Adherence To Fixed-Combination Versus Unfixed Travoprost 0.004%/timolol 0.5% For Glaucoma or Ocular Hypertension: A Randomized Trial
Adherence To Fixed-Combination Versus Unfixed Travoprost 0.004%/timolol 0.5% For Glaucoma or Ocular Hypertension: A Randomized Trial
Adherence To Fixed-Combination Versus Unfixed Travoprost 0.004%/timolol 0.5% For Glaucoma or Ocular Hypertension: A Randomized Trial
PURPOSE: To assess adherence to treatment with fixed- receiving TRAVDTIM through 12 months of
combination travoprost 0.004%/timolol 0.5% (TTFC) on-therapy evaluation. This suggests that, for patients
compared with separate containers of travoprost requiring multiple IOP-lowering medications, a fixed
0.004% and timolol 0.5% (TRAVDTIM; unfixed) using combination may provide improved long-term adherence
electronic dosing aids. compared with unfixed therapy. (Am J Ophthalmol
DESIGN: Randomized, controlled, observer-masked 2017;176:61–69. Ó 2016 The Author(s). Published by
clinical trial. Elsevier Inc. This is an open access article under the
METHODS: SETTING: Two US clinical sites. PATIENT CC BY-NC-ND license (http://creativecommons.org/
POPULATION: Eligible patients were adults diagnosed licenses/by-nc-nd/4.0/).)
with open-angle glaucoma or ocular hypertension.
Patients (n [ 81) were sequentially randomized 1:1 to
P
receive TTFC or TRAVDTIM for 12 months. INTERVEN- HARMACOLOGIC CONTROL OF INTRAOCULAR PRES-
TION: TTFC was administered once daily in the morning sure (IOP) with topical ocular hypotensive medica-
or evening with a single dosing aid. Patients randomized tions is the standard of care for the initial treatment
to TRAVDTIM administered TRAV once daily in the of open-angle glaucoma and ocular hypertension.1
evening and TIM once daily in the morning using separate Elevated IOP can lead to optic nerve head damage, visual
dosing aids. MAIN OUTCOME MEASURE: Adherence with field loss, or disability and blindness. Higher IOP levels
administered medication, as recorded by the dosing aids. are associated with increased risk for disease progression
RESULTS: Mean ± SD patient age was 60 ± 10 years; and glaucoma-related visual disability and blindness.2,3
most patients were male and white. Compared with Ocular hypotensive therapy is effective in slowing or
TRAVDTIM (n [ 40), patients receiving TTFC preventing progression of glaucoma and ocular
(n [ 41) were consistently adherent on a greater percent- hypertension,4–7 and maintaining long-term IOP reduction
age of days through month 12 (60% vs 43%). At months reduces the risk of vision loss. For some patients, a single
1, 3, 6, and 12, 80% adherence was achieved by 71% vs medication produces sufficient IOP reduction, although
38%, 53% vs 30%, 45% vs 16%, and 32% vs 11% of many patients require treatment with more than 1 medica-
patients receiving TTFC vs TRAVDTIM, respectively. tion to achieve and maintain sufficiently low IOP.5,6
Significantly more patients were adherent on ‡80% of Patient adherence with IOP-lowering treatment is crucial
days with TTFC compared with TRAVDTIM for the successful medical management of glaucoma; howev-
(P < .001 to P [ .041). Both treatments reduced er, adherence with glaucoma medications is typically low and
IOP from baseline, and no safety issues were identified decreases with time.8–10 Adherence is influenced by many
in either group. Ocular hyperemia was the most common patient- and treatment-related variables, and treatment
treatment-related adverse event (n [ 3/group). complexity (ie, number of individual medications and daily
CONCLUSIONS: Patients receiving TTFC maintained doses) is one of the only modifiable factors associated with
better treatment adherence compared with patients suboptimal treatment adherence. For example, compared
with combination medications, treatment regimens using
multiple individual medications are associated with lower
Supplemental Material available at AJO.com.
