Corneal Biomechanical Changes in Diabetes Mellitus and Their Influence on Intraocular Pressure Measurements

Abstract

purpose. To investigate possible corneal biomechanical changes in patients with diabetes mellitus and understand the influence of such changes on intraocular pressure measurements.

methods. The study group was composed of 120 eyes from 61 healthy control subjects and 81 eyes from 43 diabetic subjects. IOP was measured first with an ocular response analyzer (ORA) and subsequently with a Goldmann applanation tonometer (GAT). Central corneal thickness (CCT) was measured with an ultrasonic pachymeter attached to the ORA. Axial length (AL), anterior chamber depth (ACD), and keratometry readings were acquired with partial coherence laser interferometry during the same visit before all IOP and CCT determinations were made.

results. Corneal hysteresis (CH) was found to be significantly lower in diabetic patients when compared with healthy control subjects (9.51 ± 1.82 mm Hg vs. 10.41 ± 1.66 mm Hg, P < 0.0001). There was no significant difference in terms of corneal resistance factor (CRF; P = 0.8). Mean CCT, GAT IOP, Goldmann-correlated IOP (IOPg), and corneal-compensated IOP (IOPcc) were significantly higher in diabetic patients than in healthy control subjects (P = 0.01 for CCT, P < 0.0001 for GAT IOP, IOPg, and IOPcc).

conclusions. Diabetes affects corneal biomechanics and results in lower CH values than those in healthy control subjects, which may cause clinically relevant high IOP measurements independent of CCT.

Accurate intraocular pressure (IOP) measurement is one of the most important steps in ophthalmic practice, especially in the diagnosis and assessment of the effectiveness of glaucoma treatments. The current gold standard to measure IOP is Goldmann applanation tonometry (GAT). However, it has been clearly documented that GAT measurements can be affected by several ocular factors such as corneal curvature, axial length (AL), and central corneal thickness (CCT). ^{1 2 3 4 5 6} Recent studies focusing on ocular hypertension have reconfirmed the importance of CCT in IOP measurements. ^{1 2 7} On the other hand, during the past decade, it has been proposed that CCT is just one of several corneal biomechanical properties that affect IOP measurement. Those biomechanical properties include corneal viscosity, elasticity, hydration, connective tissue composition, and regional pachymetry. ^{8 9} In a recent study, it was shown that the level of corneal elasticity may influence the effect of CCT on IOP measurement. ⁹  

There has been considerable interest in developing new tonometers that measure IOP independent of the biomechanical properties of the eye. One such recently developed instrument, the ocular response analyzer (ORA; Reichert Inc., Depew, NY), provides a measure of corneal viscoelasticity, which is called corneal hysteresis (CH). ¹⁰ CH is described as the viscous dampening in the corneal tissue in response to a deformation pulse by an air puff from the tonometer. It has been proposed that this measurement be ascribed to corneal tissue and be independent of the corneal curvature, corneal astigmatism, visual acuity, or AL. ¹⁰ However, limited data are available regarding the effects of these parameters on IOP measurement as obtained by the ORA. 

Various structural and functional abnormalities of the cornea in patients with diabetes mellitus, called diabetic keratopathy, have been reported, including impaired epithelial and endothelial function, recurrent corneal erosions, punctate keratitis, and delayed wound healing. ^{11 12 13 14} Nonenzymatic glycosylation (glycation) of proteins (the Maillard reaction) results in the formation of advanced glycosylation end products (AGEs), and this process exhibits several harmful reactions including the liquefaction of the vitreous body leading to diabetic retinopathy and retinal detachment, the rigidification of the cornea, and retinal microvascular alterations. ¹⁵ It has also been reported that glucose can act as a collagen cross-linking agent with the help of AGEs. ^{15 16} Advanced Maillard products accumulate in collagen proteins, result in the formation of covalent cross-linking bonds, and may lead to increased corneal thickening and biomechanical changes. ^{16 17} These changes may affect the measurement of IOP in an unexpected manner, such as an overestimation of the “true” IOP. To the best of our knowledge, no published study in the literature has investigated the in vivo corneal biomechanical changes in diabetic patients. The purpose of the present study, therefore, was to investigate the effect of diabetes on measurement of corneal properties and determine the effects of the presence and duration of diabetes mellitus, CCT, keratometry, anterior chamber depth (ACD), and AL parameters on IOP values obtained by ORA and GAT. 

