Glaucoma Home Self-Testing Using VR Visual Fields and Rebound Tonometry Versus In-Clinic Perimetry and Goldmann Applanation Tonometry: A Pilot Study

Abstract

Purpose: This pilot study aimed to assess the feasibility, accuracy, and repeatability of unsupervised, at-home, multi-day glaucoma testing using the Olleyes VisuALL Virtual Reality Platform (VRP) and the iCare HOME handheld self-tonometer.

Methods: Participants were trained to use two U.S. Food and Drug Administration–registered or approved devices before conducting self-tests at home over 3 consecutive days. The iCare HOME intraocular pressure (IOP) measurements were collected four times daily per eye, and VRP visual field tests were performed once daily. The results were compared with one in-clinic Humphrey Field Analyzer (HFA) visual field test performed on the day of device training, iCare HOME measurements by the trainer, and the last five Goldmann applanation tonometer (GAT) results.

Results: Of 15 enrolled participants, nine of them (60%) completed the study. The six excluded participants could not self-measure using iCare HOME. There was significant correlation between the average mean deviation (MD) values of the at-home VRP tests and in-clinic HFA test (r² = 0.8793, P < 0.001). Additionally, the average of the sensitivities in five of six Garway–Heath sectors were significantly correlated. VRP test duration was also shorter than in-clinic HFA testing (P < 0.001). Finally, at-home tonometry yielded statistically similar values compared to trainer-obtained iCare HOME values. The mean and range of at-home tonometry were also statistically similar to those for in-clinic GAT, but at-home tonometry demonstrated higher maximum IOP values (P = 0.0429).

Conclusions: Unsupervised, at-home, multi-day glaucoma testing using two devices resulted in the capture of higher maximum IOPs than in the clinic and good MD correlation of VRP with HFA. However, 40% of participants could not self-measure IOP using iCare HOME.

Translational Relevance: The study findings suggest that at-home remote glaucoma monitoring correlates with in-office testing and could provide additional information for glaucoma management, although patients had more difficulty with the iCare HOME than the VRP.

Introduction

Glaucoma, a leading cause of irreversible blindness worldwide, requires regular in-office visual field tests and intraocular pressure (IOP) measurements to guide treatment decisions and prevent functional visual field loss.^1,2 The American Academy of Ophthalmology's Preferred Practice Patterns recommends annual visual field testing and at least one tonometry test per year for stable glaucoma, although some patients are seen every 3 to 6 months or even more frequently.^3,4 Increasing the number of visual field tests performed per year helps to establish a reliable baseline and detect glaucoma progression sooner, given the high subjective variability of perimetry.^5–8 However, multiple in-office visual fields and IOP checks can be impractical and costly due to insurance coverage or travel limitations.^7,9–11 Moreover, single in-office tonometry readings may miss unacceptable peaks in eye pressure or periods of suboptimal control, as IOP fluctuates throughout the day.¹² 

The COVID-19 pandemic expedited the adoption of telemedicine and at-home testing for glaucoma management.^13–15 This shift was facilitated not only by the emergency use authorization to reimburse video visits on par with office visits but also by the U.S. Food and Drug Administration (FDA) existing regulatory pathways that allow for the use of devices similar to those already approved.^13,15,16 Consequently, glaucoma home monitoring devices, such as head-mounted virtual reality perimeters and self-administrable tonometers, gained popularity within the private sector.^16–20 

Teleglaucoma has long been useful for glaucoma screening and monitoring remote patients.^15,17,20 However, the recent introduction of remote physiologic monitoring (RPM) and remote therapeutic monitoring (RTM) codes made remote patient monitoring more financially feasible.^21–24 During the pandemic, glaucoma clinics experimented with managing select glaucoma patients remotely using portable devices for virtual reality visual field testing and home tonometry with either FDA 510(k) clearance or approval.^19,25–30 Gold-standard in-office applanation tonometry and static automated perimetry machines could also be combined with remote physician video visits, but this still required test-only technician in-office visits. 

