Introduction

Accurate estimation of intraocular pressure (IOP) is important because it is an essential factor in precise patient classification in diagnosis and efficacy assessment of glaucoma treatment. Several studies have shown that 30–60% of ocular hypertension (OH) and normal tension glaucoma (NTG) patients are misclassified and possibly improperly treated due to a corneal thickness that is significantly deviated from the norm.1, 2, 3, 4 Goldmann applanation tonometry, in which the tonometer is calibrated based on a central corneal thickness (CCT) of 500 μm, has been the gold standard for IOP measurement for decades.5 Studies comparing measurements with manometry and tonometry indicate that measuring CCT is essential to properly interpret the results obtained with a Goldmann applanation tonometer (GAT).6, 7, 8 To estimate the true intraocular hydrostatic pressure from the GAT readings, several investigators have proposed some formulae, which we employed for the purposes of this study, that take CCT into account.6, 7, 9, 10, 11 If applying these correction formulae to the Ocular Hypertension Treatment Study cohort, the proportion of subjects with a baseline IOP≤21 mmHg in the randomly chosen eye at entry would be 52%, which would be only 10% if uncorrected by these formulae.4

The noncontact tonometer (NCT) is now widely used as a screening tool for glaucoma because corneal anaesthesia and direct corneal contact are not required to obtain a reading. NCT readings are highly correlated with those of GAT in the normal IOP range.12, 13, 14 However, little is known about the impact of CCT on the NCT measurements. Recently, the ocular blood flow tonometer (OBFT) has been introduced as another option for measuring IOP and pulsatile ocular blood flow. It comes with a disposable head that offers protection against the risk of disease transmission. The manufacturers claim that IOP measurements with the OBFT are not affected by variations in CCT, a statement that needs to be verified.

In order to test the effects of variation in CCT on readings from each of the tonometers described above, we compared the IOP measurements obtained with the GAT, NCT, and OBFT in control subjects and in patients with glaucoma or ocular hypertension, and then evaluated the relationship between CCT and these measurements. Additionally, we quantified the NCT and OBFT measurement errors related to the variation in CCT by using the CCT-corrected GAT values as the standard.

Materials and methods

Patients were recruited consecutively from the outpatient clinic of one physician (CJL). All subjects were 18 years or older, and verbal and written consent was obtained from each. Institutional Review Board (IRB) approval was also obtained for the study. All subjects had a normal cornea on biomicroscopy, a spherical equivalent between −5 and +3 dioptres, a keratometric astigmatism <2 dioptres, and a clear optic medium allowing fundus examination. Contact lens wearers and subjects with connective tissue disorders, ocular inflammation, or a history of ocular surgery were excluded. Control subjects had untreated GAT values <21 mmHg and no ocular abnormalities other than mild refractive error or senile cataract. Patients with primary open-angle glaucoma (POAG), chronic primary angle closure glaucoma (CACG), NTG or OH were enrolled if they met all the inclusion and exclusion criteria except for the glaucomatous optic neuropathy or IOP status.

One masked, experienced ophthalmologist (YK) performed the following examinations for each subject. IOP was first measured with the same NCT (Topcon CT60, Topcon Corporation, Tokyo, Japan) to get three reliable readings. After 10 min, five consecutive measurements of CCT with one ultrasound pachymeter (DGH 1000, DGH Technology, Inc., Frazer, PA, USA) were taken, and the mean of the middle three readings (in terms of numeric value) was used for analysis. After allowing the patient to rest for another 10 min, IOP was measured with the same GAT and with one OBFT (Ocular Blood Flow Ltd, Wiltshire, UK, software version 16.2). The GAT and OBFT measurements were performed in random order and separated by 15 min. Three measurements were obtained in each eye for each tonometer and the mean was used for analysis. All the instruments were calibrated according to the manufacturers' guidelines.

The corrected GAT value was calculated in three ways according to the formulae (Appendix) proposed by Ehlers et al,5 Whitacre et al,6 and Doughty et al,9 respectively. The deviations of the NCT and OBFT readings from the corrected GAT values were calculated and correlated to the CCT variables.