Subjects and Methods

A total of 61 healthy control subjects and 43 diabetic patients were enrolled. Participation was voluntary. Informed consent was obtained from every patient at the beginning of the study. The local medical ethics committee approved the study and the tenets of the Declaration of Helsinki were observed. 

Each study participant underwent a complete ophthalmic examination that included refraction, slit lamp biomicroscopy, fundus examination, and visual field testing (Humphrey HFA750i; Carl Zeiss Meditec, Inc., Dublin, CA). As inclusion criteria, the subjects had to have a best corrected visual acuity of 20/40 or better, spherical error within ±5.0 D, and cylindrical error within ±3.0 D. Subjects were excluded if they had a history of intraocular surgery or refractive surgery and contact lens use. After a complete ophthalmic examination, patients with a suspicion of corneal disorder such as early keratoconus were examined by a cornea specialist in our department, and corneal topography measurements were performed to exclude any form of keratoconus. Study participants with an IOP more than 22 mm Hg obtained by GAT were excluded from the study. All participants underwent detailed examinations to rule out the possibility of glaucoma or ocular hypertension. 

All diabetic patients were recruited from the Department of Endocrinology and Metabolic Disorders, and all healthy patients were recruited from the General Ophthalmology Clinic. All study participants were Turkish citizens (of Caucasian origin). All diabetic patients had received a diagnosis of diabetes according to the most recent guidelines published by the American Diabetes Association. ¹⁸ There were 22 patients with type I diabetes who used insulin and 21 patients with type II who used oral antidiabetes drugs. 

Two independent, masked, experienced observers (AS, AB) performed all the measurements. The ORA measurements were taken by AS, and the GAT measurements were taken by AB during the study period. The two observers worked in two separate examination rooms and did not see each other during the study hours. An experienced technician accompanied all the patients and recorded all the measurements on the patients’ charts. Therefore, the two observers were masked with regard to IOP and study groups. The same ORA and GAT devices were used throughout the study. The calibration of the GAT was checked daily during the study. 

All subjects underwent measurement while sitting. IOP was measured first with the ORA (Reichert Inc.), then with the GAT, which was attached to a slit lamp biomicroscope. The time interval between tests of each tonometer was approximately 15 minutes. Subjects were given topical anesthesia (Alcaine; Alcon, Fort Worth, TX) bilaterally before the GAT and CCT measurements. CCT was measured with a built-in ultrasonic pachymeter attached to the ORA device. Three replicate measurements were acquired for each eye with ORA. If poor-quality waveforms were obtained, they were deleted, and a new measurement was taken. AL, ACD, and keratometry readings were acquired with an ocular biometer (IOLMaster; Carl-Zeiss Meditec, Inc.) by a trained ophthalmology resident (GO) during the same visit before all IOP and CCT determinations had been made. 

The ORA uses a rapid air impulse to apply force to the cornea. During this process, an advanced electro-optical system monitors the deformation. The air-pulse causes the cornea to move inward, past the first applanation and into a slight concavity. After this, the air pulse is shut down, and the force applied starts to decrease steadily in a symmetrical fashion. As the air pressure decreases, the cornea reaches a second applanation state while returning from concavity to its healthy convex form. The two applanations occur within approximately 25 ms. An electro-optical detector system monitors the reflection of the light beam from the surface of the cornea. As the cornea reaches the first and second applanation states, the signal monitored by the detector reaches a peak level. Furthermore, as the cornea moves into concavity or convexity, the generated signal reduces significantly. This dynamic process generates two signal peaks that define two applanation states. Two corresponding pressures of an internal air supply plenum are determined from the applanation times, as derived from the detector signal peaks. These two pressures (P ₁ and P ₂) are defined as the intersection of a vertical line drawn through the peaks of the applanation curve with the plenum pressure curve. CH is defined as the difference in pressure between these two applanation pressures (CH = P ₁ – P ₂). ¹⁰ Other variables obtained by the ORA are included: corneal-compensated IOP (IOPcc), Goldmann-correlated IOP (IOPg), and corneal resistance factor (CRF). IOPcc is derived from the equation P ₂ − 0.43 × P ₁, where P ₁ and P ₂ are the first and second applanation pressures, respectively. IOPcc is proposed to be less affected by corneal properties than GAT. IOPg is the average of P ₁ and P ₂. CRF is an optimized corneal biomechanical parameter that is derived from specific combinations of the ORA-induced inward (P ₁) and outward (P ₂) applanation values. CRF is derived from the formula CRF = k ₁ · (P ₁ – P ₂) + 0.3 · k ₁ P ₂ + k ₂, where k ₁ (0.149) is a calibration constant and k ₂ (−6.12) is a calibration offset. ¹⁹ These k constants were derived using proprietary algorithms based on the results of a large-scale clinical data analysis according to the manufacturer. 