Various tools and platforms for conducting remote visual field testing have been evaluated in the literature. These include tablet-based, web-based, and head-mounted perimeters. These web- and screen-based tools have been successfully used to detect glaucoma, with a sensitivity of glaucoma detection in the literature ranging between 54% to 100% at home across all severities.^31–35 However, many of these tools are difficult to standardize, utilize a range of different hardware, and have not been studied longitudinally. The iCare HOME handheld tonometer (iCare Finland Oy, Vantaa, Finland) has been extensively studied and has been FDA approved for out-of-office use, with data supporting detection of IOP elevation not measured during clinic visits that may supplement clinic data to change management.^{29,30,36–40} Given that the healthcare environment is beginning to financially support remote chronic disease monitoring, we designed and conducted a pilot study with glaucoma patients to assess the feasibility, repeatability, and validity of unsupervised, at-home visual field testing using the Olleyes VisuALL Virtual Reality Platform (VRP; Olleyes, Inc., Summit, NJ), along with iCare HOME self-tonometry. 

Methods

Ethics Approval and Informed Consent

This study was conducted in accordance with the tenets of the Declaration of Helsinki and received approval from the Institutional Review Board at Stanford University. All participants provided digital informed consent before participating in the study. 

Participant Recruitment and Eligibility

Participants were recruited from the clinics of three glaucoma providers at the Byers Eye Institute at Stanford University from September 2021 through February 2022 during their routine office visits. Inclusion criteria were age >18 years old, English-speaking, a known diagnosis of glaucoma with a perimetric field defect of any severity, binocular vision with best-corrected acuities of 20/30 or better, and the ability to complete the initial training on VRP perimetry and iCare HOME tonometry before being sent home with the devices. Patients with corneal, retinal, or other optic nerve or visual pathway diseases that could affect the visual field; cognitive decline affecting ability to follow directions; history of poor compliance; or intraocular surgery scheduled within the study period or within 6 months prior to the study were excluded. Participants were required to have sub-25% false positives and sub-25% fixation losses on their most recent in-clinic Humphrey Field Analyzer 3 (HFA3; Carl Zeiss Meditec, Dublin, CA) visual field test using the 24-2 SITA Standard algorithm. Finally, patients who would not be able to return the devices to the clinic at the end of their testing period were excluded. 

Clinic Baseline Testing

Participants were recruited during their normal office visits, when they performed HFA 24-2 SITA Standard visual field tests on the same day for each enrolled eye. Clinical judgment was used to select patients with stable glaucoma and a history of reliable testing. Tonometry measurements were recorded using the Goldmann applanation tonometer (GAT). Best-corrected visual acuity was checked with the Snellen chart. Participants underwent in-person training following the visit, before being sent home with the two devices. 

VRP Home Perimetry

The VRP is a commercially available virtual reality headset, described in detail in previous studies.^41,42 Briefly, the VRP uses a head-mounted device (HMD) paired to a Bluetooth remote control clicker. A web-enabled device is needed for test management and administration. Within the headset, there are two independent 4K resolution displays, allowing each eye to be presented with stimuli separately with a field of view up to 100°. This eliminates the need for eye obstruction and allows both eyes to be randomly evaluated in one session, unlike the HFA. The device requires Wi-Fi to function, as tests are remotely “pushed” to the device, and all results are saved to the Olleyes Health Insurance Portability and Accountability Act (HIPAA)-compliant online platform. Test progress can be monitored remotely in real time if needed, and results are viewable by the ordering provider any time after testing. The VRP includes an optional instructional video in which a virtual assistant shows participants a preview of the test and the stimulus used. The same virtual assistant tracks the performance of the test and provides verbal feedback to minimize fixation losses and false positives. The device prompts users to fixate on a central point on the displays within the headset, relying on the Heijl–Krakau tracking method to monitor for fixation errors by periodically placing a stimulus in the blind spot and analyzing the participant’s response. In this study, we utilized the proprietary threshold strategy of the VRP, which accounts for the scotopic background and the narrower luminance range of the device, in order to best mimic the HFA's 24-2 SITA Standard test. 

