Analysis of variance between groups of 0, 1, 2, and 3 medications was statistically significant (P < .001). Paired t tests showed significant differences between all pairs of medication groups (P < .001). Error bars indicate 95% CIs.
eFigure 1. Inclusion and Exclusion Criteria
eFigure 2. IOP Percentage Increase Compared With IOP Percentage Decrease
eTable. Relationship Between CCT and IOP Parameters
Jampel HD, Chon BH, Stamper R, Packer M, Han Y, Nguyen QH, Ianchulev T. Effectiveness of Intraocular Pressure–Lowering Medication Determined by Washout. JAMA Ophthalmol. 2014;132(4):390-395. doi:10.1001/jamaophthalmol.2013.7677
While medication efficacy is well documented in clinical trials, less is known of medication effectiveness in real-world clinical settings.
To assess the effectiveness of intraocular pressure (IOP)–lowering medications in patients with open-angle glaucoma.
Design, Setting, and Participants
Prospective, multicenter, interventional cohort from the prerandomization phase of a randomized clinical trial at multiple ophthalmology clinics. A total of 603 patients (603 eyes) with primary open-angle glaucoma who were using up to 3 glaucoma medications were included.
One IOP measurement was made while the patient was using his or her usual medications to lower IOP (ON IOP). Eligible participants underwent washout of all IOP-lowering drops, and the diurnal IOP was measured 2 to 4 weeks later (OFF IOP).
Main Outcomes and Measures
Difference between OFF IOP and ON IOP. The hypothesis was formulated after data collection.
The mean (SD) ON IOPs for participants using 0 (n = 102), 1 (n = 272), 2 (n = 147), or 3 (n = 82) medications were 24.2 (3.2), 17.5 (3.2), 17.2 (3.1), and 17.2 (3.1) mm Hg, respectively. Patients not using medication had a mean (SD) IOP decrease of 0.2 (2.8) mm Hg at the OFF visit. Patients using 1, 2, and 3 medications had mean (SD) IOP increases of 5.4 (3.0), 6.9 (3.3), and 9.0 (3.8) mm Hg, respectively, at the OFF visit. The percentages of patients with less than a 25% increase in IOP were 38%, 21%, and 13% for those using 1, 2, and 3 medications, respectively.
Conclusions and Relevance
Discontinuation of 1, 2, and 3 medications was associated with a clinically significant increase in IOP, although with smaller effects for the second and third medications compared with the first medication. A substantial proportion of patients showed only small changes in IOP after medication washout, suggesting either that they were not using the medication effectively or that the medication itself, although used properly, was not lowering the IOP.
clinicaltrials.gov Identifier: NCT01085357
Lowering intraocular pressure (IOP) remains the treatment for glaucoma. Topical glaucoma medications are an important component of IOP-lowering therapy.1 Many patients are treated with 2 or more medications. Numerous randomized clinical trials (RCTs) have shown the efficacy of monotherapy2 and adjunctive therapy3,4 for glaucoma.
Quiz Ref IDRandomized clinical trials evaluate medication efficacy in a select study population, incentivized to maintain their medication regimen during a relatively short period. However, medications may not provide the same results in clinical practice as those observed in clinical trials, also called the efficacy-effectiveness gap.5 Efficacy is the extent to which an intervention demonstrates a beneficial result under ideal circumstances, eg, an RCT. Effectiveness refers to the extent to which the intervention works in practice in a real-world situation.6 In other words, efficacy answers “Can it work?” whereas effectiveness answers “Does it work?”7 In the real-world setting, factors such as lack of patient adherence and improper or inconsistent eyedrop administration8 can decrease treatment effectiveness, but these factors may be less prevalent in an RCT. Patient adherence with treatment regimens can be particularly difficult in chronic and rarely symptomatic diseases9 such as glaucoma.