In our study, all statistical analyses were performed with commercial software (SPSS for Windows, ver. 15.0; SPSS Inc, Chicago, IL). The level of significance was set at P < 0.05. Three consecutive acceptable readings within a range of ±2 mm Hg for IOP and ±5 μm for CCT were recorded. For all statistical tests, the mean IOP and CCT of the three values within the preferred range were used. The correlation between the IOP measurements obtained by the different methods was analyzed with Pearson’s correlation coefficient. The mean IOP measurement by the ORA was compared to that of the GAT using the Bland-Altman analysis. 

Results

The study included 120 eyes from 61 healthy control subjects (33 women, 28 men) and 81 eyes from 43 diabetic patients (26 women, 17 men). The mean ± SD age of the included subjects was 53 ± 9 years (range, 21–78) and 55 ± 11 years (range, 21–78 years) for the healthy control subjects and the diabetic patients, respectively (P > 0.05). The mean duration of diabetes was 13.5 ± 6.3 years. Gender distribution was similar between the groups (P > 0.05). Table 1shows the results of GAT, ORA, ultrasound pachymetry, and biometric parameters of the studied eyes. 

CH was significantly lower in diabetic patients than in healthy control subjects (P < 0.0001). There was no significant difference in CRF (P = 0.8). Mean CCT, GAT IOP, IOPg, and IOPcc were significantly higher in diabetic patients than in healthy control subjects (P = 0.01 for CCT, P < 0.0001 for GAT IOP, IOPg, and IOPcc). AL, ACD, and keratometry measurements did not show any difference between the groups (P = 0.7 for AL and ACD, P = 0.09 for keratometry). 

The correlation analyses for CH and CRF in the diabetic patients and the control groups are shown in Tables 2 and 3 , respectively. CH and CRF showed a correlation with CCT (Figs. 1 2)and GAT IOP (Figs. 3 4) . 

Tables 4 and 5show the multivariate regression analysis of the association between GAT IOP and IOPcc measurements, as well as the age, CCT, CRF, CH, AL, ACD, duration of diabetes, and keratometry for both groups. In the diabetic group, GAT IOP measurements also showed an association with keratometry (P = 0.04). There was no significant relationship between the remaining variables in any of the groups. In addition, we investigated the role of HbA1c and disease duration on CH and CRF by using a simple linear regression analysis. However, neither HbA1c nor disease duration had any statistically significant effect on CH and CRF (P > 0.05 for both). The diabetic patients were further divided into two subgroups according to their HbA1c levels (subgroup 1 = HbA1c ≤ 7.5 mg/dL, subgroup 2 = HbA1c > 7.5 mg/dL). There were no differences in mean CH and CRF between the subgroups (CH, 9.6 ± 1.9 and 9.9 ± 1.3 mm Hg; CRF, 10.4 ± 1.7 and 10.2 ± 1.7 mm Hg for subgroups 1 and 2, respectively; all P > 0.05). 