Statistical analysis

If both eyes of one patient were eligible, then one eye was chosen at random for analysis. Data were analysed with SPSS for windows, version 10.0. Group comparisons were performed with the χ2-test for categorical variables and one-way ANOVA test for continuous variables. The correlation between the IOP measurements obtained with different methods was analysed by Pearson's correlation method. The NCT and OBFT measurements were each separately compared with the GAT measurements by a paired t-test. The correlation between CCT and IOP was evaluated by a linear regression analysis. The level of significance was set at P<0.05. For multiple comparisons, the alpha levels were adjusted with the Bonferroni post hoc test.

Results

A total of 56 control subjects and 114 patients (POAG: 25, CACG: 40, NTG: 38, OH:11), all Asians, were enrolled. The demographic and clinical data for these subjects are listed in Table 1. Patients with OH were younger (P<0.001) and had thicker corneas (P<0.001) than the other groups, while NTG patients had thinner corneas than the control subjects (P=0.001).

Table 1 Demographic and clinical data
Full size table

Both the NCT (r=0.872, P<0.001) and OBFT (r=0.861, P<0.001) readings were highly correlated with the GAT readings. The NCT readings were higher than the GAT readings in control subjects (P=0.001) and in patients with OH (P=0.001) and POAG (P=0.004) (Table 2). The OBFT tended to have lower readings than the GAT, but the difference was not significant after Bonferroni correction in each subgroup (Table 2).

Table 2 Measurements with the Goldmann applanation tonometer (GAT), noncontact tonometer (NCT), and ocular blood flow tonometer (OBFT)
Full size table

Since a negative correlation existed between age and CCT (P<0.001) as well as between age and IOP (P<0.001), age was adjusted while performing linear regression of IOP on CCT. The IOP measurements with the three methods were all significantly correlated with CCT (all P<0.001) (Figure 1). The NCT measurements showed the greatest regression coefficient (β=0.063, r=0.650), while the GAT measurements showed the least regression coefficient (β=0.037, r=0.496). The deviations of the NCT and OBFT readings from the corrected GAT values were highly correlated with CCT, no matter which correction formula was used (all P< 0.001) (Figures 2 and 3). Every 10 μm change in CCT would yield a 0.98 mmHg deviation in NCT measurements (r=0.896) and a 0.81 mmHg deviation in OBFT measurements (r=0.837) according to the formula proposed by Ehlers et al.5 Based on the GAT values corrected with the formulae of Doughty et al6 and Whitacre et al,9 a 10 μm CCT change would result in a 0.65 mmHg (r=0.801) and a 0.47 mmHg (r=0.716) error in NCT measurements respectively, and a 0.48 mmHg (r=0.665) and a 0.29 mmHg (r=0.517) deviation in OBFT measurements, respectively.

Figure 1
figure 1

Correlation of CCT with the IOP measurements obtained with the NCT, OBFT and GAT.

Full size image
Figure 2
figure 2

Correlation between CCT and the deviation of NCT measurements from the corrected GAT values. ▵: deviation of NCT measurements from the corrected GAT values that are calculated according to the formulae derived from the studies of Ehlers et al, Doughty et al, and Whitacre et al, respectively.

Full size image
Figure 3
figure 3

Correlation between CCT and the deviation of OBFT measurements from the corrected GAT values. ▵: deviation of OBFT measurements from the GAT values that are corrected according to the formulae derived from the studies of Ehlers et al, Doughty et al, and Whitacre et al, respectively.

Full size image

Discussion

The realization of a wide range of CCT in normal eyes and the advent of excimer laser refractive surgery prompted ophthalmologists to pay attention to the impact of CCT on IOP measurements. Although the CCT-induced error in GAT measurements has been quantitated by manometric studies, similar studies have never been performed with the NCT or OBFT in humans. Little is known about the degree to which the NCT or OBFT readings should be adjusted in eyes with a CCT significantly deviated from the population mean. We demonstrate that the measurements using the three methods are all influenced by CCT, with the NCT readings being the ones most affected and the GAT readings, the least. Using the corrected GAT value as a standard, we found that a 10 μm change in CCT could yield a 0.47–0.98 mmHg deviation in the NCT measurements and a 0.29–0.81 mmHg deviation in the OBFT measurements.

