Accuracy and Reliability of Self-measured Intraocular Pressure in Glaucoma Patients Using the iCare HOME Tonometer

Précis: The iCare HOME tonometer is a useful tool which can be used by patients for fairly accurate and reliable self-measurement of intraocular pressure (IOP). Purpose: The aim was to compare the accuracy of IOP measurements of trained glaucoma patients using the iCare HOME tonometer before and after a week of self-monitoring at home and study the repeatability of patient measurements done at home. Materials and Methods: A total of 40 glaucoma patients had IOPs measured using the Goldmann applanation tonometer and the iCare HOME. Patients were trained to use the iCare HOME and measured their IOP twice daily over a loan period of 7 days. Bland-Altman analysis was used to assess agreement between Goldmann applanation tonometer and iCare HOME readings. Intraclass correlation coefficient compared inter-rater reliability of iCare HOME measurements, and test-retest variability for the patient’s readings over the 7-day period were analyzed. A 5-point Likert scale questionnaire was used to assess the patient’s experience with the device. Results: Up to 32 patients (80%) were able to obtain IOP readings within 3 mm Hg of the glaucoma nurse clinician’s readings at the preloan visit, and up to 36 patients (90%) at the postloan visit. Good inter-rater reliability was observed between iCare HOME measurements by the glaucoma nurse clinician and patients. At the postloan visit, inter-rater reliability was good for right eye readings and excellent for left eye readings. No distinct trend in test-retest variability was identified over the 7-day period. 65% of patients agreed the device was easy to use and 89% would recommend the device to other glaucoma patients. Conclusion: Patients trained to use the iCare HOME device can do so with fair accuracy compared with trainers with accuracy improving within 1 week of usage.

G laucoma encompasses a group of ocular disorders which result in optic neuropathy demonstrating characteristic visual field defects for which raised intraocular pressure (IOP) is a risk factor. To date, the mainstay of treatment for all glaucoma patients is reduction of IOP. Therefore, monitoring of glaucoma control invariably involves measurement of IOP. This is often done only by trained staff using noncontact tonometry or Goldmann applanation tonometry (GAT) within an eye clinic setting.
However, IOP measurement during a clinic consultation is only able to provide a reading which is not reflective of IOPs at all other points of time. Often, the ophthalmologist has to make clinical decisions such as setting target IOPs and assessing treatment responses based on the assumption that the IOP remains fairly constant throughout the day. However, diurnal variation in IOP is a well-documented fact with nocturnal increases in IOP possibly related to an increased episcleral venous pressure in the supine position. 1,2 Moreover the range of diurnal variation in patients with glaucoma has been found to be 2 to 3 times that of normal individuals. 3 Greater fluctuations in IOP have also been well documented to be related to progression of glaucoma. 1,[4][5][6] One strategy to obtain IOP measurements outside a clinic consultation involves patient self-monitoring of IOP. The iCare HOME is a handheld tonometer designed for such home tonometry. The tonometer uses the principle of rebound tonometry whereby a disposable, thin magnetized steel wire with a round plastic tip is bounced off the cornea and the impedance that is detected is used to calculate the IOP. 7 The small rounded tip minimizes the risk of corneal injury and being disposable, eliminates the risk of microbiological contamination. Furthermore, no topical anesthesia is required because of the very small size of the tip. The added portability of the device further affords patients the opportunity to self-monitor their IOP at home and at any time of the day. Compared with traditional techniques of obtaining diurnal tension curves, the iCare HOME may also be cost effective. 8 Publications have reported iCare HOME use under supervision in a clinic setting with iCare HOME tonometry slightly underestimating IOP by a mean of 0.3 9 to 0.8 10 mm Hg, or overestimating IOP by a mean of 0.7 mm Hg compared with GAT. 11 Despite the iCare HOME being meant for self-monitoring, little home usage patient data exists without the supervision in a clinic setting. One study in glaucoma patients measured IOP in both eyes on three consecutive days. They found that two-thirds of subjects had higher IOP in the morning than in the afternoon and nearly half had IOP patterns that differed on consecutive days. 12 However, questions remain about the reproducibility of patient's measurements during their learning curves and optimum durations of short-term usage to obtain good measurements.
The current clinical utility for iCare HOME tonometry includes determining variability of IOP, detection of peak IOPs outside of office hours, monitoring of treatment response and confirmation of absence of high IOPs when confirming the diagnosis of normal tension glaucoma.
Our study aims to compare the accuracy of IOP measurements of trained glaucoma patients using the iCare HOME tonometer before and after a week of selfmonitoring at home and study the repeatability of patient measurements done at home. Our secondary aim is to study the acceptability of the device by patients after a week of usage.

