Intraocular pressure, Goldmann applanation tension, corneal thickness, and corneal curvature in Caucasians, Asians, Hispanics, and African Americans

Intraocular Pressure, Goldmann ApplanationTension, Corneal Thickness, and Corneal

Curvature in Caucasians, Asians, Hispanics,and African Americans

MITSUGU SHIMMYO, MD, ANNA J. ROSS, OD, ANNA MOY, OD, ANDRAMIN MOSTAFAVI, MD

● PURPOSE: This is to investigate whether there aredifferences in Goldmann applanation tonometry (GAT),central corneal thickness, and corneal curvature amongfour racial groups. If differences are present, they mayalter GAT reading, diagnosis, and management of glau-coma in the population.● DESIGN: Observational retrospective cross-sectionalstudy.● METHODS: Charts of patients who have had keratore-fractive surgery were examined. Central corneal thick-ness, corneal curvature, refractive power, and GAT weremeasured in 1,482 Caucasian, 172 Asian, 204 Hispanic,and 118 African-American eyes (total 1,976 eyes).Refractive components and GAT were compared. Wecompared intraocular pressure (IOP) adjusted by GAT,central corneal thickness, and corneal curvature amongthe four groups.● RESULTS: There was a statistically significant differ-ence between the mean (� standard deviation) centralcorneal thickness of African American (535.46 �33.39) and Caucasian (552.59 � 34.48) eyes. Meancentral corneal thickness was near 550 �m in Cauca-sians, Asians, and Hispanics. No significant differencewas noted in corneal curvature in the four groups. Therewas a significant correlation between central cornealthickness and corneal curvature, and GAT was similaramong the four groups. When IOP was adjusted forcentral corneal thickness, it was significantly greater inAfrican Americans (16.12 � 3.27) than in Caucasians

(14.32 � 2.93). Corneas of women were significantlythinner than corneas of men.● CONCLUSIONS: African Americans had significantlythinner central corneal thickness than Caucasians,Asians, or Hispanics, causing the underreading of trueIOP. Significant correlation between central cornealthickness and corneal curvature was demonstrated. Un-corrected GAT underreading of African Americans maylead to delay in diagnosis, inadequate treatment targetsetting, and higher morbidity. Goldmann applanationtonometry needs to be corrected by central cornealthickness and corneal curvature for proper diagnosis andmanagement of glaucoma. (Am J Ophthalmol 2003;136:603–613. © 2003 by Elsevier Inc. All rightsreserved.)

T HE CLINICAL COURSE OF GLAUCOMA IN AFRICAN

Americans is more serious than in Caucasians.Recent studies note thinner corneas in African

Americans, yielding lower Goldmann applanation tonom-etry (GAT) readings.1,2 Other studies have found thickercorneas among ocular hypertensive eyes and thinnercorneas among normotensive eyes and low-tension glau-coma eyes.3–9 We still have questions as to what theaverage thickness of the human cornea is and how tocorrect the GAT reading to more accurate intraocularpressure (IOP).4,10–12 Besides differences in central cornealthickness by race, are there differences by sex or age?13–15

Are there correlations between central corneal thicknessand corneal curvature,11 central corneal thickness, andIOP,11,12 or corneal curvature and IOP?16

Goldmann’s theory is based on the assumption that thecorneal stromal thickness is a constant based on examina-tions of cadaver eyes.10 It is now widely known that centralcorneal thickness is a variable and is the major source ofsystematic artifacts in GAT.

Accepted for publication April 7, 2003.InternetAdvance publication at ajo.com April 22, 2003.From the Department of Clinical Ophthalmology (M.S.), New York

Medical College, and the Department of Ophthalmology, ManhattanEye, Ear and Throat Hospital; Laser One, Inc (A.J.R., A.M.), New York,New York; and the Department of Opthalmology (M.S., R.M.), NewYork Eye and Ear Infirmary, New York, New York.

Inquiries to Mitsugu Shimmyo, MD, 345 East 37th St, Suite 203, NewYork, NY 10016; fax: (212) 867-5731; e-mail: [email protected]

© 2003 BY ELSEVIER INC. ALL RIGHTS RESERVED.0002-9394/03/$30.00 603doi:10.1016/S0002-9394(03)00424-0

Ehlers11 in 1975 presented experiments in cannulated invivo human eyes and offered a corrective table to find IOPfrom GAT and central corneal thickness. Recently, inter-est in the significance of central corneal thickness has beenrekindled by refractive surgeons, who noted lower GATafter corneal thinning.17,18 Racial variations in centralcorneal thickness distribution were also found.1

DESIGN

THIS WAS A RETROSPECTIVE CROSS-SECTIONAL OBSERVA-

tional study.

METHODS

CHARTS OF 2,479 EYES OF CONSECUTIVE PATIENTS UNDER-

going presurgical examinations for refractive surgery with-

out ocular pathology between July 1999 and June 2002 ata multisurgeon refractive surgery center (Laser One, Inc.)in New York City were reviewed retrospectively. Pertinentinformation was transcribed into the computer database.

