups
showed more marked skewness, but further analysis showed this effect to be
spurious. The correlations of lOP with heart rate and with systolic blood pressure
were small, but stronger than with age (R = .16 and .15, respectively). Moreover,
when corrected for heart rate, the effect of age was nullified. Other factors
found to be correlated with lOP included blood glucose and hemoglobin con-
centration, smoking, and height. None of these factors significantly increased
the correlation between lOP and heart rate or blood pressure, and the skewness
was not fully explained by any of these factors or their combinations. The value
of the epidemiologic approach to detection of factors responsible for ocular
hypertension is stressed. [age, heart rate, intraocular pressure, systolic blood
pressure.] Ophthalmology 91 :311-314, 1984.

Several factors have been found to affect intraocular
pressure (lOP), including venous and arterial blood pres-
sure, heart rate, diurnal or seasonal variations and cor-
ticosteroid levels in plasma. 1-4 The effects of age and gen-
der are unclear, and conflicting findings were re-
ported. I ,2,4-8 It has been reported that the lOP in various
populations is slightly skewed to the right. 2,6,7,9
This study analyzed a large, presumably healthy pop-
ulation in Israel, composed of a mixture of employed
individuals from many employment sources and of several
ethnic groups. It was aimed at determining some of the
factors associated with increased lOP. A non-contact to-
nometer (NCT) was used. The ease and the low cost of
lOP measurement with the NCT IO, permits mass screening
for ocular hypertension (OH), early identification of per-
sons with OH or glaucoma and early introduction of
From the MOR Institute for Medical Data, Department of Preventive and
Social Medicine,· the Department of Physiology and Pharmacology, Sackler
School of Medicine, Tel Aviv University,t and Maurice and Gabriela Gold-
schleger Eye Institute, Sheba Medical Center, Tel Hashomer, Israel.t
Presented at the Eighty-eighth Annual meeting of the American Academy
of Ophthalmology, Chicago, Illinois, October 3Q-November 3, 1983.
Reprint requests to Amos D. Korczyn, MD, Dept. of Physiology and
Pharmacology, Sackler School of Medicine, Tel Aviv University, Israel.
preventive-therapeutic measures when they are most
beneficial. The identification of high-risk groups through
studies like this will make the application of this screening
method more efficient.
MATERIALS AND METHODS
STUDY POPULATION
Between 1977 and 1980, approximately 13,000 em-
ployed subjects, aged 40-75 years, underwent compre-
hensive health examinations in a multiphasic screening
center. These examinations included many physical,
physiological, biochemical and anamnestic parameters.
Most of the examinees were referred for routine testing
or for periodic health examinations from large industries
and corporations. The screening tests and examinees were
previously described in detail. II ,I2
The age-sex distribution of the study population is
shown in Table 1. The relative age distribution within
each sex group is quite similar in both sexes. This relative
age distribution (except for the somewhat low presentation
in the study sample of people over 65 years of age, and
excessive representation of males aged 40-44) was pre-
viously analyzed and compared to that of the general

311
 OPHTHALMOLOGY • APRIL 1984 • VOLUME 91 • NUMBER 4

Table 1. Age-sex Composition of the Study Group found (Carel, Rock, and Korczyn, in preparation) and
and the Mean lOP (±SO) the mean lOP commonly serves as a valid representative
value. 1,2,4
Age Group
(Years) Males Females Total