For home application, a user turns on the device and connects it to Wi-Fi. The VRP then immediately checks for any pending tests and prompts the user to begin the test by pulling the trigger on the remote control. The VRP features a 3500-mAh lithium battery capable of powering the device for about 3 hours of use before it must be recharged. The VRP can also be used as an AC-powered perimeter and therefore can be used while charging. 

All study participants were trained to use the VRP in the clinic (by ARB). Detailed training included instructions on how to power on and off the device, pair and use the accompanying remote control, wear the device properly, connect the device to Wi-Fi, charge the device, and accept and complete tests pushed to the device wirelessly by the research team. Training was completed over the course of 30 minutes. Training was considered successful if participants were able to, within 30 minutes and without assistance, power on the device, wear it, begin the pending 24-2 visual field test sent by the research team, view the instructional video within the headset, and complete the test with sub-25% false-positive check and sub-25% fixation losses. Participants’ birth years were used to obtain age-adjusted results. Participants used their own spectacles within the headset. 

iCare HOME Tonometry

The iCare HOME handheld tonometer is a second-generation device that uses a disposable probe to measure eye pressure via rebound tonometry (of note, the iCare HOME2 was released after completion of this study). Rebound tonometry is a well-established method to measure IOP, correlating with Goldmann applanation tonometry and offering the advantage of not requiring anesthetic.^{36,37,39,43–51} The device has only two buttons, one for power and one for taking measurements. It stores results locally until it is plugged into a computer or smartphone equipped with the iCare software, which then uploads the results to the online platform and deletes the data on the device. The iCare HOME has no wireless connectivity or a display; therefore, results can only be seen when it is connected to the iCare software. The iCare HOME utilizes long-lasting disposable batteries and does not require charging. Single-use sterile probes are loaded into the device before each use. The tonometer has an adjustable forehead and cheek rest that guides proper placement in relation to the cornea. During initial training, the position of these rests is typically optimized for each participant and can be easily adjusted by the patient if needed. When the device is correctly aligned parallel to the ground, a visible ring light turns green (red if misaligned). At this stage, the patient, while seated upright, can either press the measure button six times or hold it down to enable the device to capture six quick measurements. The device will then discard the highest and lowest values and average the remaining measurements to produce a single measurement value. The iCare HOME sensors detect which eye is being measured and gauge the accuracy of the bounce of the tip off the center of the cornea. In the event of acquisition errors, a series of beeps of varying lengths can assist during the troubleshooting process. Common problems include positioning the device either too close or too far away from the eye or failing to position the device parallel to the ground. 

Participants were trained (by ARB) to use the iCare HOME in line with the manufacturer's recommendations at the same clinic visit as the VRP training. Participants were taught how to power the device, load a probe, position the device on their face, and collect reliable measurements. Participants were also guided to take right-eye measurements with their right hand, and vice versa, while obscuring the fellow eye with the other hand. Participants were considered trained if they were able to get a measurement within 5 mm Hg of the trainer's measurement on their first attempt. Also, over the course of three measurements, the range of their results had to be within 7 mm Hg. These criteria are iCare's recommendations. Finally, the trainer must confirm that the placement of the device is correct. All three measurements had to be completed within 30 minutes or less. 

At-Home Testing

After successful training, participants were instructed to complete one VRP visual field test and four IOP tests per day at home over 3 consecutive days. Before leaving the training session, participants scheduled 3 consecutive days to collect measurements at home, as well as the return date to drop off the devices at the clinic. Regarding tonometry, patients were instructed to take measurements upon waking, before lunch, before dinner, and before bed per eye enrolled in the study. Equipment was generally returned within 1 to 2 weeks. The research team did not issue reminders or check-ins unless participants missed their scheduled dates to return the devices. Failure to complete the data collection on 3 consecutive days within 1 week after enrollment resulted in exclusion from the study. Participants were able to call members of the research team or device customer service for technical support if needed. Video support by a local representative for the iCare HOME was also available. 