A total of 1,976 eyes were studied, of which 1,482Caucasian (74.86%), 172 Asian (8.68%), 118 African-American (5.96%), and 204 Hispanic eyes (10.45%) hadcentral corneal thickness recorded. Demographic informa-tion, including ethnic groups, refractive parameters, cen-tral corneal thickness, and GAT were extracted fromrecords and population distribution analyses of age, sex,ethnicity, laterality, spherical power, astigmatic power,central corneal thickness, corneal curvature, and GATwere compared. Statistical analyses of variables, IOP cor-rected by GAT and central corneal thickness, and IOPcorrected by GAT, central corneal thickness, and cornealcurvature were performed. Sex and ethnicity were deter-mined by self-designation.

Spherical and astigmatic powers were determined bymanifest refraction after cyclopegia with cyclopentolate1%, by two optometrists. Central corneal thickness wasmeasured using a DGH-550 Pachette ultrasonic pachy-meter (DGH Technology Inc., Exton, Pennsylvania,USA). Corneal curvature was measured using a Canon 2autokeratorefractor and keratometric power; it then wasrecorded in dioptors. Goldmann applanation tonometrywas performed using a Haag-Streit tonometer with slitlamp. Intraocular pressure correction formulas (describedin Appendix 2) were used to correct IOP from GAT andcentral corneal thickness; then IOP was corrected againfrom GAT, central corneal thickness, and corneal curva-ture. Differences between means were tested for statisticalsignificance using an unpaired t test and one-way analysisof variance (ANOVA).

RESULTS

THE MEAN AGE OF THE POPULATION WAS 37.36 � 9.71

years. The breakdown of age by ethnic groups is shown inTable 1.

TABLE 1. Mean Age: Total Population and by Ethnic Groups

Total Population African American Asian Caucasian Hispanic

Number 1,976 118 172 1,482 204

Mean 37.36 37.20 34.84 38.08 34.21

Standard deviation 9.71 9.78 7.29 9.86 9.38

Standard error 0.22 0.90 0.56 0.26 0.66

Minimum 19.00 21.00 22.00 19.00 19.00

Maximum 101.00 69.00 56.00 101.00 66.00

Range 82.00 48.00 34.00 82.00 47.00

Median 36.00 36.00 34.00 37.00 32.00

Mode 29.00 37.00 29.00 32.00 28.00

FIGURE 1. Regression and correlation between Goldmannapplanation tonometry (GAT) and age. Age and GAT are notcorrelated. Correlation coefficient: 0.06; P � .063; Y �14.307 � 0.016* X; R^2 � 0.004.

AMERICAN JOURNAL OF OPHTHALMOLOGY604 OCTOBER 2003

Age and GAT do not appear to be related in this samplepopulation (Figure 1). Regression between age (X) andGAT (Y) yielded: Y � 14.307 � 0.016* X; Rˆ2 � 0.004.Correlations coefficient: 0.06; P � 0.063. This may be areflection of the relative youth of the population studied.

Mean GAT by total population and the four ethnicgroups (Table 2) were not significantly different. Meanspherical power of the manifest refraction (Table 3) in thetotal population and four groups showed near normal

distribution. Mean central corneal thickness by ultrasonicpachymetry in the total population and by the four groups(Table 4) showed a mean of 551.16 � 34.55 �m in thetotal population and 535.46 � 33.39 in African Ameri-cans. There were significant differences between AfricanAmericans and Caucasians (P � .0001). The differencesbetween African Americans and each of the other ethnicgroups were still significant when analyzed with ANOVAand adjusted for multiple comparisons with the Bonfer-

TABLE 2. Mean GAT: Total Population and by Ethnic Groups


Number 958 59 78 714 107

Mean 14.89 14.9 14.74 14.86 15.23


Standard error 0.08 0.34 0.27 0.09 0.22

Minimum 8 8 9 8 10.0

Maximum 25 24 21 23 25.0

Range 17 16 12 15 15.0

Median 15 15 15 15 15.0

Mode 14 16 16 14 14.0

GAT � Goldmann applanation tonometry.

TABLE 3. Mean Sphere Power of Manifest Refraction: Total Population and byEthnic Groups


Number 1,966 118 172 1,470 206

Mean �3.18 �3.58 �4.82 �2.97 �3.13


Standard error 0.06 0.25 0.21 0.07 0.19

Minimum �13 �9.5 �13 �12.5 �11

Maximum 7 6 6.5 7 5

Range 20 15.5 19.5 19.5 16

Median �3 �3.5 �4.75 �3 �3

Mode �2.25 �3.5 — �2.25

TABLE 4. Mean Central Corneal Thickness: Total Population and by Ethnic Groups


Number 1,955 116 170 1466 203

Mean 551.16 535.46 549.79 552.59 551.10


Standard error 0.781 3.1 2.48 0.9 2.49

Minimum 439.00 465 468 439 462

Maximum 690.00 624 646 690 644

Range 251.00 159 178 251 182

Median 551.50 530.5 551 553 554

Mode 541.00 527 574 — 555

EVALUATING RACIAL DIFFERENCES IN GLAUCOMA MEASUREMENTSVOL. 136, NO. 4 605

roni-Dunn procedure (adjusted P � .001; Table 5). Thedifferences among the non–African Americans were insig-nificantly small (Table 5).