treatment adherence.11–16 Most fixed-combination therapies
Accepted for publication Dec 7, 2016. provide 2 ocular hypotensive agents in a single formulation,
From the Specialty Eyecare Centre, Bellevue, Washington (H.S.B.); thereby simplifying treatment regimens and reducing the
Department of Ophthalmology (A.L.R.) and Bloomberg School of
Public Health (A.L.R.), Johns Hopkins University, Baltimore,
number of daily instillations. An inherent disadvantage of
Maryland; University of Maryland, Baltimore, Maryland (A.L.R.); and fixed combinations is that missed doses result in omission
University of Michigan, Ann Arbor, Michigan (A.L.R.). of both medications rather than of a single agent.
Inquiries to Howard S. Barnebey, Specialty Eyecare Centre, 1920 116th
Ave NE, Bellevue, WA 98004; e-mail: hbarnebey@
Because better adherence with glaucoma therapy is asso-
specialtyeyecarecentre.com ciated with reduced progression of visual field defects,17,18
Randomized (n=81)
Allocation
Allocated to fixed-combination Allocated to unfixed travoprost 0.004%
travoprost 0.004%/timolol 0.5% (n=41) + timolol 0.5% (n=40)
Follow-up
• Completed the study (n=36) • Completed the study (n=35)
• Discontinued the study (n=5) • Discontinued the study (n=5)
Inadequate intraocular pressure Inadequate intraocular pressure
control (n=2) control (n=1)
Sponsor request (n=2) Sponsor request (n=1)
Adverse event (n=1) Adverse event (n=1)
Noncompliance (n=1)
Patient decision to withdraw (n=1)
Analysis
• Intent-to-treat data set (n=41) • Intent-to-treat data set (n=40)
• Safety data set (n=41) • Safety data set (n=40)
FIGURE 1. Disposition of patients with open-angle glaucoma or ocular hypertension receiving fixed-combination travoprost
0.004%/timolol 0.5% or unfixed travoprost 0.004% D timolol 0.5%.
1 drop dispensed. The mean percentage of days on which Diabetic Retinopathy Study (ETDRS) chart at a viewing
patients adhered to their dosing schedules was the primary distance of 10 feet. Ocular signs were assessed for both
measure of adherence and was evaluated cumulatively over eyes at every visit using slit-lamp biomicroscopy. Dilated
fixed time intervals and for the overall duration of the fundus examinations were conducted for both eyes at
study. Cumulative adherence was calculated for intervals screening and month 12 visits after completion of IOP
of increasing duration (ie, through 1 month, 3 months, measurements. Systolic and diastolic blood pressures were
6 months, and 12 months). Responder analyses were recorded at every visit after 5 minutes of resting; heart
performed to assess threshold levels of adherence. Patients rate measurements were based on a 60-second count. AE
who discontinued the study early contributed data only to information was collected during all on-therapy visits;
the point of discontinuation. AEs were coded using the Medical Dictionary for Regulatory
Safety variables included IOP measurements, visual field Activities, version 13.0.
test results, gonioscopy evaluation findings, BCVA assess-
ments, ocular signs (cornea, iris/anterior chamber, lens, STATISTICAL ANALYSIS: Adherence was analyzed in the
eyelids, and conjunctiva), dilated fundus examinations intent-to-treat (ITT) population, defined as all patients
(vitreous, retina/macula/choroid, optic nerve, cup-to-disc who instilled >_1 dose of study medication using the dosing
ratio), blood pressure, heart rate, and adverse events aid and who provided > _1 day of adherence information
(AEs). IOP was measured in both eyes at all study visits from the dosing aid. Safety variables were assessed in the
by an independent operator and reader using a calibrated safety population, which was defined as all patients who
Goldmann applanation tonometer, after completion of instilled >
_1 dose of study medication using the dosing aid.