Data Collection and Statistics

VRP visual field reports were downloaded from Olleyes’ online cloud storage, whereas the iCare HOME devices were connected to the iCare client via USB to upload collected data and clear the device. Time-stamped tonometry data were downloaded from the iCare online client. PDF reports of the visual field results for VRP and PDF reports of the visual field results for both VRP and HFA were scanned with a custom script and manually checked to extract the numerical sensitivity plot values for each eye. Statistical analysis was conducted using SPSS Statistics 28.0.1.1 (IBM, Chicago, IL). 

Visual field sensitivity data for each of the 54 points were grouped into sectors corresponding to the Garway–Heath visual field map, which identifies six distinct regions corresponding to structures on the optic disc.⁵² The means of the sensitivity values for each of the six Garway–Heath sectors were calculated for both HFA and VRP visual field reports for every eye in the study. The average values for each sector across the three VRP tests were compared to the sector values obtained in the clinic using the HFA by calculating the Spearman correlation for each sector. 

To compare the mean deviation (MD) and pattern standard deviation (PSD) values across the two devices, the average MD and PSD values across the three VRP tests for a given eye were compared to the values from the single in-clinic HFA using the Spearman correlation. A pointwise comparison for each of the 54 points per eye was done by taking the average of the three VRP tests and comparing them to the HFA using the Wilcoxon signed-rank test. 

The IOP obtained during the at-home period for each eye was compared to the IOP measurements obtained by ARB with the iCare HOME during the training session and by GAT from the last five in-clinic visits using Spearman correlation and the Wilcoxon signed-rank test. 

Finally, in an attempt to address inter-eye correlation of IOP and VF defects, two linear mixed-effects models (LMMs) were generated to compare Garway–Heath sector and MD values obtained from the HFA and VRP. A third LMM was used to compare average IOP values obtained from the iCare HOME self-administered at home and the GAT measured in the clinic. 

For each LMM, the device type (for Garway–Heath sectors and MD, VRP and HFA; for IOP, iCare HOME and GAT) was treated as a fixed effect. A random intercept for each participant was included to account for the inter-eye correlation, acknowledging that most participants contributed measurements from both eyes. 

The models were fitted using restricted maximum likelihood (REML), and the significance of the fixed effects was evaluated. For each sector and parameter, we extracted the estimated coefficients, standard errors, z values, P values, and 95% confidence intervals to estimate the magnitude and significance of the differences in measurements between the devices. 

Results

A total of 28 eyes from 15 participants (73.3% male), with a mean age of 60.2 ± 16.4 years, were enrolled in the study. Of these, 16 eyes from nine participants successfully completed the study, with two participants only having one eye diagnosed with glaucoma. The six participants who were unable to complete the study reported difficulties using the iCare HOME device despite training. Four of the 15 participants could not be trained to collect eye pressure measurements in the clinic using the iCare HOME due to incorrect placement of the device on the face. Two of the 15 participants were successfully trained but could not complete 3 days of data collection at home due to difficulty using the device. The most cited issues were difficulty positioning the device on the face and difficulty interpreting the error tones, which made troubleshooting frustrating for participants. No participants were excluded due to difficulty with the VRP, and no participants met the threshold for exclusion due to high false positives or fixation losses; all but one participant maintained sub-15% false positives and sub-15% fixation losses. 

In this cohort, 12 eyes exhibited advanced glaucomatous visual field changes with either superior and inferior hemifield involvement (six eyes) or one hemifield with fixation involvement (six eyes). An additional three eyes had defects confined to one hemifield. One eye had a visual field within normal limits. The average MD value for the HFA tests was –10.16 ± 6.94 dB (range, –2.16 to –27.72), and the average IOP from Goldmann applanation tonometry was 16 ± 5.18 mm Hg (range, 9 to 42). Six of these patients were prescribed two different glaucoma eyedrops, and three patients had three glaucoma eyedrops. Average logMAR was 0.0375 ± 0.0592, and average spherical equivalent was −3.13 ± 4.02 D, with 18.75% of the eyes in the study having undergone some type of glaucoma surgery longer than 6 months before enrollment into the study. 