Central corneal thickness vs GAT: significant correla-tion exists between central corneal thickness (X)and GAT (Y), regression formula: Y � 5.825 � 0.016* X;Rˆ2 � 0.057. Correlation coefficient: 0.238; P � .001, asreported by many authors (Figure 2).

Central corneal thickness vs age: no correlation existsbetween central corneal thickness (Y) and age (X) in thispopulation: Y � 545.395 � 0.154* X; Rˆ2 � 0.002.Correlation coefficient: 0.238; P � .001. This may reflectthe relative youth of the population.

Central corneal thickness vs corneal curvature (K1):significant correlation exists between keratometric powerK1 (Y) and central corneal thickness (X). Regressionformula: Y � 46.995 � 0.006 X. P � .001. Thicker eyes areflatter, and thinner eyes are steeper (Figure 3).

Central corneal thickness vs corneal curvature (K2):

significant correlation exists between keratometric powerK2 (Y) and central corneal thickness (X), as in centralcorneal thickness vs K1. Regression formula: Y � 48.266� 0.007 X. P � .001.

Central corneal thickness, right eye (OD) vs left eye(OS): There was no significant difference in centralcorneal thickness between OD (551.16 � 34.55) and OS(552.55 � 35.84). P � .289.

Central corneal thickness, male vs female: male corneas(553.98 � 34.37) were found to be significantly thickerthan female corneas (547.72 � 34.48). P � 0.001 (Table6).

Goldmann applanation tonometry, male vs female: nosignificant difference was found in GAT between male(14.97 � 2.57) and female (14.8 � 2.36). P � 0.2801.

There was no significant difference in primary cornealcurvature (K1) distribution by ethnicity (Table 7).

Central corneal thickness vs K1 in African Americans:Y (central corneal thickness) � 663.018 � 3.016* X (K1);Rˆ2 � 0.027. Correlation coefficient: �0.164; P � .157.Although a similar trend seen in the total populationexists, statistical significance was lost in this segment,which represents the relative flatness of their corneasdespite the thinness.

Intraocular pressure adjusted by central corneal thick-ness: total population (Figure 4) and by four ethnic groups(Table 8): the adjusted IOP of African Americans (15.94� 3.16) was significantly higher than that of Caucasians(14.59 � 3.02) and other races. The difference betweenCaucasians and African Americans was still significantwhen analyzed with ANOVA and adjusted for multiplecomparisons using the Bonferroni-Dunn procedure (P �.001). This was an unexpected finding.

Intraocular pressure adjusted by central corneal thick-ness and corneal curvature, total population, and by fourethnic groups (Table 9): Adjusted IOP of African Amer-icans (16.12 � 3.32) was significantly higher than that ofCaucasians (14.32 � 3.04; P � .001) and other races.

DISCUSSION

A LARGE-SCALE POPULATION STUDY OF CENTRAL COR-

neal thickness, corneal curvature, and GAT in differentethnic groups is difficult. As a substitute for such a study,we chose to review charts of patients who underwentpresurgical examinations at a refractive surgery center.That is not an ideal unbiased population in terms ofcomparing racial variations, but our primary intention wasto collect a sufficiently large database for statistical com-parison of other parameters. It was our concern that therepresentative numbers of subjects were not evenly distrib-uted racially, as the majority of the subjects who came tothe center was Caucasians. Rather than selectively reduc-ing the number of Caucasian subjects to match the numberof other groups and reduce the total number of observa-

TABLE 5. P Values (Unpaired t Tests) of Mean CentralCorneal Thickness Differences by Ethnic Groups

African

American Asian Caucasian Hispanic

African American .0003 �.0001 .0001

Asian .0003 .315 .712

Caucasian �.0001 .315 .567

FIGURE 2. Regression and correlation between central cor-neal thickness and Goldmann applanation tonometry (GAT).There is a significant correlation between central cornealthickness and GAT. Correlation coefficient: 0.238; P < .001;Y � 5.825 � 0.016* X; R^2 � 0.057.


tions, we chose to include them all for the sake ofcollecting a sufficient number for greater statistical signif-icance in seeking correlations between parameters such ascentral corneal thickness vs corneal curvature, as this wasone of our main interests. Studying only one eye for eachsubject may reduce bias by race in population analysis, butto collect a sufficient number of observations, we includedthe data of both eyes of each subject as two different eyes,as we considered the combinations of parameters in eacheye to be unique.

