This randomized, observer-blinded, prospective trial study was performed in accordance with the Helsinki Treaty and was approved by the University of Yamanashi Ethical Review Board. Written informed consent was obtained from all of the patients (see Figure 1 for the design of this study and Multimedia Appendix 1 for the CONSORT checklist; trial registration number UMIN000006982).

The GSS is a secure online PHR system. Registered patients can use the GSS to view their IOP values, visual field test results, current and prior medications, and other findings from their examination histories related to glaucoma. Figure 2 shows the first page of the GSS. Graphs are used to facilitate patient understanding of their medical records. Patients can view the detailed results of visual field tests on different pages (see Figures 3 and 4).

The GSS is based on the concept of an information security management system. The data are stored in a database server that is located in a facility at the University of Yamanashi. The registered GSS data are periodically updated by automated medical chart extraction or manual data entry by physicians and medical staff. All the data transferred from medical charts to the GSS are managed offline using locally developed data management programs. Patient registration for the GSS began in 2005, and approximately 1600 glaucoma patients were registered as of November 2011. Our plan was to allow registered patients to access their medical records by themselves. The GSS was the first PHR system routinely used in Japanese clinical ophthalmology care.

Glaucoma patients who had been periodically treated at the University of Yamanashi Hospital Glaucoma Outpatient Clinic and expressed interest in registering with the GSS were subsequently registered for the GSS. Patients were informed of the benefits and risks of the GSS by the ophthalmologist and the medical staff. The patients were also informed that the registered medical records would not be available until security was well established. Registration for the GSS began in 2005. By 2008, only ophthalmologists were allowed to browse the registered medical records, and patients could view their registered medical records in the presence of the attending ophthalmologist at the clinic.

Internet access by patients began in 2008. We randomly selected patients for Internet access between January 2008 and December 2008. The inclusion criteria included the following: patients who had been registered in the GSS database for 2 years and who were diagnosed with either primary open angle glaucoma (POAG), normal tension glaucoma (NTG), or ocular hypertension (OH). The exclusion criteria included the following: patients who were under 20 years old at the time of registration; patients who had a history of intraocular surgery in both eyes; patients for whom accurate IOP measurement was deemed difficult; patients who had received oral glaucoma therapy, such as carbonic anhydrase inhibitors; patients who had a disease other than glaucoma that caused visual field defects; patients with a visual acuity of less than 20/60 or a mean deviation (MD) value lower than –20 dB in the worse eye, as determined by the Humphrey Field Analyzer (HFA) central 30-2 program (Zeiss Inc); and patients diagnosed with dementia whose use of GSS was judged by an ophthalmologist to be difficult.

After re-confirming the patients’ preferences for browsing their registered medical records, identification numbers and temporary passwords were sent to the patients to allow them Internet access. The patients began to browse the GSS after attending lectures by the system staff.

The users were asked to replace their temporary passwords and register their user names with the system. To maintain patient anonymity, no information that could identify a patient—such as name, age, gender, telephone number, home address, or business address—was accessible through the Internet.

The outcomes of the patients who were assigned to the Internet access (IA) group were compared with those of patients who viewed their medical records only with an ophthalmologist during their medical examinations (the non–Internet access group, or NIA group).

Patients who met any of the following criteria were excluded during the 2-year period post Internet access: those who died; had glaucoma surgery, including laser treatment; received oral glaucoma therapy, such as carbonic anhydrase inhibitors; developed a disease other than glaucoma that caused visual field disturbances or IOP changes; and developed severe visual impairment or dementia that caused GSS use to become difficult as judged by an ophthalmologist.

All of the patients visited the Glaucoma Outpatient Clinic at the University of Yamanashi Hospital approximately every 3 months. A best-corrected visual acuity (BCVA) measurement, an IOP measurement using a Goldmann applanation tonometer, a slit-lamp examination, and a fundus examination were performed as part of the routine examinations. The HFA visual field test was usually performed every 6-10 months and disc photography was performed every year. No differences were found in the ophthalmological examination protocol or follow-up schedule between the NIA and IA groups. The patient and the attending ophthalmologist discussed the results of the glaucoma examination using data displayed by the GSS. The ophthalmologists did not know whether the patient browsed the GSS data at home or elsewhere.