MATERIALS AND METHODS
The study was approved by the National Healthcare Group Domain Specific Review Board, which is overseen by the National Healthcare Group Research Ethics Committee. All patients provided informed consent and the study adhered to the tenets of the Declaration of Helsinki. Patients with a clinical diagnosis of glaucoma were recruited for the study and informed consent was taken from each patient. Exclusion criteria were very elderly patients above 80 years of age, patients with significant hand deformities, weakness or tremors which would result in difficulty handling the iCare HOME device, patients who were unable to understand or remember instructions, and those with visual acuity poorer than 6/60. Baseline measurements of best corrected Snellen visual acuity, central corneal thickness (CCT) and autorefraction were obtained.
At the preloan visit, patients had their IOPs for both eyes measured by a trained glaucoma nurse clinician using the GAT, following which a measurement using the iCare HOME (iCare Finland Oy, Vantaa, Finland) was taken by the same nurse clinician. Each patient was then taught by the nurse clinician how to use the iCare HOME device for self-measurement of IOP.
Training initially focused on being able to set up, handle and troubleshoot the device. Thereafter, emphasis was placed on positioning the probe correctly at the central cornea and being able to meet manufacturer IOP validation criteria. When the patient was successful in obtaining a valid reading with the device, those readings for both eyes were recorded.
The iCare HOME device was then loaned to the patient for a period of 7 days. During that period, the patient was instructed to record their IOPs for both eyes twice a day, with 3 separate readings taken at each time point for each eye.
The postloan visit was scheduled at the end of the 7 days. At this visit, the IOPs for both eyes were again taken first by the nurse clinician using the GAT and the iCare HOME. In addition, iCare HOME readings taken by the patients themselves were recorded. All iCare HOME readings were downloaded using the iCare LINK software and analyzed.
Finally, a 5-point Likert scale questionnaire was administered at the postloan visit to assess the patient's acceptance and perception of the iCare HOME device.
Statistical analysis was performed using SPSS 21.0 (SPSS Inc, Chicago, IL). Statistical significance was defined by P-value <0.05. Bland-Altman difference analysis was used to assess the agreement between GAT and iCare HOME readings taken by the glaucoma nurse clinician at both preloan and postloan visits, while intraclass correlation coefficient (ICC) assessed the inter-rater reliability of iCare HOME measurements by the glaucoma nurse clinician and patients. The test-retest variability for the patient's readings over the 7-day period were analyzed by taking the SD of the 3 readings taken at each time point per day with 2 readings taken each day over the 7 days.
Univariate regression analysis was carried out to explore the effect of patient demographic and clinical factors on the accuracy of IOP measurements.
ICC estimate values <0.5, between 0.5 and 0.75, between 0.75 and 0.9, and > 0.9 are indicative of poor, moderate, good, and excellent reliability/agreement respectively. A minimum of 29 participants is required to detect a difference between 0.6 (moderate) and 0.81 (good) reliabilities with a power of 0.8 and a significance level of 0.05 for 3 repeated measurements. We recruited 40 subjects to ensure adequate sample size for analysis even with a 20% dropout rate.

RESULTS
A total of 40 patients were recruited, 18 female and 22 male, with a median age of 58 years (range: 21 to 73 y). The time taken to teach each patient how to obtain a valid reading with the iCare HOME ranged from 15 to 75 minutes.
The mean log-MAR visual acuity was 0.16 ± 0.25 in the right eye, and 0.23 ± 0.41 in the left eye. Mean CCT readings were 557.58 ± 27.92 micrometers for the right eye, and 556.88 ± 27.91 µm for the left eye. Five eyes (3 right and 2 left) had CCT measurements above 600 µm, while 3 eyes (1 right and 2 left) had CCT readings of < 500 µm. IOPs ranged from 9 to 22 mm Hg in the right eye and 8 to 23 mm Hg in the left eye.
Right hand dominance was reported by 36 patients (90%) while the remaining 4 patients (10%) reported left hand dominance. Most of the patients operated the iCare HOME using their right hand for their right eye and their left hand for their left eye during the preloan and postloan visits. The glaucoma nurse clinician who is left-handed used her dominant hand when taking iCare HOME measurements on the patient's left eye, and her nondominant hand when taking measurements on the patient's right eye. We were not able to obtain accurate detailed information on the patients' use of the dominant hand versus the nondominant hand during the loan period.
Bland-Altman plots were used to assess the agreement between the GAT and iCare HOME measurements recorded by the glaucoma nurse clinician at both the preloan and postloan visit. At the preloan visit, the mean IOP by GAT was 14.1 ± 3.0 mm Hg for the right eye and 14.5 ± 3.2 mm Hg for the left eye. iCare HOME readings were generally slightly higher for both eyes compared with GAT, with a mean bias of 1.38 mm Hg (95% limits of agreement −7.27 to 4.52) in the right eye, and 1.48 mm Hg (95% limits of agreement −8.89 to 5.94) in the left eye ( Figs. 1 and 2).