The mean age of the population (37.36 � 9.71 years) inthis study is relatively younger than in other populationanalysis, such as the Reykjavik Eye Study13 in which 925eyes of Caucasians of older subjects aged 50 to 85 yearswere studied. They also used different instrumentation, aNidek NT 2000 noncontact air-puff tonometer and Nidek

EAS 1000 automated eye analysis system, to measurecentral corneal thickness and corneal curvature based onScheimpflug slit-lamp images. The mean GAT in ourpatients was 14.89 mm Hg vs 15.5 mm Hg in the Reykjavikstudy. No correlation was found between GAT and age(Figure 1) in our study, in agreement with the Reykjavikstudy. Other studies in Asian populations report higherIOP with age.14

A significant correlation was seen between centralcorneal thickness and GAT (Figure 2) in all four ethnicgroups and in both sexes, as shown in many reports.Significant correlation between central corneal thicknessand corneal curvature (Figure 3) found in other animals11

was also found among the subjects in this study.Although Foster14 found a decrease in central corneal

thickness with increasing age in Mongolians using opticalpachymetry, we did not see a correlation between centralcorneal thickness and age in our population, agreeing withthe Reykjavik13 and Rotterdam15 studies. Observation inthe Mongolian population may reflect genetic or environ-mental factors. We found a significant difference in centralcorneal thickness between African Americans and theother groups (Table 4) as reported by La Rosa.1 We alsofound corneas of women to be significantly thinner thanthose of men (Table 6). In contrast to the Reykjavikstudy,13 we did not find GAT to be higher in females.

Many published studies suggest correction factors basedon central corneal thickness and GAT,11,12,19 however,corneal curvature can also influence GAT readings.10,16,20

Orssengo and associates21 recently discussed errors in GATby theoretical mathematical models. They discussed the

FIGURE 3. Regression and correlation between central corneal thickness and primary corneal curvature (K1). There is a significantcorrelation between central corneal thickness and K1. Correlation coefficient: �0.147; P < .001; Y � 46.955 � 0.006* X; R^2

� 0.022.

TABLE 6. Mean Central Corneal Thickness: Male VersusFemale

Female Male P Value

Number 880 1076

Mean 547.72 553.98 �.001

Standard deviation 34.48 34.37

Standard error 1.16 1.05

Minimum 450.00 439.00

Maximum 690.00 665.00

Range 240.00 226.00

Median 545.00 555.00


deformation of central cornea flattened by pressure of theprism and bulging outward from the middle to peripheralcornea by the inner pressure of the eye. They concludedthat central corneal thickness, corneal curvature, and areaapplanated should be considered in adjusting errors ofGAT and alluded to the role of variations in Poisson rateof the corneal tissue of different individuals. They consid-ered the results of Ehlers to be the most accurate. Asians,particularly Japanese, are known to have a disproportion-ately high incidence of normotensive glaucoma. As theGoldmann applanation tonometer was calibrated withCaucasian eyes, calibration with Asian eyes and studies ofdifferences in physical properties of ocular protein, namelyPoisson rate, would be necessary in elucidating the enigma.

In 1978, Johnson and associates8 reported one patientwith apparently benign ocular hypertension with a cornealthickness of 0.90 mm and IOP by GAT in the range of 30

to 40 mm Hg. On cannulated manometry, the patient wasfound to have an actual IOP of 11.0 mm Hg with asimultaneous Perkins applanation reading of 40.0 mm Hg.This must have been a surprise at the time. However, if weuse the correction formula of P � A � (550 � T) / 11.8,we get 40 � (550 � 900) / 11.8 � 40 � 29 � 11, ascalculated by correction formula 3 (see Appendix 2).

To illustrate the impact of central corneal thickness onGAT, GAT values were calculated from the formula attrue IOP levels of 5 to 40 mm Hg at central cornealthickness levels between 400 and 700 �m (Table 10).

When GAT (uncorrected by central corneal thickness)is considered as one of the diagnostic criteria, many eyeswith thicker corneas may be erroneously considered tohave glaucoma or ocular hypertension3–9 and may receiveunnecessary therapy. For example, an eye with centralcorneal thickness of 700 �m with GAT of 30.7 mm Hg has

TABLE 7. Keratometric Power by Total Population and Ethnicity


Number 1,390 76 114 1,036 164

Mean 43.49 43.34 43.11 43.56 43.35


Standard error 0.04 0.19 0.13 0.05 0.11

Minimum 37.50 37.50 39.75 38.50 39.25

Maximum 50.75 48.25 47.00 50.75 47.00

Range 13.25 10.75 7.25 12.25 7.75

Median 43.50 43.25 43.13 43.50 43.50

Mode 43.50 43.25 43.50 43.50

FIGURE 4. Histogram: intraocular pressure (IOP) adjusted by central corneal thickness (CCT) in total population.


a true IOP of only 20 mm Hg and may not need treatmentif the optic nerve is intact.