40-44 1888 (21.1)" 451 (11.8) 2339 (18.3) RESULTS

13.3 ± 3.1 13.0 ± 2.9 13.1 ± 3.1
45-49 1524 (17.0) 443 (11.5) 1967 (15.4)
13.4 ± 3.2 13.7 ± 3.0 13.6 ± 3.1 The mean lOP (±SD) in the various age groups for
50-54 2018 (22.5) 1064 (27.7) 3082 (24.1) each sex separately and for the whole population is shown
13.6 ± 3.2 13.5 ± 3.1 13.5 ± 3.2 in Table 1. In all age groups the mean lOP did not differ
55-59 1733 (19.3) 886 (23.1) 2619 (20.5) significantly between the sexes. Thus, further analyses
13.7 ± 3.4 13.8 ± 3.4 13.7 ± 3.4 were made on pooled data for both sexes. There was a
60-64 1064 (11.9) 635 (16.6) 1699 (13.3) slight tendency of mean lOP to increase with age. This
13.5 ± 3.4 13.8 ± 3.3 13.6 ± 3.4 trend was statistically significant but of limited clinical
65-69 528 (5.9) 282 (7.4) 810 (6.3) importance, since it was small compared to the size of
13.9 ± 3.7 14.0 ± 3.4 14.0 ± 3.5
70-75 210 (2.3)
the standard deviation.
77 (2.6) 287 (2.2)
14.1 ± 3.8 13.3 ± 3.2 13.7 ± 3.5 Figure 1 shows the relative frequency distribution of
lOP in males and females. The similarity between the
Total 8965 (100) 3838 (100) 12803 (100) two curves is obvious. The curves are unimodal and
13.5 ± 3.3 13.6 ± 3.2 13.5 ± 3.3 skewed to the right, indicating more people with relatively
Mean age 52.4 ± 8.1 54.5 ± 7.4 53.0 ± 8.0 high lOP values, with respect to the mean, than low values.
SO = standard deviation. Simple correlation coefficient values between lOP and
" Number in parentheses indicates percentage of the same sex group. various health parameters were calculated (see Table 2).
Even though all R values were small (less than 0.20), the
corresponding P values were highly significant (P < .001),
populationY We were not primarily concerned in this reflecting the size of the study group. The highest cor-
study in prevalences of various variables in the general relation coefficients were found with heart rate (HR),
population but rather in evaluation of the effects of dif- systolic blood pressure (SBP), diastolic blood pressure
ferent factors on the distribution of lOP in presumably and serum glucose level. Age was found to be only weakly
healthy people. correlated with lOP. Smoking habits, serum cholesterol
and uric acid also have positive, but very weak, correlation
with lOP (not shown in the table). Height is negatively
EYE EXAMINATION
correlated with lOP (R = -.07). Other parameters such
Each examinee went through the same battery of tests. as weight (obesity index; weightjheight2), sex or ethnicity
lOP was tested with an AO Non-Contact Tonometer. No were not significantly correlated with lOP. When con-
medication was used and there was no fixed order for trolling for age the R values of the various parameters
examining the eyes. Each eye was tested twice, and if the were unaltered, indicating that their effect was not exerted
difference between consecutive measurements was greater through age to a significant degree. As these correlations
than 2 mmHg, the procedure was repeated. The value were low, very little variance could be partitioned to age
reported was the highest of the two (differing by 2 mmHg in any case. Stated differently, because of the low cor-
or less). This criterion was used in order to minimize relation of lOP with age, it could be expected that age
false negative rate. Unless otherwise specified the values will not contribute much to the R values with those other
of lOP reported are the means of both eyes. This was parameters which are known to be correlated with age
done as only a small difference between the 2 eyes was (like SPB, blood sugar or height).
Analysis of variance (ANOYA) with main effects of
sex, ethnic origin, and smoking habits, and covariates
Table 2. Simple Correlation Coefficients (R) Between lOP and Various age, SBP, HR, and height, has demonstrated that lOP is
Relevant Health Parameters primarily related to HR (F = 116.6), SBP (F = 103.3),
height (F = 21.7), and smoking more than one pack of
Variable Males Females Total cigarettes per day (F = 19.7). All these F values were
significant at the. P < 0.001 level. No significant rela-
HR (beats/min) 0.14 0.19 0.16 tionship was found with age, sex or ethnic origin.
SBP (mmHg) 0.14 0.16 0.15 Table 3 demonstrates that by performing stepwise
OBP (mmHg) 0.09 0.12 0.10 multiple regression only a small increase in multiple R
Glucose (mg%) 0.07 0.13 0.09
Age (years) 0.05 0.07 0.06 or R2 was obtained over simple linear correlation with
Hemoglobin (gr%) 0.06 0.11 0.05 HR. By utilizing three variables (HR, SBP and height)
the value ofR increased from 0.16 to 0.20. R2 increased
HR = heart rate; SBP = systolic blood pressure; OBP = diastolic from 0.03 to 0.04, indicating almost no improvement in
blood pressure. the "explained" variance. Additional terms did not add