Figure 5shows the Bland-Altman plot of the agreement between IOPcc and GAT IOP in healthy control subjects and diabetic patients. The mean ± SD difference between IOPcc and GAT IOP was 1.29 ± 2.86 mm Hg (95% limits of agreement: −4.32 to 6.89 mm Hg) for healthy control subjects and 2.08 ± 3.27 mm Hg (95% limits of agreement: −4.32 to 8.48 mm Hg) in diabetic patients. The mean differences were significantly different from 0 in both groups (both P < 0.0001). As mean IOP increased, ORA IOPcc overestimated the GAT IOP significantly. The correlation in the differences between IOPcc and GAT IOP with regard to CCT was negative in healthy control subjects, whereas there was no correlation in the diabetic group (Fig. 6) . 

Discussion

There has been considerable interest in the impact of some corneal parameters, especially CCT, as a potential determinant of measured IOP and glaucoma risk and/or progression. ^{2 7} Measured IOP is affected by CCT in different tonometers. ^{3 4 5 6} However, this relationship has not been precisely specified, and has been thought to be nonlinear in the range of typical IOP. Recently, it has been shown that tonometry is affected by overall corneal biomechanical characteristics other than CCT. ⁹ However, in vivo measurement of these parameters is currently not easy. ORA is a recently introduced device that provides a measure of corneal viscoelasticity called CH. ¹⁰ Studies that have been conducted to investigate CH and CRF in healthy subjects, ^{20 21} as well as in patients with glaucoma, ²² keratoconus, ²³ and Fuchs’ endothelial dystrophy. ¹⁰ However, no published data have been available in the literature that indicate the changes in measures of corneal viscoelasticity. 

In the present study, we found mean CH and CRF of 10.41 and 10.36 mm Hg, respectively, in healthy control subjects, similar to those in previous reports. ^{10 20 23 24 25} CH can be described as the dynamic response of the cornea and reflects the capacity of corneal tissue to absorb and dissipate energy. ¹⁰ CH and CRF exhibited similar associations with CCT and GAT IOP in healthy control subjects, but CH did not show a positive correlation with GAT IOP in diabetic patients (Figs. 1 2 3 4) . Perhaps as important as the relatively low correlation of CH to CCT in diabetic patients was the high degree of variability in CH for any specific CCT. In other words, diabetic eyes with the same CCT varied greatly in CH when compared to normal eyes. Several investigators have reported that CH is correlated with CCT, as we saw in our study. ^{19 20 21 22 24} However, there have been ongoing concerns about the relationship between CH and CCT. It has been reported that the measurements of corneal properties decreased significantly after LASIK, ^{26 27} but it still remains unclear what clinical and structural factors affect CH. From this point of view, further studies are needed to elucidate the exact relationship of CH with ocular variables (e.g., CCT, keratometry). The association between CRF and CCT has been shown by us. ²⁸ and is in agreement with the prediction made by the manufacturer (Luce D, et al. IOVS 2006;47:ARVO E-Abstract 2266). CRF was originally calculated as a linear function of pressure peaks 1 and 2 to maximize the correlation between CRF and CCT on the basis of a large-scale data analysis, as obtained by the manufacturer (Luce D, et al. IOVS 2006;47:ARVO E-Abstract 2266). These findings indicated that variations in CCT were associated with changes in CH and CRF. With regard to IOP, higher GAT IOP and possibly some compromised aspects of the biomechanical properties in the diabetic group may play a complex role that contributes to the observed CH change in response to IOP change. 

In the present study, CH was significantly lower in diabetic patients, whereas CRF was not significantly different from that of control subjects (Table 1) . Although the exact physiological properties of CH and CRF are still unclear, CH is thought to be an indicator of corneal viscous dampening, and CRF is thought to be a correction factor that reduces the effect of CCT on IOP measurement. ^{10 26 28} Lower CH in diabetic patients may be explained by an alteration in the collagenous components due to collagen cross-linking. The lower CH in diabetic patients suggests that the dampening effects of the cornea decrease due to diabetes and are induced to increase during the cross-linking of collagen fibrils. ¹⁶ CH has also been shown to be reduced in patients with Fuchs’ dystrophy, keratoconus, and congenital glaucoma, and especially in those with marked Haab’s striae. ^{10 29} The biomechanical behavior similarities of these corneas to the corneas of patients with diabetes in the present study may be indicative of alterations in the corneal microstructural properties known to occur in these corneal diseases. ^{30 31} As mentioned, CRF was not significantly different between the groups. We could not find an exact explanation for this finding. Since our study may be accepted as a preliminary report, this finding deserves more research. Taken altogether, lower CH in diabetic patients could indicate less damping ability. 