Conversely, eyes with thinner corneas with truly highIOP may escape detection. An eye with central cornealthickness of 400 �m with GAT of 13.7 mm Hg has a trueIOP of 25 mm Hg and will most likely need treatmentwhen optic nerve damages are recognized.

The common practice of relying on unadjusted GATresults in misdiagnosis and mismanagement. In apparentlynormotensive glaucoma, more aggressive lowering of IOPseems to be indicated when the corrected IOP is known tobe high.

To discuss the significance of distribution of centralcorneal thickness in the general population, we observed

TABLE 8. Intraocular Pressure Adjusted by Central Corneal Thickness: Total Population andby Ethnic Groups


Number 947 59 76 708 104

Mean 14.74 15.94 14.75 14.59 15.16


Standard error 0.1 0.41 0.36 0.11 0.26

Minimum 4.38 10.07 7.06 4.38 8.81

Maximum 27.00 26.03 21.53 27.00 23.50

Range 22.63 15.96 14.47 22.63 14.68

Median 14.73 15.80 14.63 14.60 15.39

Mode 15.00 16.80 14.40 16.67

TABLE 9. Intraocular Pressure Adjusted by Central Corneal Thickness and Keratometry:Total Population and by Ethnic Groups


Number 693 44 52 514 83

Mean 14.50 16.12 14.16 14.32 14.95


Standard error 0.11 0.49 0.39 0.13 0.30

Minimum 5.93 9.92 7.19 5.93 8.76

Maximum 26.05 26.05 21.20 24.11 23.27

Range 20.12 16.12 14.01 18.18 14.51

Median 14.44 15.85 14.26 14.33 14.95

Mode 16.41

TABLE 10. GAT Calculated by CCT and IOP

CCT Calculated GAT

700 13.8 19.4 25.0 30.7 36.3 41.9 47.7 53.4

650 10.9 16.3 22.4 27.1 32.5 37.9 43.5 48.9

600 7.9 13.1 18.3 23.6 28.8 34.0 39.2 44.5

550 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0

500 2.1 6.9 11.7 16.4 21.2 26.0 30.8 35.5

450 0.9 3.8 8.3 12.9 17.5 22.1 26.5 31.1

400 0.0 0.6 5.0 9.3 13.7 18.1 22.3 26.6

IOP 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0

CCT � central corneal thickness; GAT � Goldmann applanation tonometry; IOP � intraocular

pressure.


mean central corneal thickness to be 551.16 � 34.55;following normal distribution patterns, 68.26% of the eyeswill fall within the first standard deviation, and 31.74% ofthem will be outside and have GAT errors greater than 2.3mm Hg; 95.45% of the eyes will fall within the secondstandard deviation, and 4.85% of them will be outside andwill have GAT errors greater than 4.6 mm Hg; only 0.26%will be outside of the third standard deviation and willhave GAT errors greater than 6.91 mm Hg. Half of 0.26%,or 0.13%, of the eyes will have central corneal thicknessless than 413 �m. It is still very important to know the trueIOP for detection and treatment of glaucoma. Therefore,pachymetry seems to be a necessary examination. The costof the instrument and the time required for pachymetry areless than 10% of those for perimetry.

Formulas for adjusting GAT by central corneal thick-ness and corneal curvature are discussed in Appendix 2.The correction formulas presented here are statisticalapproximations. Without direct and invasive manometry,we still do not know the true IOP. Currently availablemanometry is based on certain assumptions. Averagecentral corneal thickness differs by different populationgroups. Optical pachymetry and ultrasonic pachymetryhave differences in measurement results. Ehlers used opti-cal pachymetry and found that the GAT was most accuratewhen central corneal thickness was 520 �m. We found themean central corneal thickness to be 551.25. To minimizethe impact of adjusting GAT in the majority of thepopulation, statistical approximation formulas are pre-sented, taking 550 as a statistical norm (average). As theerror in measurement increases nonlinearly, correctionformulas with exponential function (formulas 4, 5, and 6)are introduced.

Approximately 50% to 60% of eyes considered to haveocular hypertension3–9 may have normal IOP. Approxi-mately 30% to 40% of eyes considered to have normoten-sive glaucoma may have high IOP3–7 after correction ofGAT by central corneal thickness. When the Schiøtztonometer was in use, Friedenwald discussed the signifi-

cance of scleral or corneal rigidity. In Goldmann applana-tion tonometry, as the effect of deformation of cornea issignificantly smaller than in Schiøtz tonometry, discussionof the impact of tissue rigidity or elasticity has dissipated.Goldmann discussed the deformation of the cornea andshifting of fluid in anterior chamber extensively withmathematical models and also discussed the role of cornealcurvature mathematically.