312
 CAREL, et al • FACTORS AFFECTING lOP

significantly to the regression equation or multiple cor- Table 3. Multiple Regression of lOP on the Various Relevant
relation coefficient. Age was the fourth contributing factor Parameters for the Whole Population
but it did not change the multiple R value at all (R = 0.20
and P = 0.22). Multiple

Step No. Variable R R2 P*

DISCUSSION 1 HR 0.16 0.03 0.001

In recent years, screening for OH and glaucoma has
been more widely performed, leading to early detection
and possible prevention of the damage afflicted by long
standing disease process. Glaucoma is currently one of
the major causes of blindness in adults. Because the ben-
eficial effect of early treatment was demonstrated, screen-
ing and surveillance of populations at risk is indicated.
To perform such a task efficiently and economically it is
imperative to better understand the distribution of the
concerned parameter and other relevant health deter-
minants in the general population. It is also necessary to
establish "normal" ranges as well as criteria for identifYing
diseased individuals or high risk groups. Such epidemi-
ologic investigations can also assist in planning of services
and programs and even in consideration of etiologic fac-
tors in the development of OH. The available medical
technology (NCT) makes the testing of lOP a simple,
rapid and economical process. No previous normal ranges
with this instrument were reported, even though in quite
a few studies a good concordance was demonstrated be-
tween readings of the NCT and other tonometers.
In several articles a trend for increasing average lOP
with age is emphasized. 1,2.4-6 However, none of these pre-
vious works further analyzed this apparent trend. In our
2 SBP 0.19 0.04 0.001
3 Height 0.20 0.04 0.001
4 Age 0.20 0.04 0.220
HR = heart rate; SBP = systolic blood pressure.
* P = significance of the marginal contribution to regression of the
concerned variable.
work, a similar apparent trend is also seen (Table 1) and
only further statistical analyses show it to be a spurious
effect. In each age group the individuals with higher lOP
values are more likely to be those with higher HR and
SBP. Thus, when first taking into account HR and SBP,
the age effect on lOP was nullified (Table 3). Therefore,
the apparent effect of age on lOP merely reflects the greater
frequency of people with higher SBP and HR in older
age groups. The tendency of SBP to increase with age in
western population is well documented.
Low order correlations ofIOP with SBP were also found
by previous workers. 1,3,4,13,14 It is of interest to indicate
that Klein and Klein4 also found that SBP was the variable
best correlated with lOP. In our study, a stepwise multiple
regression has shown that age contributes only minimally
to the regression after SBP was taken into account in the
first step (R increased from .17 to .19 and R2 from .03
to .04).

25 MALES
FEMALES ---------

20

15 ,,
,
,'/
10 ,,
,I'
5 / "
,I','

,','

2 8 10 12 14

IO P (mmHg)
Fig 1. Distribution of lOP values of the 12.803 examinees, expressed as percentages for males and females. Note the skewness towards higher
values.