Any glaucoma medication changes were determined by glaucoma specialists with the patient’s consent. The medical glaucoma treatment was determined by glaucoma specialists using a targeted IOP strategy. The specialists increased the number of glaucoma medication treatments in response to the following test results: two or more consecutive IOP values exceeding the target IOP, glaucomatous neuropathy deterioration suggested by a visual field test, and imaging tests focused on the optic nerve head and nerve fiber layer thickness. If two or more consecutive IOP values were sufficiently below the target IOP and the glaucoma specialists judged that reducing the glaucoma medication would not elevate the IOP over the target value, the glaucoma medication was reduced.

This study compared the amount of antiglaucoma ophthalmic solution used, MD values, and BCVA values between the two groups. In addition, we performed a subanalysis of IOP changes over the study period in those patients who did not report any changes in the amount of antiglaucoma ophthalmic solution used during the study. The IOP profiles were compared in the NIA and IA groups. IOP changes from the pre-IA to the post-IA period were also compared in the IA groups. The right eye was chosen for the analysis. If the right eye met exclusion criteria, then the left eye was subject to the analysis.

We analyzed changes in the amount of antiglaucoma ophthalmic solution used for individual patients and individual eyes. If the change in the amount of antiglaucoma ophthalmic solution used during the study period differed between the right and left eyes, the eye that showed greater change was used in the analysis. Any change in the concentration of the antiglaucoma ophthalmic solution used was considered to be a change in the antiglaucoma medication. However, a change to a different antiglaucoma ophthalmic solution having the same pharmacological action and similar IOP-reducing potential was not considered to be a change. For example, changing from a latanoprost ophthalmic solution to another prostaglandin-related antiophthalmic solution (other than an isopropyl unoprostone solution) and changing from a timolol maleate ophthalmic solution to another beta-blocker ophthalmic solution were not considered to be changes in antiglaucoma medication.

We investigated the change in medication possession ratio (MPR) among patients whose medication was not changed through the study period. A useful definition of MPR as a parameter to measure adherence is the ratio of the days of prescription supply dispensed over the number of days between the first and last prescription refill [2,21,22]. The number of days of supplied medication was calculated from the actual number of drops per bottle, based on Fiscella’s report [23] or from the manufacturer’s estimates for the products, summarized by Friedman et al [2]. See Multimedia Appendix 2 for information on the days of supplied medication for the antiglaucoma ophthalmic solutions used.

The data were analyzed using the JMP 8.0 software package (SAS Institute Inc), and the values are presented as the mean and standard deviation (SD). Changes in the amount of antiglaucoma ophthalmic solutions used were analyzed using the Wilcoxon signed rank test or a contingency table analysis. The IOP, BCVA (expressed as logMAR), MD values of the HFA central 30-2 program, and MPR were compared between the IA and NIA groups using the Mann-Whitney U test. The IOP changes within a group were analyzed using the Student t test, and the type of glaucoma and patient gender were compared using Fisher’s exact probability test. The effect of Internet access on the IOP changing trend was analyzed using the analysis of covariance (ANCOVA) and Pearson correlation coefficient; P<.05 was considered to be statistically significant.

In total, 194 patients were randomly assigned to the two groups, and 81 IA and 90 NIA patients completed the study. Tables 1 and 2 show detailed information on patient dropouts and the characteristics of the patients who completed the study. The mean age, type of glaucoma, and gender did not differ significantly between the two groups. Both the IA and NIA patients had no significant changes in their BCVA and MD values obtained with the HFA central 30-2 program between the initial and final examinations.

