Different Effects of Aging on Intraocular Pressures Measured by Three Different Tonometers

This study aimed to compare intraocular pressures (IOP) using different tonometers, Goldmann applanation (IOPGAT), non-contact (IOPNCT), and rebound (IOPRBT), and to assess the effects of aging and central corneal thickness (CCT) on the measurements. The IOPGAT, IOPNCT, IOPRBT, mean patient age (65.1 ± 16.2 years), and CCT (521.7 ± 39.2 µm) were collected retrospectively from 1054 eyes. The differences among IOPs were compared by the paired t-test. Possible correlations between devices, age, and CCT were assessed by linear regression analyses. The effects of age and CCT on the IOP reading were assessed by mixed-effects regression models. The IOPGAT values were 2.4 and 1.4 mmHg higher than IOPNCT and IOPRBT, respectively; the IOPNCT was 1.0 mmHg lower than IOPRBT (p < 0.0001 for all comparisons). The IOPs measured by each tonometer were highly correlated with each other (r = 0.81–0.90, t = 45.2–65.5). The linear regression analyses showed that age was negatively correlated with IOPNCT (r = −0.12, t = −4.0) and IOPRBT (r = −0.14, t = −4.5) but not IOPGAT (r = 0.00, t = −0.2); the CCT was positively correlated with IOPGAT (r = 0.13, t = 4.3), IOPNCT (r = 0.29, t = 9.8), and IOPRBT (r = 0.22, t = 7.2). The mixed-effect regression models showed significant negative correlations between age and IOPNCT (t = −2.6) and IOPRBT (t = −3.4), no correlation between age and IOPGAT (t = 0.2), and a significant positive correlation between CCT and the tonometers (t = 3.4–7.3). No differences between IOPGAT and IOPRBT were seen at the age of 38.8 years. CCT affects IOPs from all tonometers; age affects IOPNCT and IOPRBT in different degrees. IOPRBT tended to be higher than IOPGAT in young subjects, but this stabilized in middle age and became higher in older subjects.


Introduction
Intraocular pressure (IOP) is the only known modifiable risk factor relevant to the treatment of glaucoma. Goldmann applanation tonometry (GAT) has been considered the "gold standard" for IOP measurement, although its readings are affected by central corneal thickness (CCT), corneal curvature, the modulus of corneal elasticity, and tear film [1]. Noncontact tonometry (NCT) using air-puff pressure has several favorable characteristics, including no corneal contact and no requirement for local anesthesia, which facilitates convenient use [1]. Rebound tonometry (RBT) uses the impact rebound principle by launching a magnetized probe against the cornea using a solenoid; the speed of deceleration of probe is measured and converted into the IOP [2]. There is no need for an air puff, corneal anesthesia, and slit-lamp mounting, and the measurement skill enables affordable, quick, and repeated IOP measurements even in children and very old patients [3].
During the routine use of the various tonometers in the clinic, we realized that RBT may yield higher IOP readings than GAT in young patients, while this scenario was reversed in older patients. To test our suspicion, we compared the IOP readings of GAT, NCT, and RBT and investigated the effects of age and CCT on the IOP readings in subjects who visited our glaucoma clinic.

Subjects
This retrospective study adhered to the tenets of the Declaration of Helsinki; the institutional review board (IRB) of Shimane University Hospital reviewed and approved the research. Based on the approval, written informed consent from each subject was waived; instead, the study protocol was posted at the study institutions to notify participants about the study. Among the 716 subjects who visited the glaucoma clinic of one author (MT) during April 2018 and March 2019, a review of the medical charts identified 1054 eyes of 544 subjects that fulfilled the inclusion criteria and were included in the analyses. The inclusion criteria were the measurement of IOPs using GAT (IOP GAT ), NCT (IOP NCT ), and RBT (IOP RBT ) on the same day and the recording of the CCT. In our glaucoma clinic, the IOPs obtained using the three different devices and CCT were recorded as routine examinations during the initial patient visit; most data collected were obtained at the initial visit; however, when multiple records of a subject were eligible, the most recent data were collected. No exclusion criterion was set in this real-world data analysis study; accordingly, all subjects who met the inclusion criteria were consecutively included irrespective of glaucoma or non-glaucoma, newly diagnosed or follow-up patients, treated or untreated, and the presence or absence of corneal and other eye diseases. Typically, one glaucoma specialist (MT) used the GAT and RBT (iCARE Rebound Tonometer TA01i, M.E. Technica, Tokyo, Japan) to record the IOPs. One of nine certified orthoptists in our department recorded the IOP using a non-contact air-puff tonometer (TonoRef III, Nidek, Aichi, Japan), and the CCT was recorded using a corneal pachymeter equipped in a specular microscope (EM-3000, Tomey, Nagoya, Japan). No pre-planned calibration of the tonometers was performed for this study.