In our study, mean CCT and GAT IOP, IOPcc, and IOPg were significantly higher in diabetic patients than in healthy control subjects. This result was consistent with those in large population-based studies of diabetic patients in which a consistently higher IOP was found. ^{32 33 34 35} However, there is still controversy about whether diabetic patients really have higher IOPs in light of the finding that thicker corneas yielded higher IOP measurements with various tonometers. As mentioned, corneal collagen cross-linking may lead to biomechanical changes, which can lead to an overestimation of the true IOP with GAT. ^{1 4 5 6 7} The altered collagen structure may be additive to the effect of a thicker cornea, thereby further increasing the measured IOP. This additional effect was illustrated in a recent study in which the cross-linking of corneal collagen was observed experimentally. ³⁶ In this study, a pneumotonometer and a hand-held tonometer (Tono-Pen; Reichert) measured IOP at approximately the same level in a cadaveric eye when the eyes were held at a constant pressure of 30 mm Hg. The pressure was monitored after cannulation of the vitreous chamber with a transducer. After cross-linking, both devices measured the IOP at more than twice that level. Liu and Roberts ⁹ analyzed the effect of CCT, corneal radius, and Young’s modulus on GAT IOP measurements through a mathematical model and showed that variation in the elasticity of the cornea within a range predicted to occur in a normal population would result in an IOP measurement error of as much as 17 mm Hg. If the impact of these altered corneal properties in diabetic patients is extrapolated with regard to IOP measurement, the results of these large epidemiologic studies could certainly be looked at from a different perspective. The majority of these studies, such as the Beaver Dam Eye Study, ³⁷ the Los Angeles Latino Eye Study, ³⁸ and the Blue Mountains Eye Study, ³⁹ demonstrated that diabetes is a risk factor for the development and progression of glaucoma. At the same time, other large studies, including the Rotterdam Study, ⁴⁰ the Baltimore Eye Study, ³² and the Barbados Eye Study, ⁴¹ showed that diabetes was not a significant risk factor for development and progression of glaucoma. At least one explanation can be postulated for the protective effect of diabetes in the development of glaucoma. As mentioned, glucose-mediated corneal collagen cross-linking may have been responsible for the overestimation of IOP. If we consider the possibility that the true IOP in these patients is statistically lower, then it is possible that the threshold for significant pressure-dependent damage to the optic nerve may have not been reached. ⁴² Therefore, the exact relationship between of IOP and glaucoma to diabetes remains controversial. 

One of our objectives in this study was to investigate the agreement between GAT IOP and ORA IOPcc. We wanted to know how much the ORA IOPcc is likely to differ from the GAT IOP. If the difference is not sufficient to cause problems in clinical interpretation, we can use the two interchangeably. Correlation analysis is one of the statistical approaches used when the reference method is low in error. However, all IOP measurement techniques have an inherent error. A Bland-Altman analysis ⁴³ is a statistical method that allows the clinician to compare two different measurement techniques. When the agreement between GAT IOP and IOPcc was determined by Bland-Altman analysis, the discrepancy between the tonometers became significantly larger in the diabetic group, especially for the higher GAT-determined IOPs (Fig. 5) . Although IOPcc and GAT IOP demonstrated a strong linear correlation in both groups, we found a high degree of variability in the difference between IOPcc and GAT IOP in diabetic patients. For pressures >20 mm Hg, ORA IOPcc showed a significant tendency to overestimate the GAT IOP readings. The difference between IOPcc and GAT correlated negatively with CCT in healthy control subjects (Fig. 6) . The difference approached zero for a CCT of around 590 μm. In diabetic patients, at a thin CCT the difference was negative, which indicated an overestimation of the IOP as measured by GAT when compared to the IOP measured by ORA. The difference approached 0 and became positive for a thick CCT. Our findings for healthy control subjects are similar to those reported previously ^{24 26} and further support reports that differences between IOPcc and GAT are significantly related to CCT in healthy control subjects. This further suggests that the effect of corneal biomechanical alterations is not completely eliminated from the manufacturer’s ORA IOPcc measurement. 