The number of antiglaucoma ophthalmic solutions used at the initial examination in the IA group was mean 1.2 (SD 1.2) in the right eye and mean 1.3 (SD 1.1) in the left eye, while the same number in the NIA group was mean 1.0 (SD 1.0) in the right eye and mean 1.1 (SD 1.0) in the left eye. The number of antiglaucoma ophthalmic solutions used did not differ significantly between the IA and NIA groups. In the NIA group at the final examination, it was mean 1.2 (SD 1.1) in the right eye and mean 1.2 (SD 1.0) in the left eye, which were significantly greater than the values at the initial examination (P<.001 for the right eye, P=.03 for the left eye, Wilcoxon signed-rank test). In contrast, the number of antiglaucoma ophthalmic solutions used in the IA group decreased by mean 0.1 (SD 0.4) in the right eye and by mean 0.1 (SD 0.6) in the left eye. The changes in the number of antiglaucoma ophthalmic solutions used in the NIA and IA groups were mean 0.3 (SD 0.4) and mean –0.1 (SD 0.4), respectively, which is significantly different (P<.001, Wilcoxon signed-rank test). None of the patients reported using less antiglaucoma ophthalmic solution in one eye and more in the other eye during the investigation period.

Figure 5 shows the distribution of the medication used by the glaucoma patients during the course of the study. In the NIA group (n=90), the amount of antiglaucoma ophthalmic solution used did not change in 55 patients (61.1%), increased in 16 patients (17.8%), and decreased in 3 patients (3.3%). No medications were administered to 16 patients (17.8%). In the IA group (n=81), the amount of antiglaucoma ophthalmic solution used did not change in 46 patients (56.8%), increased in 7 patients (8.6%), and decreased in 11 patients (13.6%). No medications were administered to 17 patients (21.0%). Compared to the NIA group, 10.3% more IA patients reported using decreased number of antiglaucoma ophthalmic solutions and 9.2% fewer patients reported using increased number of antiglaucoma ophthalmic solutions. A significant difference was found in the distribution of medications received between the two groups of glaucoma patients (P=.008, contingency table analysis).

IOP values at the first examination of the IA and NIA groups were mean 15.3 (SD 4.2) mmHg and mean 15.1 (SD 3.1) mmHg, respectively (P=.69). IOP values at the last examination of the IA and NIA groups were mean 14.6 (SD 3.6) mmHg and mean 15.1 (SD 3.3) mmHg, respectively (P=.26). Changes in IOP over the study period were not significant in either the IA (P=.23) or the NIA group (P=.83).

We examined the patients whose antiglaucoma ophthalmic solution use did not change over the course of the study. The subjects included 46 patients in the IA group and 55 patients in the NIA group. The distribution of the cases is shown in Table 3. The two groups did not differ significantly by age, gender, type of glaucoma, or the amount of antiglaucoma ophthalmic solutions used. The two groups also did not differ significantly by their BCVA and MD values measured at the beginning of the study or by changes in their BCVA and MD values during the study period. See Multimedia Appendices 3 and 4 for information on the antiglaucoma ophthalmic solutions used.

The IOP values in the IA and NIA groups were mean 15.2 (SD 4.2) mmHg and mean 14.5 (SD 2.1) mmHg at the first examination, respectively. No significant IOP differences were observed between the IA and NIA groups (P=.31).

Differences in the IOP values between the first examination and the final examination in the IA group and NIA group were mean 0.2 (SD 1.4) mmHg (P=.32) and –0.8 (SD 3.0) mmHg (P=.10), respectively. The IA group showed a tendency toward IOP reduction during the study period.

In the IA group, the IOP of both eyes changed significantly between the pre-IA period and the post-IA period. Figure 6 shows the changing IOP profiles during the study period. In the IA group, the IOP was increasing before the start of Internet access (R ²=.790, P=.04) and decreased after the start of Internet access (R ²=.800, P=.04). By contrast, no significant IOP changes were observed during the study period in the NIA group (R ²=.055, P=.57). The ANCOVA showed that Internet access significantly shifted from an increasing IOP trend to a decreasing trend in the IA group (P=.002). The IOP increased by mean 0.1 (SD 0.1) mmHg/month during the pre-IA period, while the IOP decreased by mean 0.1 (SD 0.1) mmHg/month during the post-IA period.