An opportunity to observe long-term glaucoma therapy effectiveness was available from data of the prerandomization period of the COMPASS phase 3 randomized study of the supraciliary CyPass Micro-Stent (Transcend Medical). Until the patients were approached about the trial, they were using their medication in a real-world setting and were subject to the uncertainties of medical therapy not typically seen in RCTs. In contrast to efficacy trials that monitor the response of IOP to adding 1, 2, or 3 medications, our study assessed the effectiveness of therapy by comparing IOP response before and after standardized medication discontinuation. Compared with data collected from an RCT, this discontinuation method produces measurements that more closely approximate the real-world clinical effectiveness of glaucoma therapy than does an RCT, as it provides a snapshot of clinical use and outcomes before clinical trial entry and introduction of investigation-specific selection and ascertainment biases.
Prerandomization data were analyzed from the COMPASS study, a randomized, double-masked, comparative trial of the CyPass Micro-Stent for the treatment of open-angle glaucoma (OAG). All patients provided written informed consent at the initial screening visit before study enrollment and subsequent medication washout. The COMPASS study protocol was approved by the institutional review board at each site, adhered to the tenets of the Declaration of Helsinki, and was compliant with the Health Insurance Portability and Accountability Act of 1996. The study was conducted across 7 sites in the United States.
At the screening visit, the diagnosis of glaucoma was made based on glaucomatous optic neuropathy detected by direct ophthalmoscopy and/or glaucomatous visual field defect tested with the Humphrey automated perimeter (Carl Zeiss Meditec) using the Swedish Interactive Threshold Algorithm standard 24-2 program. Participants had to meet the following inclusion criteria: mild to moderate primary OAG diagnosis with a mean deviation score of−12.0 or greater to less than 0 dB, presence of operable cataract, and IOP not exceeding 33 mm Hg with 0 to 3 IOP-lowering medications. There was no minimum IOP cutoff used for enrollment. Fixed combination medications, such as a topical carbonic anhydrase inhibitor and a β-adrenergic antagonist, were counted as 2 medications. Exclusion criteria included presence of angle closure, grade 1 or 2 by Shaffer classification, secondary OAG, uveitic and neovascular glaucoma, or any other discernible congenital anomalies of the anterior chamber and angle, including presence of other ocular pathology judged significant by discretion of the investigator. If both eyes qualified for the study, a random number generator was used to designate the study eye.
All patients under consideration for inclusion in the study, including those not using any medications, underwent IOP measurement at a screening visit while they were using their usual glaucoma medication regimen (ON). At baseline, another set of IOP measurements was taken after complete IOP-lowering medication washout (OFF). The washout period was a minimum of 5 days for carbonic anhydrase inhibitors, 14 days for α-adrenergic agonists, and 28 days for all other medications. At the discretion of study investigators, a subset of patients (n = 11) received brinzolamide during the washout period; these patients were excluded from this analysis (eFigure 1 in Supplement).
All IOP measurements were taken by 2 trained individuals using Goldmann tonometry. One individual performed the measurement, while the other read the scale. The IOP was measured twice. If the 2 measurements differed by no more than 2 mm Hg, then the mean was used as the IOP. If the 2 measurements differed by more than 2 mm Hg, then a third IOP was taken and the median IOP was used. The ON IOP was recorded at a single time, while the OFF IOP was measured at 8 am, 12 pm, and 4 pm. We defined the mean OFF IOP as the average of these 3 IOP measurements. The mean of the 3 measurements was used to determine the baseline IOP for purposes of qualifying for the study and for comparison with posttreatment IOP measurements. The change in IOP was defined as the mean diurnal OFF IOP minus the single-screening ON IOP. An additional analysis was conducted defining OFF IOP as the postwashout IOP measurement most close in time of day to the ON IOP measurement (eg, if ON IOP was taken at 9:30 am, the 8 am OFF IOP measurement was selected). The IOP percentage increase was defined as IOP change divided by ON IOP, and the IOP percentage decrease was defined as IOP change divided by OFF IOP. The IOP change, IOP percentage increase, and IOP percentage decrease were calculated for each individual and averaged for groups of patients using 0, 1, 2, and 3 medications.
Central corneal thickness (CCT) was measured by ultrasonography (Sonomed, Inc) or optical coherence tomography (Visante; Carl Zeiss Meditec). Both eyes were measured 3 times, and the study eye’s average CCT was used.