At the postloan visit, the mean IOP by GAT was 13.9 ± 3.4 mm Hg for the right eye and 14.7 ± 4.1 mm Hg for the left eye. The mean bias in the right eye was higher at 2.08 mm Hg (95% limits of agreement −8.48 to 4.33), but the left eye was lower at −1.25 mm Hg (95% limits of agreement −7.05 to 4.55) (Figs. 3 and 4).
At the preloan visit, the glaucoma nurse clinician obtained iCare HOME readings within 3 mm Hg of GAT readings in 31 patients (78%) for the right eye, and 25 patients (63%) for the left eye. At the postloan visit, this remained fairly similar at 29 patients (73%) for the right eye but improved to 31 patients (78%) for the left eye.
The glaucoma nurse consistently used her dominant hand to take iCare HOME readings on the patient's left eye, and her nondominant hand for the right eye. At the preloan visit, the mean difference between iCare HOME and GAT readings was −1.37 ± 3.01 in the right eye and −1.47 ± 3.78 in the left eye. At the postloan visit, the mean difference was −2.08 ± 3.27 for the right eye and −1.25 ± 2.96 for the left eye. There was no significant difference comparing the mean difference between the right and left eyes at both the preloan visit (P = 0.875) and the postloan visit (P = 0.170).
The mean iCare HOME readings taken by the nurse clinician were 15.4 ± 4.1 mm Hg for the right eye and 16.0 ± 5.8 mm Hg for the left eye at the preloan visit. At the postloan visit, the mean readings were 16.0 ± 5.0 mm Hg for the right eye and 15.9 ± 5.8 mm Hg for the left eye.
The mean iCare HOME readings taken by the patients were 16.5 ± 5.3 mm Hg for the right eye and 16.3 ± 5.0 mm Hg for the left eye at the preloan visit. At the postloan visit, the mean readings were 16.7 ± 5.0 mm Hg for the right eye and 16.4 ± 6.0 mm Hg for the left eye.
At the preloan visit, 29 patients (73%) were able to obtain iCare HOME readings within 3 mm Hg of the glaucoma nurse clinician's iCare HOME readings on their right eye, and 32 patients (80%) on their left eye. At the postloan visit, this dropped slightly to 27 patients (68%) for the right eye but increased to 36 patients (90%) for the left eye.
On univariate analysis, there was no statistically significant association between the accuracy of the patient's FIGURE 1. Bland-Altman plot of agreement between intraocular pressure measured by the glaucoma nurse clinician using Goldmann applanation tonometry (GAT) and iCare Home (iCareN) for the right eye at the preloan visit. The x-axis: mean of intraocular pressure measurements by iCare Home and GAT. The y-axis: difference between intraocular pressure measurements by GAT and iCare Home. Figure 1 can be viewed in color online at www.glaucomajournal.com.
iCare HOME readings and age, sex, CCT, visual acuity or severity of glaucoma.
Two-way mixed, single measure ICC was used to assess agreement between iCare HOME readings by the glaucoma nurse clinician and the patient's readings. The ICC between iCare HOME readings obtained by the glaucoma nurse clinician and the patients at the preloan visit was 0.748 [95% confidence interval (CI): 0.567-0.859] for the right eye, and 0.840 (95% CI: 0.719-0.912) for the left eye. ICC at the postloan visit was 0.796 (95% CI: 0.648-0.886) for the right eye, and 0.908 (95% CI: 0.834-0.950) for the left eye (Table 1).
SDs were computed for the 3 readings at each time point (2 sets of readings per day) for each eye over the 7 days of use. Repeatability was assessed by within-subject SD (Sw). The differences between repeated measurements for the same subject is expected to be less than the specific values (under repeatability) for 95% of the subject. The repeatability was fairly constant at 3.60 to 6.06 across the 7 days of IOP readings after removing the outliers, <5 and > 50.
From questionnaire responses 79% of the patients strongly agreed or agreed that the iCare HOME device is useful for checking their own IOPs, and 65% of patients agreed the device is easy to use although only 33% could get the reading at their first attempt. 95% strongly agreed or agreed that using the device was painless and 85% agreed that it is safe. Although 32.5% strongly agreed or agreed that the device is inconvenient to use, 87% of them would recommend the device to other glaucoma patients.   Bland-Altman plot of agreement between intraocular pressure measured by the glaucoma nurse clinician using Goldmann applanation tonometry (GAT) and iCare Home (iCareN) for the left eye at the postloan visit. The x-axis: mean of intraocular pressure measurements by iCare Home and GAT. The y-axis: difference between intraocular pressure measurements by GAT and iCare Home. Figure 4 can be viewed in color online at www.glaucomajournal.com.