BCVA and slit-lamp biomicroscopy assessments. Central Patient demographics and baseline characteristics,
threshold visual fields were assessed at screening and month adherence outcomes, and safety variables were summarized
12 with habitual correction by achromatic automated peri- descriptively. Cumulative adherence was calculated for
metry with a Humphrey Field Analyzer using a 24-2 various thresholds (50%, 60%, 70%, 80%, 90%, and
threshold field test. Gonioscopy was performed at screening 95%). All threshold levels were included in adherence cal-
and month 12 before instillation of dilating or miotic drops, culations for each cumulative time interval. Post hoc statis-
and gonioscopy was graded using the modified Shaffer tical analyses of potential between-group differences in the
grading scale. BCVA was assessed at every study visit, percentage of patients achieving various adherence thresh-
before IOP measurement, using an Early Treatment olds were performed using Fisher exact tests and x2 tests.
Age, years
RESULTS Mean 6 SD 58.7 6 10.2 61.5 6 9.3 .213a
Range 29–76 44–80
PATIENTS: Eighty-one patients were enrolled, random-
Distribution .439b
ized to treatment, and included in the ITT data set <65 27 (65.9) 23 (57.5)
(TTFC, n ¼ 41; TRAVþTIM, n ¼ 40; Figure 1). Mean >
_65 14 (34.1) 17 (42.5)
6 SD patient age was 60 6 10 years; 38% of patients Sex, n (%) .753b
(31/81) were > _65 years old. Most patients were male Male 28 (68.3) 26 (65.0)
(67%; 54/81) and white (89%; 72/81; Table 1). Patient Female 13 (31.7) 14 (35.0)
demographic and baseline characteristics were similar Race, n (%) .781c
between treatment groups. White 35 (85.4) 37 (92.5)
Ten patients discontinued the study (n ¼ 5 per treat- Black or African 4 (9.8) 3 (7.5)
American
ment group). Reasons for discontinuation in the TTFC
Native Hawaiian 1 (2.4) 0
group were inadequate IOP control (n ¼ 2), sponsor
or Pacific Islander
request (n ¼ 2), and AE (n ¼ 1). Reasons for discontinua- Other 1 (2.4) 0
tion in the TRAVþTIM group were inadequate IOP con-
trol, AE, nonadherence, sponsor request, and patient a
2-sided t test.
decision (n ¼ 1 each). x test.
b 2
c
Fisher exact test.
ADHERENCE: Throughout 12 months of on-therapy
evaluation, patients receiving TTFC were adherent with
their dosing schedule on a greater percentage of days
compared with patients receiving TRAVþTIM points. The threshold of adherence with dosing on >
_80%
(Figure 2). In the first month of dosing, the cumulative of days was maintained through 12 months by 32% of
mean 6 SD percentage of days patients were adherent patients receiving TTFC (12/37) compared with 11% of
with dosing was 79% 6 26% in the TTFC group and patients receiving TRAVþTIM (4/36; P ¼ .028).
67% 6 28% in the TRAVþTIM group. The cumulative
percentage of days through month 12 that patients were SAFETY: The mean 6 SD duration of exposure to study
adherent with dosing was 60% 6 28% and 43% 6 27% medications was 341 6 84 days in the TTFC group and
with TTFC and TRAVþTIM, respectively. 345 6 66 days in the TRAVþTIM group. Mean 6 SD
Cumulative treatment adherence at thresholds of baseline IOP at the postwashout eligibility visit was similar
50%–95% was consistently higher with TTFC compared between treatment groups (TTFC, 26.8 6 4.4 mm Hg;
with TRAVþTIM (Figure 3). Significantly greater per- TRAVþTIM, 25.5 6 4.1 mm Hg). IOP change from base-
centages of patients receiving TTFC demonstrated cumu- line was also similar between groups, ranging
lative adherence with dosing schedules on > _80% of days from 7.7 mm Hg and 8.7 mm Hg at month 1
through treatment months 1, 3, 6, and 12 compared with to 7.4 mm Hg and 7.3 mm Hg at month 12 with
patients receiving TRAVþTIM (P < .001 to P ¼ .041). TTFC and TRAVþTIM, respectively (Figure 4). There
However, in both treatment groups, cumulative adherence were no trends or significant differences between treatment
declined throughout the duration of the study. At 1 month groups with regard to changes from baseline in systolic or
of treatment, 37% of patients in the TTFC group (15/41) diastolic blood pressure or heart rate at any visit.