Initially, statistical comparisons were between patients using different numbers of medications with 1-way analysis of variance. All tests were 2-sided. If analysis of variance was statistically significant at P < .05, paired t tests were performed between the groups of patients using different numbers of medications with Tukey-Kramer t tests for post hoc multiple pairwise comparisons. Univariate and multivariate linear regression analyses were performed for IOP change as predicted by CCT, adjusted for age, sex, and IOP after washout (OFF IOP). Statistical analyses were performed with JMP statistical software (SAS Institute, Inc).
Six hundred nineteen patients had both ON and OFF IOP measurements. Sixteen patients were excluded from this analysis: 5 were using 4 glaucoma medications and 11 did not adhere to the washout instructions (eFigure 1 in Supplement). The remaining 603 patients had a mean (SD) age of 70.3 (8.1) years, were 55% female (334 of 603 patients), and had a mean (SD) Humphrey visual field mean deviation of −3.5 (3.1) dB.
A total of 102 (17%), 272 (45%), 147 (24%), and 82 (14%) patients were using 0, 1, 2, and 3 medications, respectively (Table 1). Prostaglandin analogues were the most common medication class, being used by 81%, 78%, and 100% of patients with 1-, 2-, or 3-medication regimens, respectively.
Table 2 summarizes IOP data by number of medications, with OFF IOP as the average of 3 measurements. The IOP percentage increase is the change directly observed in this study after discontinuing medications, whereas the IOP percentage decrease is the calculated percentage if medications had been added instead of removed (eFigure 2 in Supplement). Participants using no medications at the screening visit served as validation for using 1 measurement for ON IOP measurements and 3 measurements for OFF IOP measurements. On average, for those using 0 medications, there was no IOP change (mean [SD], −0.2 [2.8] mm Hg). The IOP change, IOP percentage increase, and IOP percentage decrease were each statistically different (P < .001) across 0-, 1-, 2-, and 3-medication groups. Analysis using the OFF IOP closest in time to the ON IOP measurement produced similar results (data not shown).
In paired comparisons, IOP difference (Figure), IOP percentage increase, and IOP percentage decrease were each statistically different between 0-, 1-, 2-, and 3-medication groups (P < .001). Table 3 displays paired comparisons for IOP change and IOP percentage increase after medication washout between 0-, 1-, 2-, and 3-medication groups. Quiz Ref IDThe largest change in IOP was observed with the removal of 1 medication, while the removal of a second or third medication produced smaller but additional changes in IOP. Removing 1 medication, compared with 0 medications, increased IOP by 5.6 mm Hg. The difference between 2 medications and 1 medication, or the additional IOP change with a second medication, was 1.5 mm Hg. The difference between 3 and 2 medications, or the additional IOP change with a third medication, was 2.1 mm Hg. In terms of IOP percentage increase, removing 1 medication provided a 33% increase in IOP compared with 0 medications. Removing a second medication provided an additional 9% increase in IOP compared with 1 medication. Removal of a third medication provided an additional 13% increase in IOP compared with 2 medications. Similar results were obtained when using only the OFF IOP measurement closest in time to the ON IOP measurement (data not shown).
Table 4 demonstrates the percentage of patients with an IOP increase of less than 20%, 25%, or 30% after medication washout. Discontinuing treatment produced less than a 20% increase in IOP for 29%, 16%, and 4% of patients using 1, 2, and 3 medications, respectively. A 25% increase in IOP was not seen in 38%, 21%, and 13% of patients using 1, 2, and 3 medications, respectively. A 30% IOP increase was not seen in 47%, 31%, and 17% of patients using 1, 2, and 3 medications, respectively. Similar results were seen when OFF IOP was defined by the measurement closest in time of day to the ON IOP measurement (data not shown).
Quiz Ref IDThe mean (SD) CCT in our study was 548.2 (36.5) μm. Patients with a greater CCT did not have a statistically different OFF IOP but did have a higher ON IOP. Thicker corneas had significantly smaller IOP change, IOP percentage decrease, and IOP percentage increase (eTable in Supplement). In univariate and multivariate analysis, a greater CCT was associated with a higher IOP while using IOP-lowering medication.