Ocular biometric variables such as AL and keratometry can affect the accuracy of IOP measurements as obtained by GAT. ⁴ In the present study, we found that GAT IOP measurements correlated with AL and keratometry in both groups. On the other hand, IOPcc measurements were not associated with age, CCT, AL, ACD, and keratometry in multivariate models for both groups. This result was similar to those reported in studies that seemed to indicate that the effect of keratometry was also taken into account when measurements of the corneal properties were obtained by the ORA. 

This study has limitations that must be addressed. One is that the groups were not well matched by GAT IOP. A recent study proposed that CH is IOP dependent and decreases in a given eye when IOP is elevated. ²¹ On the contrary, in a study by Luce, ¹⁰ CH was found to be IOP independent. There are still inconsistencies regarding the relationship between CH and IOP. The measurements were not performed during the same office hours in all patients, to minimize the effect of diurnal IOP changes. However, since both IOP measurements were performed consecutively within 15 minutes, we think that the effect of diurnal variation can be excluded. In addition, no significant 24-hour change in corneal viscoelasticity was detected with ORA. ⁴⁴ Another limitation was that we did not include patients with ocular hypertension (nondiabetic patients) and/or glaucoma in the study. 

In conclusion, diabetes causes significantly lower CH, most likely due to biomechanical changes in the eye. Clinicians should take this finding into account in routine practice because clinically relevant IOP measurement errors may occur independent of CCT. Further studies are needed to explain the complex relationships among IOP, ocular biomechanics, glaucoma, and diabetes. 

 Presented in part at the annual meeting of the American Academy of Ophthalmology, Atlanta, Georgia, November 2008.

 Submitted for publication August 25, 2008; revised October 20, 2008, and January 10, February 12, and March 25, 2009; accepted July 22, 2009.

 Disclosure: A. Şahin, None; A. Bayer, None; G. Özge, None; T. Mumcuoğlu, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

Clinical and Ocular Characteristics of the Study Participants

	Healthy Control			Diabetic			P
		Mean ± SD	Range	Mean ± SD	Range
Age, y	53.15 ± 9.74	21–78	55.39 ± 11.64	21–78	0.14
CCT, μm	535.46 ± 39.20	460–643	550.07 ± 40.76	433–624	0.01
GAT IOP, mm Hg	14.55 ± 3.12	8–22	16.71 ± 3.14	11–22	0.0001
IOPg, mm Hg	15.34 ± 3.66	7–25	17.68 ± 4.42	10–30	0.0001
IOPcc, mm Hg	15.85 ± 3.24	7–24	18.81 ± 4.71	12–32	0.0001
CRF, mm Hg	10.36 ± 1.97	5–15	10.32 ± 1.76	6–13	0.8
CH, mm Hg	10.41 ± 1.66	5–15	9.51 ± 1.82	4–13	0.0001
AL, mm	23.24 ± 0.97	20–25	23.19 ± 0.78	20–24	0.78
ACD, mm	3.12 ± 0.35	2–3	2.97 ± 0.48	1–3	0.09
K, D	43.68 ± 1.46	40–47	43.72 ± 1.37	40–46	0.87
HbA1c, mg/dL	N/A	N/A	7.31 ± 1.53	5–11	N/A

 Bold type indicates statistical significance.

Table 2.

 

Results of Correlation Analyses for CH and CRF in Healthy Control Subjects

	Age	CCT	GAT IOP
CRF
r	0.20	0.71	0.74
P	0.03	0.0001	0.0001
CH
r	0.13	0.56	0.40
P	0.15	0.0001	0.0001

 Bold type indicates statistical significance.

Table 3.

 

Results of Correlation Analyses for CH and CRF in Diabetic Patients

	Age (y)	CCT (μm)	GAT IOP (mm Hg)
CRF
r	−0.12	0.56	0.31
P	0.29	0.0001	0.001
CH
r	−0.11	0.30	−0.28
P	0.33	0.01	0.01

 Bold type indicates statistical significance.