Statistical Analysis
The differences among the IOPs assessed using the three tonometers were compared using the paired t-test. Possible correlations between three devices, their differences, i.e., NCT minus GAT (IOP NCT-GAT ), RBT minus GAT (IOP RBT-GAT ), and RBT minus NCT (IOP RBT-NCT ), age, and CCT were assessed by linear regression analyses. The effects of age and CCT on each tonometer were further assessed using a mixed-effects regression model in which each patient's identification number was regarded as a random effect, and both age and CCT were regarded as fixed effects. All continuous data were expressed as the mean ± standard deviation (SD). All statistical analyses were performed using the JMP version 11.0 statistical software (SAS Institute, Inc., Cary, NC). p < 0.05 was considered statistically significant.
The p values are calculated by using a paired t-test between each pair of tonometer groups. SD = standard deviation; CI = confidence interval. Finally, the effects of age and CCT on the IOPs measured by the three tonometers were assessed by mixed-effects regression models to adjust the interaction between age and CCT and cancel the bias resulting from the inclusion of both eyes of a subject (Table 3). Significant negative correlations were also seen between age and the IOP NCT (t = −2.6) and IOP RBT (t = −3.4), a non-significant correlation between age and the IOP GAT (t = 0.2), and significant positive correlations between the CCT and all three tonometers (t = 3.4-7.3) ( Table 3). The t and p values are calculated by mixed-effect regression models to adjust the interaction between age and CCT and cancel the bias resulting from the inclusion of both eyes of a subject. CI = confidence interval; r = Pearson's correlation coefficient.
We identified a significant negative correlation between age and IOP NCT or IOP RBT , while the correlation between age and IOP GAT was not significant ( Table 2). Since the IOP RBT-NCT was negatively correlated with age (Table 2), the impact of age is the greatest on the RBT among the tonometers tested. The absolute t-value was the largest for age with the IOP RBT-GAT (t = −7.6) among the models that included CCT and age ( Table 3), suggesting that age determines the difference in IOP readings between GAT and RBT more than CCT. A recent report has found a negative correlation between IOP RBT and age [17], and thus our results are in agreement with the previous report. Subject age and the detected difference between GAT and RBT readings in previous and current studies are summarized in Table 4. Including the current study, some studies have reported minus IOP RBT-GAT values [10,[12][13][14], while others have reported plus IOP RBT-GAT values [4][5][6][7][8]11,15,16]; this discrepancy is not fully explained by the difference in the CCT. Scatterplots of the subjects' ages and IOP RBT-GAT from previous studies (Table 4and Figure 1) clearly suggest the roles of age and IOP RBT-GAT . Previously, 0 IOP RBT-GAT was reported in subjects with a mean age of 59.3 years [9]. In the current subjects, based on linear regression analyses ( Figure S3E), the age of subjects with 0 IOP RBT-GAT was calculated to be 38.8 years. Thus, a lower/higher association of IOP readings between GAT and RBT is reversed based on the ages of the subjects. Other than the CCT, it has been proposed that corneal biomechanical properties such as corneal hysteresis (CH) and corneal resistance factors (CRF) affect the RBT and GAT differently [7,9,13]; both the CH and CRF decreased with aging [18]; thus, age-dependent changes in corneal biomechanical properties may be associated with our observation but need to be confirmed.  The limitations of the current study included the retrospective design and the inclusion of eyes with various types of glaucoma and glaucoma suspects. Because of the retrospective nature of the study, the methods of tonometry and examiners were not predetermined, although one examiner recorded the GAT and RBT using specific devices. We reviewed all patients who visited during the indicated period and included all patients who fulfilled the inclusion criteria, thus minimizing the selection bias. The inclusion of both eyes of a patient may have introduced bias, although we minimized this by using a mixed-effects regression model. Other than age and CCT, the modules of corneal elasticity [1] should affect the results. When the IOP elevates, the deviation between IOP NCT and IOP GAT becomes larger ( Figure S2A); this may be explained by the effect of changes in corneal elasticity. Despite the various backgrounds of subjects included and the retrospective study design, we believe that our real-world data analysis is reasonable to test our suspicion, described in the introduction section.

Conclusions
In conclusion, CCT affects the IOP readings of GAT, NCT, and RBT, while age affects the NCT and RBT by different degrees. The RBT readings tended to be higher than the GAT readings in young subjects, but this stabilized in middle age and was reversed in older subjects.

Informed Consent Statement:
The IRB approval did not require each patient to provide written informed consent for publication; instead, the study protocol was posted at the study's institutions to notify participants about the study.

Data Availability Statement:
The data are fully available upon reasonable request to the corresponding author.

Conflicts of Interest:
The authors report no conflict of interest in this work.