This study reports the effectiveness of IOP-lowering topical agents in patients already under treatment and under consideration for participation in a clinical trial. The study design differs from almost all other reports in the literature, which would be characterized as trials of efficacy. In the only other discontinuation study in the literature of which we are aware, conducted in patients using both latanoprost and timolol, Kobayashi10 found that discontinuing latanoprost resulted in a mean (SD) IOP increase of 4.3 (1.6) mm Hg (mean [SD], 27.2% [11.8%]), whereas discontinuing timolol resulted in a mean (SD) increase of only 1.6 (1.2) mm Hg (mean [SD], 10.3% [8.3%]).
We found that when patients already using 1, 2, or 3 IOP-lowering agents had all medications discontinued, the mean (SD) IOP increased by 5.4 (3.0), 6.9 (3.3), and 9.0 (3.8) mm Hg, respectively. The IOP increased by 33%, 43%, and 55% when removing 1, 2, or 3 medications, respectively. The increase in IOP was statistically different between the 0-, 1-, 2-, and 3-medication groups. The use of more medications was associated with greater IOP change, although the second and third medications were not as effective as the first medication. In a retrospective record review of patients from a single glaucoma specialist, 40% to 60% of patients already using either 2 (n = 67) or 3 (n = 29) glaucoma medications had an initial IOP decrease greater than 20% after addition of 1 or more medications, but this decrease was sustained for at least a year in only 30% of eyes.11
In addition to analyzing the effect of the number of medications on mean IOP, we also looked at the proportion of patients achieving a specific increase in IOP. After discontinuing 1, 2, or 3 medications, 38%, 21%, and 13% of patients had less than a 25% increase in IOP and 47%, 31%, and 17% had less than a 30% increase in IOP, respectively. From both the mean IOP-lowering data and the proportion of eyes with a given percentage of IOP lowering, we conclude that in the real-world setting, both second and third IOP-lowering agents do provide additional IOP lowering.
The ideal conditions in an RCT, which measures efficacy, have distinct differences from the conditions in the real-world clinical situation that might influence effectiveness. Clinical trials use several means to minimize confounding effects and optimize treatment efficacy, including specific dose regimens, inclusion and exclusion criteria for study participants, and efforts by trial staff to improve treatment adherence.5 A number of potential barriers can decrease glaucoma treatment effectiveness, including increased or inconvenient dosing frequency,12,13 medication adverse effects,14 patient forgetfulness,14,15 improper administration of eye medications,12,15,16 lack of visual symptoms without medication,12 poor patient-physician communication,8 cost of medications,8 and poor health literacy.17
We hypothesized that the IOP lowering of medications observed in this analysis of effectiveness would be less than that reported in RCTs measuring efficacy. In an RCT comparing latanoprost and bimatoprost, patients had a decrease of 20% or more in IOP for 69% to 82% of participants using bimatoprost and 50% to 62% of participants using latanoprost.18 In another RCT, 74% of participants using latanoprost had a decrease of 20% or more in IOP.19 In our study, 147 of 219 patients (67%) using a single prostaglandin analogue had a decrease of 20% or more in IOP. Therefore, contrary to our hypothesis, the effectiveness of prostaglandin analogues in our data set appears to be similar to the efficacy reported in RCTs.
The comparability of the effectiveness data we present and the efficacy data from RCTs appears to hold true for 2 and 3 medications as well. In one meta-analysis, the additional IOP reduction from adding a β-blocker to a prostaglandin analogue was between 10% and 21%.20 In a second meta-analysis,3 the addition of a second medication (β-blocker, carbonic anhydrase inhibitor, or α-agonist) to a prostaglandin analogue improved IOP lowering by an additional 15% or less. The removal of 3 medications in our study was associated with an increase in IOP of 9.0 mm Hg, compared with 6.9 mm Hg for removal of 2 medications, an incremental benefit of 2.1 mm Hg. This is comparable to a randomized study of 153 patients showing that addition of a carbonic anhydrase inhibitor reduced IOP by 13.8% (mean [SD], 20.3 [2.0] to 17.5 [2.6] mm Hg) in patients already using a prostaglandin analogue and β-blocker combination, compared with a 7% reduction with placebo (mean [SD], 20.9 [2.7] to 19.4 [3.8] mm Hg).21 Similarly, in another study of 204 patients already using a prostaglandin analogue, IOP reduction improved by an additional 6.1 mm Hg (26.5%) for adjunctive β-blocker therapy compared with 8.3 mm Hg (35.5%) for adjunctive β-blocker and α-agonist therapy.22
Central corneal thickness has become an important parameter to measure in glaucoma trials. In the Ocular Hypertension Treatment Study,23 thicker corneas were associated with smaller responses after initiating ocular hypotensive medications. Similarly, our study showed that patients with a greater CCT had a smaller change in IOP after discontinuing medications.