Perimetry

Among the 16 eyes that completed the study, the mean of the three MD values from the at-home VRP visual field testing significantly correlated with the MD value from the single in-clinic HFA test, with a correlation of 0.8793 (P < 0.001, Spearman correlation) (Fig. 1A). Figure 1B presents a Bland–Altman plot demonstrating agreement between the HFA and VRP. The intraclass correlation coefficient (ICC) and concordance correlation coefficient (CCC) for the MD values from both devices were 0.947 and 0.843, respectively (Table 1). 

Figure 1.

 

Agreement between the HFA versus VRP mean deviation. (A) Scatterplot showing the MD values from the in-clinic HFA test plotted against the mean of three at-home tests using the VRP. Each point represents an eye enrolled in the study. Correlation was calculated using Spearman's rho. (B) Bland–Altman plot. Horizontal dashed lines correspond to the 95% limits of agreement, and the solid line is the bias.

Table 1.

 

Correlation Between MD and PSD Values from the In-Clinic HFA Test and Mean of Three At-Home VRP Tests

 

To perform a more detailed regional comparison, Garway–Heath sectors were utilized to compare specific visual field areas for each eye, as shown in Figure 2. The average of the sensitivities within the six resultant subsections for each visual field were then compared between the HFA and VRP (Table 2). 

Figure 2.

 

A representation of Garway–Heath sectorization of a right visual field. Black denotes the blind spot. The labels are based on their corresponding optic nerve sectors.

Table 2.

 

Correlation Between Mean Sensitivities of Garway–Heath Sectors From In-Clinic HFA Testing and Average of Three At-Home VRP Tests

 

Five of the six sectors had a significant correlation between the in-clinic HFA and the average of three at-home VRP visual field tests, with the highest correlation being in the inferotemporal region (Table 2). The only sector that was not significantly correlated was the superonasal sector. 

The time taken for each eye to complete each visual field test was recorded and compared, as well (Fig. 3). The VRP at-home tests were found to be significantly shorter (P < 0.001) than the in-clinic HFA tests, with a median difference of 69.33 seconds. 

Figure 3.

 

Violin plot comparing the duration of VRP and HFA visual field testing per eye. Time is measured in seconds. Quartiles are marked by red dotted lines, and the median is marked by a black solid line. The Wilcoxon matched-pairs signed-rank test yielded a significant difference (P = 0.0002).

A pointwise comparison for each eye was conducted to compare each of the 54 sensitivity values within each visual field (Fig. 4). The average sensitivity values obtained per point over the course of 3 days at home were compared to the sensitivity values obtained from one HFA test in the clinic. This comparison revealed an r² value of 0.775 and demonstrated that the VRP tended to overestimate sensitivity values relative to the HFA. Manual examination of each visual field test confirmed this; in four eyes in the study, at least two of the three VRP tests did not show defects that were shown in the HFA report, and on one occasion no defects were seen on the first and third VRP tests but the second VRP test reproduced the defect on the HFA. Visually, correlation of sensitivity values between devices appeared to decrease below 20 dB; however, the Spearman correlation below 20 dB proved to be higher than above 20 dB. 

Figure 4.

 

A pointwise comparison of the average sensitivities obtained from three at-home VRP tests compared to one in-clinic HFA test. Two lines of best fit were calculated, one for HFA values below 20 dB and one for values above 20 dB. Data points are color coded by Garway–Heath sector.

Over the course of 3 consecutive days of testing, repeatability of the VRP was demonstrated with significant ICC across all six Garway–Heath sectors (Table 3). The superonasal sector had the highest ICC value at 0.97, although it was also the least correlated with HFA (Table 2). Conversely, the temporal sector had the lowest ICC value. MD values had an excellent ICC of 0.90, indicating good repeatability of measurements over time. 

Table 3.