In the IA group, the MPRs before and after the start of Internet access were mean 82.3% (SD 30.7) and 91.1% (SD 40.3), respectively, which indicates a significant improvement (P=.03). In the NIA group, the MPRs before and after the start of Internet access were mean 84.0% (SD 28.5) and 82.9% (SD 31.1), respectively, which does not indicate significant improvement (P=.56).

The utility of patient education and medical information services has been previously discussed [7,8]. Despite multiple efforts, however, there is insufficient evidence supporting the utility of patient education and medical information services in glaucoma treatment [8,24,25]. The current study showed that allowing glaucoma patients to self-browse their clinical data contributed to two favorable effects: using less antiglaucoma ophthalmic solution and improving the effectiveness of the medication in reducing IOP.

It was common for the patients to increase their medication use during the treatment period. Indeed, 17.8% (16/90) of the NIA patients increased their glaucoma medications during the study period, while only 8.6% (7/81) of the IA patients did. In addition, 13.6% (11/81) of the IA patients reduced their medications during the study period. Notably, among the patients whose antiglaucoma ophthalmic solution use did not change during the study period, the Internet access group shifted from an increasing IOP trend to a decreasing trend. It is possible that the amount of antiglaucoma ophthalmic solutions used was increased in the patients who did not start Internet access. It is unclear whether the current outcomes were the direct results of Internet access or were due to other mechanisms, such as improved patient-physician communication during office visits. Improving adherence may contribute to these outcomes, although we were unable to monitor changes in adherence using subjective and quantitative methods, such as electronic monitoring systems, which are not available in Japan.

The current study showed that Internet access significantly improved the MPR among patients whose medication was not changed during the study and that Internet access may contribute to a reduction in IOP over time. The MPR in the current study is higher than that in Friedman’s report [2]. A possible explanation for this difference could be that the understanding of glaucoma and the motivation for glaucoma treatment may have been stronger for the patients in this study than those in the previous study. In the present study, patients had been treated for glaucoma before their enrollment and were willing to use the GSS.

Health literacy is an important consideration in using electronic PHRs appropriately. We examined glaucoma literacy by administering a questionnaire to another set of patients who had been registered with the GSS for more than 2 years [26]. The patients who had been self-checking their registered data for more than 1 year exhibited a much better understanding of aspects of their glaucoma, such as their glaucoma severity, IOP values, and medications, than did the patients who had been provided their registered data by the ophthalmologists (see Figures 7 and 8). The IA patients in the current study may have improved their eye health literacy by checking their data through the Internet.

Although the current study demonstrated the effects of Internet access on glaucoma treatment, these effects are not consistent with those observed in previous studies. In their review, Tenforde et al [19] concluded that despite PHR’s potential to improve chronic disease management and patient outcomes, the evidence supporting the clinical value of PHR remains limited. Many of the previous studies have enrolled patients with diabetes mellitus or other systemic diseases in which interventions may influence clinical outcomes through complicated processes. By contrast, glaucoma is a disease in which proper medication is the only intervention proven to influence IOP control. This difference could be a possible explanation for inconsistencies with previous studies.

We used a prospective, randomized study design, but the NIA patients were slightly older than the IA patients. The effect of age on adherence is controversial. Dietlein et al have reported that adherence to therapy with antiglaucoma ophthalmic solutions deteriorated with age [27,28], while other previous studies have reported that younger patients have poor adherence [29,30]. In the present study, no significant differences were found in the MD value, the number of antiglaucoma ophthalmic solutions used, or the BCVA value, suggesting that there were no marked differences in glaucoma severity between the two groups. Although we eliminated the patients who met the exclusion criteria, no significant differences were found in the reasons for loss to follow up between the two groups. Altogether, the influence of any factors, except for Internet access, between the two groups may have been limited.

The patients in this study were GSS users. Given that many patients cannot use a personal computer or the Internet due to their age, poverty, or other reasons, it is possible that the patients in this study are not representative of general glaucoma patients. The present study did not examine how frequently the patients accessed the GSS. Therefore, further investigations should be performed to verify the relationship between the frequency of use and the improvements in glaucoma treatment associated with using the system.

The current study confirmed that patients’ understanding of their glaucoma care status may play an important role in better management.