There were several limitations in this study. As a retrospective analysis, patients were not equally divided into subgroups at each level of medication, such as equal numbers of patients using prostaglandin analogues, β-blockers, carbonic anhydrase inhibitors, and α-agonists. Instead, most patients were using a prostaglandin analogue, limiting comparisons by various 1-, 2-, and 3-medication regimens. It has been observed that patients are more likely to use their eyedrops in the days leading up to their appointment than at other times,24 and we were unable to account for this possibility. Although the data were collected prospectively in a standardized manner, the study was not designed with the primary intent of determining medication effectiveness.
The COMPASS study is a US Food and Drug Administration–approved clinical trial comparing patients with primary OAG randomized to either combined CyPass Micro-Stent and cataract surgery or cataract surgery alone. The study’s primary objective is to demonstrate therapeutic efficacy of supraciliary stenting for IOP lowering in the setting of cataract surgery. Nevertheless, the prerandomization medication washout period presents a significant and unique opportunity to investigate the effectiveness of IOP-lowering medications in that population. This inherent study design is also the reason for the single IOP measurement before washout rather than a matching diurnal study as is the case after washout. It is possible that 1 IOP measurement for ON IOP may have been insufficient to obtain a true baseline while using medications. We tried to control for this by showing that the change in IOP between ON IOP and OFF IOP was not significantly different in patients using 0 medications and by comparing the single ON IOP with the single OFF IOP closest in time to the single ON IOP.
This study used medication discontinuation and washout period in assessing IOP effectiveness rather than initiation and addition of IOP-lowering medications. As initiation of IOP treatment in the context of a clinical trial will introduce selection bias and confounding through study intervention, the washout method may be a more accurate measure of real-world, population effect. In addition, it captures a cross section of real-world treatment duration where there is no arbitrary cutoff in follow-up. This study allowed for a method of measuring clinical effectiveness of medication and has shown that despite many of these potential barriers, it appears that, at least in this patient population, clinical efficacy and effectiveness are similar. While there are numerous RCTs on the efficacy of various glaucoma therapies, it remains important to show the effectiveness of medications for patients, physicians, and policy makers.
Quiz Ref IDWe studied the clinical utility of multiple IOP-lowering medications in patients with OAG who underwent standardized total topical medication washout. We found that there was an incremental effect on IOP with 1, 2, and 3 medications. The largest effect on IOP was with the first medication, while the second and third medications produced significant but smaller changes in IOP. We were unable to demonstrate that the effectiveness of topical therapy for lowering IOP was less than the efficacy reported in prospective clinical trials.
Corresponding Author: Henry D. Jampel, MD, Woods 155 B1, Johns Hopkins Hospital, 1800 Orleans St, Baltimore, MD 21287 (email@example.com).
Submitted for Publication: July 2, 2013; final revision received September 18, 2013; accepted September 27, 2013.
Published Online: January 30, 2014. doi:10.1001/jamaophthalmol.2013.7677.
Author Contributions: Dr Jampel had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Jampel, Stamper, Packer, Ianchulev.
Acquisition of data: Packer, Nguyen, Ianchulev.
Analysis and interpretation of data: All authors.
Drafting of the manuscript: Jampel, Chon, Han, Ianchulev.
Critical revision of the manuscript for important intellectual content: Jampel, Chon, Stamper, Packer, Nguyen.
Statistical analysis: Chon, Han.
Administrative, technical, and material support: Jampel, Stamper, Ianchulev.
Study supervision: Jampel, Stamper, Packer.
Conflict of Interest Disclosures: Drs Jampel, Stamper, and Packer are consultants of Transcend Medical.
Funding/Support: Mr Chon was supported by a research grant from Transcend Medical.
Role of the Sponsor: Transcend Medical had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.