The area of applanation in different instrument isanother variable that causes artifacts. Electronic tonom-eters22 have a smaller area of applanation and pneumatictonometers have a larger area of deformation than Gold-mann’s apparatus. Pneumatic tonometers23 measure IOPnot statically but dynamically with the cornea in motion,in contrast to other methods. Each of the current methodsof IOP measurement needs separate correction and adjust-ment for appropriate care of glaucoma. The Ehlers exper-iments may require some scrutiny. Pachymetry should havebeen done at the time of manometry, as sudden change inintracameral pressure may alter central corneal thickness.However, living eyes11 should yield more reliable informa-tion than cadaver eyes.10 The thinnest part of the corneamay lie slightly off the geometric or optical center of thecornea. Hydration state may alter the measurementquickly. However, these measurement errors create lessthan 10 �m of error, far smaller than the possible structuralerrors in the magnitude of as great as 150 �m.

The Ocular Hypertension Treatment Study2 recognizedthe significance of central corneal thickness on GAT.They found that eyes with thinner corneas had a higherrisk of developing primary open-angle glaucoma than eyeswith thicker cornea. As GAT of 24 to 32 was the inclusioncriterion for the study, eyes with thinner corneas hadhigher true IOP than the eyes with thicker corneas.Although central corneal thickness is important, it shouldbe considered as a masking factor, hiding elevated IOP,rather than an independent risk factor.

In conclusion, we concur with other authors’ findingsthat central corneal thickness causes systematic errors inGAT. Correlation between central corneal thickness andcorneal curvature was discovered in human eyes, with thethicker cornea being flatter and thinner cornea beingsteeper. Female cornea is thinner than male cornea.African Americans have thinner and relatively flattercorneas, compounding the underreading of GAT. Thismay contribute to delay in diagnosis and inadequate targetIOP setting for glaucoma therapy in this population andalso in normotensive and low tension glaucoma patients.Goldmann applanation tonometry was similar in all fourethnic groups, but when GAT was corrected by centralcorneal thickness and further by central corneal thicknessand corneal curvature, African Americans had signifi-cantly higher IOP than Caucasians. This puts them at adisadvantage even before glaucoma is diagnosed. A recentreport by Guzek and associates in the poster presented atthe American Academy of Ophthalmology meeting in

TABLE 11. GAT (A) Correction for Error (X) and 1/X

GAT (A) X 1/X

5 17 0.0588

10 16 0.0625

15 15 0.0666

20 14 0.0714

25 13.3 0.0751

30 12.6 0.0791

35 11.8 0.0847

40 11.2 0.0892

45 10.4 0.0961

GAT � Goldmann applanation tonometry.


2002 shows a 7.5% corrected prevalence of glaucoma in asample population in Ghana. Glaucoma screening ofAfrican Americans should be performed much earlieramong the younger population. Central corneal thick-nesses of Asians and Hispanics were found to be similar tothat of Caucasians. We present mathematical formulas (3–6) that are based on those of Ehlers and on our statisticalanalyses.

The diagnosis of glaucoma and determination of a targetIOP should be based on IOP adjusted by available correc-tion methods. We concur with the assertion by Orssengo21

that true IOP is found by adjusting GAT by pachymetry(central corneal thickness), keratometry (corneal curva-ture), and area of applanation. We present correctionformulas that offer statistical approximations. Theseshould be considered as ongoing working hypotheses sub-ject to further revisions based on elucidation of otherfactors involved such as hardness or softness of the tissue.

ACKNOWLEDGMENTS

We acknowledge statistical consultation by Ms Melissa L.Earl, MPH, Director, Innovative Medical and Epidemiol-ogy Data Solutions, Moreno Valley, California.

REFERENCES

1. La Rosa FA, Gross RL, Orengo-Nania S. Central cornealthickness of Caucasians and African-Americans in glauco-matous and non glaucomatous populations. Arch Ophthal-mol 2001;119:23–27.

2. Gordon MO, Beiser JA, Brandt JD, et al. The ocularhypertension treatment study. Arch Ophthalmol 2002;20:714–720.

3. Shah S, Chatterjee A, Mathai M, et al. Relationship be-tween corneal thickness and measured intraocular pressure ina general ophthalmology clinic. Ophthalmology 1999;106:2154–2160.

4. Argus WA. Ocular hypertension and central corneal thick-ness. Ophthalmology 1995;1102:1810–1812.

5. Herndon LW, Choudhri SA, Cox T, Damji KF, Shields MB,Allingham RR. Central corneal thickness in normal, glau-comatous, and ocular hypertensive eyes. Arch Ophthalmol1997;115:1137–1141.

6. Copt RP, Thomas R, Mermoud A. Corneal thickness inocular hypertension, primary open-angle glaucoma, and nor-mal tension glaucoma. Arch Opthalmol 1999;117:104–105.

7. Thomas R, Korah S, Muliyil J. The role of central cornealthickness in the diagnosis of glaucoma. Indian J Ophthalmol2000;48:107–111.

8. Johnson M, Kass MA, Moses RA, Grodzki WJ. Increasedcorneal thickness simulating elevated intraocular pressure.Arch Ophthalmol 1977;96:664–665.