313
 OPHTHALMOLOGY • APRIL 1984 • VOLUME 91 • NUMBER 4

Several factors could account for the relationship be-
tween lOP and HR or SBP. SBP could increase the fil-
3.
tration pressure in the eye and thus lOP, but this is an
unlikely explanation since the correlation of lOP was
stronger with HR than with SBP. A common mechanism
could account for the increase of all three, such as sym- 4.
pathetic tone, serum corticosteroids or sclerotic changes
occurring both in arteries and in the outflow channels 5.
from the eye. Whatever the mechanism, it should be
stressed that its contribution-though highly significant
statistically-is quantitatively small. Therefore other 6.
factors (ie, genetic) must be responsible for the skewness
to the right of the lOP value distribution observed in
7.
Figure 1.
The effect on lOP of parameters other than age and 8.
BP was investigated earlier. HR was previously found to
be related to IOP. 3 ,4 The negative correlation ofIOP with 9.
height was demonstrated by Kahn and associates. 3 Positive
(low) correlation of lOP with smoking habits was shown 10.
by Morgan and Drance (13). Others have shown lOP to
be correlated with blood sugar or diabetes. 3,15 The rela-
tionships of these and other predisposing factors to disease 11.
occurrence or development, is reminiscent of the known
web of causation in cardiovascular diseases. 16 This concept
12.
could be beneficial in investigating epidemiologic, etio-
logic and intervention problems related to OH and glau-
coma. 13.
14.
REFERENCES
15.
1. Bengtsson B. Some factors affecting the distribution of intraocular
pressures in a population. Acta Ophthalmol 1972; 50:33-46. 16.
2. Hollows FC, Graham PA. Intra-ocular pressure, glaucoma, and glau-
coma suspects in a defined population. Br J OphthalmoI1966; 50:570-
86.
Kahn HA, Leibowitz HM, Ganley JP, et al. The Framingham Eye
Study. II. Association of ophthalmic pathology with single variables
previously measured in the Framingham Heart Study. Am J Epidemiol
1977; 106:33-41.
Klein BE, Klein R. Intraocular pressure and cardiovascular risk variables.
Arch Ophthalmol1981; 99:837-9.
Kahn HA, Leibowitz HM, Ganley JP, et al. The Framingham Eye
Study. I. Outline and major prevalence findings. Am J EpidemioI1977;
106:17-32.
Armaly MF. On the distribution of applanation pressure. I. Statistical
features and the effect of age, sex, and family history of glaucoma.
Arch Ophthalmol 1965; 73: 11-8.
Leydhecker W. The intraocular pressure: clinical aspects. Ann
Ophthalmol 1976; 8:389-99.
Armaly MF. Age and sex correction of applanation pressure. Arch
Ophthalmol 1967; 78:480-4.
Bankes JLK, Perkins ES, Tsolakis S, Wright JE. Bedford glaucoma
survey. Br Med J 1968; 1:791-6.
Forbes M, Pico G Jr, Grolman B. A noncontact applanation tonometer;
description and clinical evaluation. Arch Ophthalmol 1974; 91: 134-
40.
Carel RS, Leshem G. Evaluation of the cost-effectiveness of an au-
tomated multiphasic health testing system. Prev Med 1980; 9:689-
97.
Tartakovsky MB, Carel RS, Luz Y. A comparison of the body height
of the Israeli-born and immigrants to Israel. Hum Hered 1983;
33:73-8.
Morgan RW, Drance SM. Chronic open-angle glaucoma and ocular
hypertension; an epidemiological study. Br J Ophthalmol 1975;
59:211-5.
Kaskel 0, Baumgart W, Metzler U, Fink H. Blood pressure, blood
flow and intraocular pressure. Ophthalmic Res 1974; 6:338-45.
Becker B. Diabetes mellitus and primary open-angle glaucoma. Am
J Ophthalmol1971; 71:1-16.
Friedman GO. Primer of Epidemiology, 2nd ed. New York: McGraw-
Hill, 1980; 3-4.

Discussion
by
R. Stamper, MD

The authors have presented an epidemiologic study of intra-
ocular pressure in a "healthy", working population aged 40-
75 years. They correlated the intraocular pressures with age,
sex, ethnicity, heart rate, blood pressure, serum glucose, serum
cholesterol, serum uric acid, smoking habits, height, and weight.
They found that intraocular pressure correlated only weakly
with age. In fact, when they used multivariate analysis to factor
out the more significant correlations, the weak correlation with
age could be explained entirely on the basis of the independent
association of those factors (i.e. systolic blood pressure and heart
rate) with age. The positive correlations with systolic blood pres-
sure and heart rate, as well as the negative correlation with
height, were small but statistically significant because of the
large numbers in the study (12,800). Some other small corre-
lations were also demonstrated, such as smoking, blood glucose
and blood hemoglobin. The clinical significance of these findings
is questionable.
While the total numbers are large enough to make a small
correlation statistically significant, the relatively small numbers
in the higher risk groups (i.e., >65) may have contributed to
their conclusion that lOP does not increase with age. Presumably
excluded were those at the higher end of blood pressure readings
and blood glucose levels, as these persons may have less like-
lihood of working. This may also have biased their results.

314