 

ICC of VRP Testing Across 3 Days Split by Garway–Heath Sectors

 

LMMs comparing average Garway–Heath sector sensitivities obtained from the HFA and VRP, as well as MD values, were generated to account for inter-eye correlation (Table 4). The superonasal sector exhibited a coefficient of 4.000 (P = 0.027), and the inferotemporal sector had a coefficient of 4.013 (P = 0.024), indicating significantly higher measurements with the VRP. The other sectors showed non-significant differences, but the direction of the coefficients suggested a trend toward higher measurements with the VRP. Analysis of MD values revealed a non-significant difference between the devices, with a small coefficient of 1.685 (P = 0.259), suggesting agreement with a trend for VRP to yield slightly higher MD values. 

Table 4.

 

Linear Mixed-Effects Models Comparing Average Garway–Heath Sector Sensitivities and MD Values Obtained from HFA and VRP

 

Tonometry

The IOP values obtained for each eye at home were compared to those obtained by the iCare HOME during the training session by the trainer and by GAT performed by clinic providers from the last five visits preceding the study. In our study, the central corneal thickness (CCT) for the right eye had a mean value of 550.0 µm with a standard deviation of 27.34 µm, ranging from 518 to 598 µm. Similarly, the left eye exhibited a mean thickness of 546.44 µm with a standard deviation of 21.51 µm, ranging from 522 to 594 µm. The Spearman correlation between the in-clinic iCare HOME IOP measurements collected by ARB and the average at-home measurement taken at the time of day nearest to the in-clinic measurement was 0.871 (95% CI, 0.652–0.956; P < 0.001). 

Spearman's r for the mean pressures obtained at home versus GAT in the clinic was 0.6284 (95% CI, 0.2144–0.8510; P = 0.0052). ICC3 for the mean pressures at home versus GAT in the clinic was 0.863 (95% CI, 0.633–0.949; P < 0.001), indicating a high degree of agreement between measurements from the iCare HOME and GAT. 

Twelve of 16 eyes had higher maximum values from home tonometry than GAT. The at-home testing yielded statistically significant higher maximum values compared to the last five in-clinic visits, with a Wilcoxon matched-pairs signed-rank test yielding P = 0.0429 (Fig. 5). The differences (maximum home IOP – maximum GAT IOP) ranged from –5 to 9 mm Hg in all patients. 

Figure 5.

 

The mean and maximum IOP values obtained over 12 measurements in 3 days of at-home testing compared to the mean and maximum values obtained in the last five visits to the clinic. Quartiles are marked by red dotted lines, and the median is marked by a black solid line. The Wilcoxon matched-pairs signed-rank test yielded a significant difference for maximum values (P = 0.0429), with a median difference of 3.5 mm Hg. There was no significant difference in the mean or range between the GAT IOP values obtained in the last five clinic visits and after 3 days of at-home tonometry, with P = 0.5874 and P = 0.6293 for the Wilcoxon matched-pairs signed-rank test, respectively.

At-home pressure measurements typically peaked in the morning, with 37.5% of maximum measurements occurring between 8 AM and 12 PM, 18.75% occurring between 12 PM and 4 PM and 4 PM to 8 PM, and 25% occurring between 8 PM and 12 AM. Nadirs typically occurred at night, with 43.6% of nadirs occurring between 8 PM and 12 AM, 25% occurring between 8 AM and 12PM, 12.5% occurring between 12 PM and 4 PM, and 18.75% occurring between 4 PM and 8 PM, with a mean difference between peaks and nadirs of 8.22 mm Hg. Diurnal variation testing was not done with GAT; however, the mean difference between peaks and nadirs over the last five clinic visits was 7.67 mm Hg. There was no significant difference in the mean or range between the GAT IOP values obtained in the last five clinic visits and after 3 days of at-home tonometry, with P values of 0.5874 and 0.6293 for the Wilcoxon matched-pairs signed-rank test, respectively. 

LMM analysis comparing IOP obtained from the iCare HOME to the GAT revealed a non-significant effect of the device type on IOP readings. The estimated difference (coefficient) in IOP measurements between devices was 0.422 ± 1.470 mm Hg (95% CI, –2.459 to 3.304). The z-value for this effect was 0.287, resulting in P = 0.774, suggesting agreement between devices. 