9. Stodtmeister R. Applanation tonometry and correction ac-cording to corneal thickness. Acta Ophthalmol Scand 1998;76:319–324.

10. Goldmann VH, Schmidt T. Uber Applanationstonometrie.Ophthalmologica 1957;123:221–242.

11. Ehlers N, Bramsen T, Sperling S. Applanation tonometryand central corneal thickness. Acta Ophthalmol 1975;53:34–43.

12. Whitacre MM, Stein RA, Hassanein K. The effect of cornealthickness on applanation tonometry. Am J Ophthalmol1993;115:592–596.

13. Eyesteinsson T, Jonasson F, Sasaki H, et al. Central cornealthickness, radius of corneal curvature and intraocular pres-sure in normal subjects using non-contact techniques: Reyk-javik Eye Study. Acta Ophthalmol Scand 2002;80:11–15.

14. Foster PJ, Bassanhu J, Alsbirk PH, Munkhbayar D, Ur-anchimeg D, Jonson GJ. Central corneal thickness andintraocular pressure in a Mongolian population. Ophthal-mology 1998;105:969–973.

15. Wolfs RC, Klaver CC, Fingerling JR, Grobee DE, HoffmanA, deJong PT. Distribution of central corneal thickness andits association with intraocular pressure: Rotterdam Study.Am J Ophthalmol 1997;123:767–772.

16. Mark HH. Corneal curvature in applanation tonometry.Am J Ophthalmol 1973;74:223–224.

17. Mardelli PG, Piebenga LW, Whitacre MM, Siegmund KD.The effect of eximer laser photorefractive keratectomy onintraocular pressure measurements using Goldmann applana-tion tonometer. Surv Ophthalmol 1997;104:945–949.

18. Rashad KM, Bahnassy AA. Changes in intraocular pressureafter laser in situ keratomileusis. J Refract Surg 2001;17:420–427.

19. Doughty MJ, Zamen ML. Human corneal thickness and itsimpact on intraocular pressure measures: a review and meta-analysis approach. Surv Ophthalmol 2000;44:367–408.

20. Hafner A, Seitz B. Durch falsch normle Applanationstonom-etrie nach Goldmann maskiertes primaer chronisches Offen-winkelglaukom bei extremer Cornea plana. Klin MonatsblAugenheilkd 2001;218:621–625.

21. Orssengo GJ, Pye DC. Determination of the true intraocularpressure and modulus of elasticity of the human cornea invivo. Bull Math Biol 1999;61:551–572.

22. Mok KH, Wong CS, Lee VW. Tono-pen tonometer andcorneal thickness. Eye 1999;13:35–37.

23. Matsumoto T, Makino H, Uozato H, Saishin M, MiyamotoS. The influence of corneal thickness and curvature on thedifference between IOP measurements obtained with a non-contact tonometer and those with a Goldmann applanationtonometer. Nippon Ganka Gakkai Zasshi 2000;104:317–323.

24. Imbert A. Theorie sur ophthalmotonometre. Arch Ophthal-mol (Paris) 1885;5:358–363.

25. Fick A. Uber Messung des Druckes im Auge. Arch fur DieGesammte Physiologie des Menschen & der Thiere 1888;142:86–90.

26. Feltgen N, Leifert D, Funk J. Correlation between centralcorneal thickness, applanation tonometry, and direct intra-cameral IOP readings. Br J Ophthalmol 2001;85:85–87.


APPENDIX 1

● DERIVATION OF GOLDMANN FORMULA: The ImbertMaklakoff-Fick24,25 principle (1) describes the relationshipof forces working on an ideal, dry, perfect sphere withinfinitely thin wall. However, the cornea suffices none ofthe above conditions. Goldmann discussed the limitationsof applying the Imbert-Fick principle on the cornea.10 Awet surface creates surface tension (S), thickness of thewall creates counterforce as modulus of elasticity (E)against the force (W) applied on the sphere surface area(A).

The Imbert-Fick principle states: P � E � W/A � S.(Formula 1)

In which P � IOP, E � modulus of elasticity of cornealdeformation, corneal thickness being a major factor, W �the force acting on the tonometer tip, A � area of contactbetween the tonometer tip and flattened cornea, and S �the attractive force of surface tension.

By measuring various variables involved, Goldmanndesigned and calibrated the applanation prism with adiameter of 3.06 mm canceling surface tension (S) andeffect of the thickness of the cornea (E), simplifying theequation to: P � W/A. (Formula 2)

As A is a known value, knowing W by turning the dial,we know P, which equals IOP. Goldmann measured thethickness of cadaver eyes in Switzerland and assumed thecorneal thickness to be a constant value of a half amillimeter. He added 50 �m for combined thickness ofepithelium and endothelium but thought that they werenot significant in IOP determination. This was a greatinvention, but it was necessary for him to eliminate twovariables for this formula to work. He wrote that theoret-ically the variation in corneal thickness will affect thereading of IOP.