DISCUSSION
In our study, we compared the accuracy of IOP measurements of trained glaucoma patients using the iCare HOME tonometer before and after a week of selfmonitoring at home using an experienced glaucoma nurse clinician as a reference standard. We found that the correlation between readings of the glaucoma patient and nurse clinician readings improved from moderate to good (right eye) and from good to excellent (left eye) after the 7-day loan period. This may indicate that there is a modest learning curve from a 1 week period of usage. It is, however, unclear what factors accounted for the better accuracy with increased usage. It might be suspected that it is because of better ocular alignment with increasing familiarity with the device. It is also uncertain whether the modest gain in accuracy with short-term use is clinically meaningful enough to recommend disregard of early IOP readings with the device.
Previous studies looking at first session of iCare HOME use found high inter patient reliability of measurements. 13 We assessed the reliability of patient IOP measurements over a 7-day period and found no significant trend in test-retest variability across the 7 days of use. Therefore, measurement reliability is established early despite the apparent improvement in accuracy with short-term use.
We did find a difference in the agreement in IOP measurements between glaucoma patients and a glaucoma nurse clinician both using iCare HOME versus the agreement of glaucoma patients using iCare HOME compared with GAT. The former showed better agreement which is in keeping with Realini et al's 14 report of better intraoperator reproducibility of ICare model ic100 measurements compared with Goldmann measurements. They attribute this to Goldmann tonometry having many sources of error that vary over time making it unpredictable as opposed to the iCare rebound tonometer.
This may have implications for judging the success of training to use the device. One study had a criterion where the first of 3 iCare HOME measurements obtained by the patient had to be within 5 mm Hg of the measurement obtained with GAT. 13 This resulted in 16% of their subjects failing to meet certification to use the device. The manufacturer's IOP validation guidelines do not compare IOP readings of subject to GAT but instead compares IOP readings by iCare HOME between trainer and subject. In another study using manufacturer's guidelines, failure to meet IOP validation criteria was 5.2%. 9 On the basis of our study findings where the agreement of glaucoma patients using iCare HOME compared with GAT is lower than agreement to a trained user, we suggest using the manufacturer's guidelines which do not consider agreement with GAT for certification.
We compared iCare HOME and GAT readings taken by the glaucoma nurse clinician at the first and second visits. 63% to 78% of iCare HOME measurements were within 3 mm Hg of GAT. This is in agreement with other studies that have compared iCare HOME to GAT where ∼70% of iCare HOME measurements were within 3 mm Hg of GAT measurements in healthy subjects 15 and in another study up to 93% were within 5 mm Hg involving glaucoma patients. 13 We found, however, that the iCare HOME consistently slightly overestimated IOP compared with GAT. In this we were similar to Takagi et al 11 but differed from others showing an underestimation. 13,15,16 We also explored other factors which could possibly have an impact on the patient's ability to use the iCare HOME device accurately such as age, sex, highest educational level, visual acuity, severity of glaucoma, and CCT. On univariate analysis, there was no statistically significant impact of age, sex, severity of glaucoma, or visual acuity on the accuracy of readings.
Hand dominance did not appear to affect the accuracy of the glaucoma nurse clinician's iCare HOME readings compared with GAT. The glaucoma nurse clinician consistently used her dominant hand to take iCare HOME measurements for the patient's left eye and her nondominant hand for the right eye, and there was no significant difference between the 2 eyes when compared with GAT readings. This is in agreement with another study which also suggested that hand dominance did not affect the accuracy or reliability of measurements. 16 Our secondary aim was to study the user experience after a week of iCare HOME usage. Overall, the device was well received by patients. The majority agreed that measurements were painless, the device was safe, and it was easy to use. In terms of safety, user experience concurred with more objective evidence. 13 For ease of use, we found 65% of patients agreed with the statement but this was lower than the 93% reported in another study. 9 A possible reason for the difference is that we administered the questionnaire at the end of 7 days of use whereas the other study administered the questionnaire after the training session. Our results were however closer to another training session questionnaire result where 71% felt it was easy to use. 16 Looking at the clinical use proposition, the majority of glaucoma patients found it useful for checking their own IOPs and would recommend the device to other glaucoma patients. However, less than half found it convenient to use, perhaps explaining withdrawals and noncompliance found in 1 other study. 13 The strengths of this study are that we looked at patient accuracy and reliability beyond a single clinic training session to 1 week of usage. This was useful to evaluate the learning curve of the device. We also included patients with visual acuity up to 6/60 making our data more generalizable