The correlation between the NCT and GAT measurements has been widely explored, but articles dealing with the relationship between CCT and the NCT measurements are scanty. In a population study on subjects aged 50 years or older, Eysteinsson et al15 found that every 10 μm increase in CCT is associated with a rise of 0.22–0.28 mmHg in NCT (Nidek NT 2000) readings. However, their study did not take into account individual variation in true intraocular hydrostatic pressure while estimating the influence of CCT on IOP readings. Therefore, those results might not truly reflect the amount of the CCT-induced error. Another study reported a significant intraindividual difference between the NCT (Topcon CT60) and GAT measurements in 60 glaucoma patients (120 eyes).16 The difference was positively correlated with CCT in eyes with a corneal thickness greater than 574.5 μm. This indicates that the CCT-induced overestimation is greater with the NCT than that with the GAT in eyes with a thick cornea. Our results support this conclusion. In addition, we found that the NCT readings could be overestimated by as great as 0.98 mmHg with a 10 μm increase in CCT. Conversely, the linear regression analyses demonstrate that the NCT readings could be underestimated in eyes with thin corneas.

The OBFT is a pneumotonometer that is designed by applying the elastic plate theory, and taking into account the radius and bending characteristics of the cornea.17 The manufacturer claims that the OBFT measurements are not affected by CCT, which is based on the results of two studies that use pneumotonometer to measure IOP before and after photorefractive keratectomy.18, 19 However, our results show that the OBFT measurements are influenced by CCT. By analysing the relationship between CCT and the OBFT readings, several recent studies also showed that thicker corneas are associated with higher OBFT readings.20, 21, 22 Bhan et al20 found that the OBFT measurements were more affected by variations in CCT than the GAT and Tono-Pen measurements, with a mean increase in IOP with increasing CCT of 0.28 mmHg/10 μm. Morgan et al21 showed that a 10 μm change in CCT was equivalent to a 0.3 mmHg change in OBFT readings. Again, the above-mentioned studies did not take individual pressure variations into account while evaluating the effect of CCT on OBFT measurements. In the present study, we found an error of 0.29–0.81 mmHg in OBFT readings with a 10 μm change in CCT.

This study recruited eyes with normal corneas, a spherical equivalent between −5.0 and 3.0 dioptres, and a keratometric astigmatism <2 dioptres. We did not consider the corneal curvature because the effect of it on applanation tonometer readings remains uncertain. Using 925 randomly sampled right eyes, Eysteinsson et al15 could not establish any relationship between the radius of corneal curvature and the NCT measurements. Paranhos et al23 also found no significant influence of corneal curvature on IOP readings obtained with the GAT and NCT in 404 eyes with keratometric readings between 8.6 and 6.39 mm. However, some suggested that corneal curvature should be taken as a correction factor in order to obtain true IOP.

Shimmyo et al24 derived a correction formula for adjusting GAT measurement by CCT and corneal curvature from statistical approximations. They found each millimetre of difference in corneal radius from an average of 7.85 mm caused 0.8 mmHg of error in GAT measurements. Orssengo and Pye25 developed a theoretical mathematical model of the cornea, and proposed a correction factor between GAT measurement and true IOP according to corneal dimensions including central thickness and anterior radius of curvature. If applying Orssengo's and Shimmyo's formulae (Appendix) to calculate the corrected GAT values in our study cohort,24, 25 a 10 μm CCT change would result in a 0.59 mmHg (r=0.728) and a 0.78 mmHg (r=0.825) deviation in NCT measurements respectively, and a 0.45 mmHg (r=0.642) and a 0.65 mmHg (r=0.757) deviation in OBFT measurements respectively. Thus, the CCT-induced deviation in NCT or OBFT measurements from the corrected GAT values that are adjusted by both CCT and anterior corneal curvature is within the range derived from this study. This could probably be explained by the fact that the GAT measurement error induced by corneal curvature is much less than CCT-induced error, a notion supported by the studies of Shimmyo et al, and Orssengo and Pye.

In this study, we used the corrected GAT values as the standard to evaluate the impact of CCT on the NCT and OBFT measurements. Although this model may not be as accurate as a manometric study, it provides information which is more useful than that derived from simply correlating the relationship between CCT and IOP readings. Our findings may help clinicians interpret pressure readings obtained with the NCT and OBFT in each individual.

In conclusion, measurements with the GAT, NCT, and OBFT are all affected by variations in CCT, with the NCT readings being the ones most affected and the GAT readings, the least. In this way, our results concur with the previous studies and indicate the importance of adjusting IOP readings according to individual corneal thickness to avoid pressure overestimation or underestimation, both of which could lead to misclassification of a patient in diagnosis and affect the efficacy assessment of prescribed treatment, which is relevant and important in the clinical evaluation of glaucoma.