APPENDIX 2

● DERIVATION OF IOP CORRECTION FORMULAS: In1975, Ehlers and associates11 studied the relationshipbetween central corneal thickness and GAT in rabbits and29 human eyes. They measured central corneal thicknessby optical means, cannulated human eyes in vivo, andcompared the directly defined IOP and simultaneous GATmeasured with Perkins or Draeger hand-held applanationtonometers calibrated against a standard Goldmanntonometer. They found a statistically significant correla-tion between central corneal thickness and error of GAT(�P). Having seen a linear relationship between the twovariables, they calculated the intermediate pressure levelsfrom �P 10 and �P 30 by linear interpolation. Theypresented a correction table to obtain IOP from centralcorneal thickness and GAT readings. They eliminatedeyes with astigmatism greater than 1.5 diopters to avoid

the effect of corneal curvature on GAT. They found alinear relation between central corneal thickness andcorneal curvature in rabbit eyes but not in human eyes.

Whitacre12 presented one adjustment formula with cen-tral corneal thickness across the board based on regressionanalysis of 15 eyes. Argus4 offered one formula based onthe Whitacre study. The Doughty formula19 was derivedfrom meta-analysis of 300 data sets in the literature bydifferent authors, different methods, and different groups ofsubjects. Seeing a nonlinear relation between GAT anderrors, we cannot make adjustment with a single ratio or alinear formula. In extreme corneal curvature values, errorsin GAT become greater.20 The diverse variation in reportsof the effect of central corneal thickness on GAT seem tooccur from studying skewed populations, small samples,not taking other factors such as corneal curvature intoconsideration, and variations in instruments used.26

Extrapolating from the data of Ehlers and our observa-tions, the following statistical approximation formulas arepresented to calculate IOP from measurement outside thetable they offered. Ehlers and associates measured centralcorneal thickness by optical means and found that GATwas most accurate at a central corneal thickness of 520�m. As measured central corneal thickness values differ byoptical means and by ultrasonic pachymetry, the Ehlersexperiment needs to be repeated with ultrasonic pachym-etry of central corneal thickness, which is a more prevalentmethod of measurement. We propose a statistical adjust-ment formula of GAT to minimize the impact to currentpractice of measuring IOP by GAT by taking the statisticalmean central corneal thickness of 550 �m as a startingpoint of adjustment. The following correction formula isfor use by clinicians in facilitating clinical judgment ofmore accurate IOP. The difference (�T) between patient’scentral corneal thickness (T) and the average of 550 �m(550 � T) is divided by the ratio (X) of �T/�P.

We arrive at the following correction formula by centralcorneal thickness (T): P � A � (550 � T) / X. (Formu-la 3)

Where, P � IOP in mm Hg, A � Goldmann applana-tion reading in mm Hg, T � central corneal thickness byultrasonic pachymetry in �m, X � correction ratio be-tween differences in CCT (�T)/and differences in theGAT (�P): (�T/�P).

When one uses manual calculation, the following cor-rection ratio X may be used (Table 11). In an average eyewith GAT around 20, a correction factor of 14 will givereasonable correction to most GAT. Formula 3, above, wasused to correct GAT in our patient populations to com-pute IOP from GAT and central corneal thickness. Theerror of GAT increases at higher GAT level and thecorrection factor X was found to be a function of GAT(A), but the relation is nonlinear. On semilog graph, alinear relation was found, and the X is now expressed as anexponential function of A as 18e�0.005A. This may be usedto replace X in formula 3 for computerized calculation.


Ehlers also found that when IOP was corrected bycentral corneal thickness and corneal curvature, there wasa higher level of statistical significance in calculation ofadjusted IOP in multiple regression analysis. Each milli-meter of difference in corneal radius (r) from an average of7.85 mm causes 0.8 mm Hg of error in IOP.

Thus we extrapolate and get correction formula 4 bycentral corneal thickness and corneal curvature: P � A �(550 � T) / 18e�0.005A � 0.8 (r � 7.848837). (Formula 4)

Where, r � radius of curvature in millimeters.To transcribe the radius of curvature in millimeters to

dioptric power of the cornea, using Gulstrand’s formula: D� (n � 1) / R, or R � (n � 1)/D.

Where, D � dioptric power of the cornea, n � index ofrefraction of cornea � 1.3375, R � radius of cornealcurvature in meter.

Then adjusting decimals from meter to millimeters,r � 337.5/D, formula 4 is transcribed to: P � A �(550 � T)/18e�0.005A � 0.8 (337.5/D � 7.848837).(Formula 5)

Replacing D by averaging 2 keratometric values K1 andK2 in diopters, formula 5 is transcribed to: P � A � (550� T) / 18e�0.005A � 0.8 (675/(K1 � K2) � 7.848837).(Formula 6)

Formula 6 was used in our study to adjust GAT bycentral corneal thickness and corneal curvature.


Documents

Intraocular pressure, Goldmann applanation tension, corneal thickness, and corneal curvature in Caucasians, Asians, Hispanics, and African Americans