Figure 1.

 

The relationship between CH and CCT in healthy control subjects and diabetic patients (CH = −2.27 + 0.024 × CCT, R ² = 0.3, P < 0.0001 in healthy control subjects; CH = 2.02 + 0.014 × CCT, R ² = 0.09, P = 0.006 in diabetic patients).

Figure 2.

 

The relationship between CRF and CCT in healthy control subjects and diabetic patients (CRF = −8.79 + 0.036 × CCT, R ² = 0.50, P < 0.0001 in healthy control subjects; CRF = −2.9 + 0.024 × CCT, R ² = 0.31, P < 0.0001 in diabetic patients).

Figure 3.

 

The relationship of CH and CRF with GAT in healthy control subjects (GAT IOP = 6.79 + 0.74 × CH, R ² = 0.15, P < 0.0001; GAT IOP = 2.49 + 1.16 × CRF, R ² = 0.54, P < 0.0001).

Figure 4.

 

The relationship between CH and CRF and GAT in diabetic patients (GAT IOP = 21.29 − 0.48 × CH, R ² = 0.07, P = 0.01; GAT IOP = 11 + 0.55 × CRF, R ² = 0.09, P = 0.005).

Table 4.

 

Results of Multivariable Linear Regression Analysis between IOP Measurements and Other Ocular and Clinical Variables in Healthy Control Subjects.

	GAT IOP (mm Hg)							IOPcc (mm Hg)
		Coefficients (SE)	P	95% CI			Coefficients (SE)	P	95% CI
Age, y	0.11 (0.02)	0.11	−0.01	0.08	−0.01 (0.03)	0.75	−0.08	0.06
CCT, μm	0.06 (0.01)	0.52	−0.01	0.02	0.01 (0.008)	0.08	−0.002	0.03
CRF, mm Hg	1.44 (0.22)	0.0001	1.87	2.75	Not included
CH, mm Hg	−0.90 (0.24)	0.0001	−2.21	−1.26	Not included
AL, mm	−0.15 (0.24)	0.05	−0.94	0.00	0.39 (0.38)	0.31	−0.37	1.17
ACD, mm	0.03 (0.55)	0.62	−0.83	1.38	−0.41 (0.93)	0.65	−2.27	1.43
K, D	−0.04 (0.15)	0.58	−0.38	0.21	−0.16 (−0.08)	0.51	−0.66	0.33

 Bold type indicates statistical significance. CI, confidence interval.

Table 5.

 

Results of Multivariable Linear Regression Analysis between IOP Measurements and Other Ocular and Clinical Variables in Diabetic Patients

	GAT IOP (mm Hg)							IOPcc (mmHg)
		Coefficients (SE)	P	95% CI			Coefficients (SE)	P	95% CI
Age, y	−0.10 (0.03)	0.44	−0.09	0.04	0.03 (0.06)	0.58	−0.101	0.178
CCT, μm	0.21 (0.02)	0.33	−0.02	0.06	0.06 (0.03)	0.27	−0.004	0.134
CRF, mm Hg	0.88 (0.34)	0.0001	0.99	2.39	Not included
CH, mm Hg	−0.57 (0.32)	0.0001	−1.96	−0.64	Not included
AL, mm	0.07 (0.74)	0.67	−1.20	1.83	−0.93 (1.55)	0.55	−4.10	2.23
ACD, mm	−0.23 (1.08)	0.13	−3.91	0.54	−0.42 (2.31)	0.85	−5.15	4.30
K, D	0.37 (0.43)	0.04	0.06	1.84	0.01 (0.91)	0.98	−1.85	1.89

 Bold type indicates statistical significance.

Figure 5.

 

Bland-Altman analysis of the agreement between ORA IOPcc and GAT IOP measurements in both groups. Dotted lines and straight lines: the means and 95% limits of agreements in healthy control subjects and diabetic patients, respectively.

Figure 6.

 

The difference between ORA IOPcc and GAT IOP measurements versus CCT.