Discussion

Feasibility

Our pilot results demonstrate that patient self-testing at home with virtual reality–based visual field testing and tonometry is feasible and potentially able to supplement in-office testing to expedite treatment decisions. The protocol was rigorous, requiring participants to measure eye pressures four times a day per eye, with each measurement requiring the probe to rebound off the eye six times, which equates to the device contacting the eye 48 times per day. Participants were also required to complete one visual field test per day. Of the enrolled participants, 62.5% adhered to this schedule for 3 consecutive days, suggesting that participants had both the desire and the ability to complete the study. A study by Huang et al.,¹² which used the iCare HOME and had 27 participants, suggested that four daily measurements over the course of a week would be sufficient to characterize a participants IOP, but our results suggest that only 3 days of data collection can still capture statistically significant spikes in data with a nonsignificant difference in range of pressures between in the clinic and at home. One prior study on home testing using the iCare HOME with a head-mounted device for visual field testing focused on feasibility and satisfaction.³⁰ That study reported that 73.7% of 20 patients thought the iCare HOME was easy to use, whereas all of their participants felt that the visual field headset was easy to use. Our study yielded similar conclusions, given that all participants were successfully trained to use the VRP, but six out of 15 participants were ultimately excluded due to difficulties with the iCare HOME. 

Perimetry

The visual field data collected on the VRP correlated strongly with the results obtained using the HFA in the clinic with respect to MD, PSD, and five of the six visual field sectors. The VRP also demonstrated excellent repeatability at home. However, the sectors along the inferior rim of the visual field appeared to correlate less strongly, with the VRP yielding higher sensitivity values than the HFA. LMM analysis also suggested a significant difference was present in two of six sectors, as well, with a trend for the VRP to overestimate sensitivity values. This may be because participants were permitted to use their own glasses within the VRP headset, as permitted by the manufacturer, which may have included bifocals. However, there was evidence of overestimated sensitivity values in the VRP testing in younger patients without bifocals, as well. Another potential explanation could be that the tests may have been conducted in different lighting conditions, as sometimes there is a small amount of light leak toward the bottom of the headset regardless of fit, although one may suspect this would decrease sensitivity levels on the VRP. Additionally, the VRP utilizes white points on an illuminated pixel screen (1 cd/m²), whereas the HFA uses full white on white (10 cd/m²), potentially making it easier for patients to see the stimuli. Another hypothesis is that greater variability is seen in patients with visual field defects compared to those without visual field defects, and the bias toward overestimating sensitivity values inferiorly may be a spurious statistical finding among our small cohort. Finally, despite using a stimulus of the same size, the range of intensity on the VRP (3–120 cd/m²) is much narrower than the range of intensities possible on the HFA (10–3183.1 cd/m²). With this current limitation, advanced glaucoma may be more difficult to assess with the VRP compared to the HFA. Despite this shortcoming within our small cohort, the use of at-home monitoring could result in an increase in the number of tests taken, which could increase test–retest reliability and increase confidence in the data, potentially allowing for early detection of test changes from baseline through clustered testing; however, further investigation of the quality of data with increased testing at home is warranted.^33,53 

Tonometry

The iCare HOME has FDA approval and can already be used clinically. Several studies have evaluated the accuracy of the iCare HOME compared to the GAT, both in the clinic and at home.^{43,45,54–58} This study offers another piece of evidence suggesting that the iCare HOME demonstrates good correlation with the GAT and excellent agreement based on ICCs of average values obtained, as well as LMM analysis. However, it is worth noting that there is evidence that IOP fluctuates in a diurnal pattern and that the diurnal pattern observed by the iCare HOME may differ from the pattern seen with the GAT.^12,40,58,59 Within our cohort, the majority of the peaks were during the morning and the majority of the nadirs were in the late evening, following the most typical diurnal pattern; however, peaks were seen throughout the testing periods, including in the late evening or out-of-office hours. In our study, the iCare HOME on average agreed with the in-clinic iCare HOME and GAT measurements but captured statistically significantly higher pressures that were not visible in the clinic, suggesting that at-home testing during or outside of office hours may provide information to physicians that could warrant a change in care or further evaluation, based on clinical judgment, as seen in other studies.^{36,38–40,60} Further, these data were present during the 3 days, a shorter loan period compared to prior studies, which suggests that even a short loan period of home monitoring devices can be helpful. 

There also remains the concern of the differential impact of CCT on the IOP readings from these devices, given that they both use different mechanisms to collect IOP, making a direct comparison difficult and necessitating further studies. In our study, all patients had CCT between 500 and 600 µm, a range in which the GAT and iCare HOME appear to have better correlation compared to patients with CCT < 500 µm and > 600 µm.⁵⁸ 

The iCare HOME proved to be difficult for participants to use, particularly for those with central visual field loss, making it less useful for more advanced glaucoma patients. Making the device easier to place on the face correctly would likely address many of the concerns that arose during the study. An updated version of the iCare HOME, the iCare HOME2, was released after the data collection phase of this study was complete. The iCare HOME2 features a screen that can better convey error messages and has a more ergonomic body; however, the portion contacting the face is fundamentally similar and may still be difficult to align on the face correctly, but this is yet to be seen. 

Limitations

Our study provides valuable insights into the feasibility, reliability as compared to gold standard in-clinic testing, and repeatability of at-home testing for glaucoma, but it has several limitations that should be acknowledged. The cohort size in our study was relatively small (meant to be an exploratory pilot), which may limit the generalizability of our findings. Larger studies are needed to confirm our results and to further explore the potential of at-home testing in a broader population of glaucoma patients. The small sample size also limited our ability to perform analyses stratified by the severity of glaucoma (mild, moderate, or severe). 

Furthermore, much of the analysis done in this study assumes that each subject is independent of one another. However, given that we are treating each eye as a subject, this assumption is not true, as there is some degree of inter-eye correlation of IOP and visual field defects. Although we did perform LMM analysis, introducing random effects to attempt to account for this inter-eye correlation, such an analysis is hindered by the small sample size. Therefore, future studies with a larger sample size could provide clearer insights into how the inter-eye correlation affects these results. 

Also, our study did not include the use of reminders for patients to perform their at-home tests. Although this may simulate a real-world scenario where patients may forget or neglect to perform their tests, it is worth noting that the VRP platform has the capability to automatically send reminders to patients. However, for the sake of patient privacy during our study, we limited the information provided to the device. Future studies could explore the impact of reminders on adherence to at-home testing. Despite these limitations, our study provides a foundation for further exploration into the use of at-home testing devices in glaucoma care. 

Furthermore, glaucoma medication adherence in relation to device use at home may be further explored in the future, as well, which was not done in this study; however, all nine patients documented to be using either two or three types of drops throughout the day were still able to complete the rigorous protocol, suggesting that patients may be sufficiently motivated to complete testing throughout the day in addition to the burden of instilling eyedrops. 

In addition, our study design excluded participants when they were unable to complete data collection for either device. However, given the higher rate of failure to collect data with the iCare HOME relative to the VRP in our cohort, potential data for the VRP was lost. Therefore, future studies should consider allowing participants to continue with data collection unless they are unable to collect data for both devices. Future studies would also benefit from utilizing the iCare HOME2, which has been updated to improve ease of use, potentially allowing for a higher study completion rate. 

In conclusion, our pilot study suggests that patients are motivated and capable of performing both perimetry and tonometry unsupervised at home. The VRP shows a strong correlation with MD from the in-clinic HFA and good repeatability at home across three tests. The iCare HOME captures potentially clinically significant spikes in IOP that are missed in the clinic within 3 days of proper usage. With the increasing affordability, ease of use, and portability of these devices, it is likely that they will become more widely used in glaucoma management. 

Acknowledgments

Supported by a grant from the National Institutes of Health (P30 EY026877), by a Research to Prevent Blindness Unrestricted Grant (Stanford Ophthalmology), and by an American Glaucoma Society Mentoring for the Advancement of Physician Scientists (MAPS) grant to AS. 

Disclosure: A.R. Berneshawi, None; A. Shue, None; R.T. Chang, None