and 10% of patients in the TRAVþTIM group (4/40) A similar number of patients in each treatment group
achieved adherence at a threshold level of 95% of days experienced AEs, most of which were ocular in nature
(P ¼ .005). Patients receiving TTFC maintained adher- and mild or moderate in intensity (Table 2). The most
ence on > _95% of days through month 3 (25%; 10/40), common treatment-related AE in both groups was ocular
month 6 (21%; 8/38), and month 12 (16%; 6/37), whereas hyperemia (n ¼ 3 per group). Eye pain and eye irritation
no patients receiving TRAVþTIM maintained cumulative were each reported for 2 patients receiving TTFC. Serious
daily adherence above the 95% threshold at these time AEs unrelated to treatment were reported for 3 patients:
100
Cumulative Adherence,
Percentage of Patients 80
79
73
60 67 65
60
56
40 47 43
20
(41) (40) (40) (40) (38) (38) (37) (36)
0
1 Month 3 Months 6 Months 12 Months
FIGURE 2. Cumulative treatment adherence to fixed-combination travoprost 0.004%/timolol 0.5% vs unfixed travoprost
0.004% D timolol 0.5% by patients with open-angle glaucoma or ocular hypertension. Data reflect the percentage of patients consid-
ered adherent through each time point; results are presented as mean and standard deviation. Patient percentages are indicated within
top of bars; group sizes are indicated within bottom of bars.
90
Percentage of Patients, 3 Months
Fixed-combination travoprost 0.004%/timolol 0.5%, n=38 Fixed-combination travoprost 0.004%/timolol 0.5%, n=37
Percentage of Patients, 6 Months
90 Unfixed travoprost 0.004% + timolol 0.5%, n=38 Unfixed travoprost 0.004% + timolol 0.5%, n=36
Percentage of Patients, 12 Months
90
80 80
70 70
60 P=0.037 68
63 60
61
50 55 P=0.006 50
40 47 50 49
45 P=0.003 40 P=0.028
40
30 P=0.005* 30 38
32 32 32
20 28 P=0.025*
20
21
10 16 5 19 19
10 16
0 11 3
0
0 0
0.5 0.6 0.7 0.8 0.9 0.95 0.5 0.6 0.7 0.8 0.9 0.95
Threshold Level of Adherence Threshold Level of Adherence
FIGURE 3. Adherence to fixed-combination travoprost 0.004%/timolol 0.5% vs unfixed travoprost 0.004% D timolol 0.5% by
threshold level for patients with open-angle glaucoma or ocular hypertension at 1 month (Top left), 3 months (Top right), 6 months
(Bottom left), and 12 months (Bottom right). Patient percentages are indicated within bars. P values reflect the results of x2 tests
unless otherwise indicated. *P value from Fisher exact test.
–2
12 months.
–4 The once-daily dosing of TTFC and TRAVþTIM was
–6 the simplest treatment regimen possible (ie, 1 daily instil-
–7.4 –7.4 –7.3
–7.7 –7.7 lation and 2 daily instillations, respectively) and, as such,
–8 –8.7
enabled the most direct comparison of adherence rates
–10
between the fixed and unfixed medications without intro-
ducing additional variables. Compared with patients
–12 receiving TRAVþTIM, patients receiving TTFC were
–14
Fixed-combination travoprost 0.004%/timolol 0.5% adherent with their dosing schedule on a greater percent-
Unfixed travoprost 0.004% + timolol 0.5%
age of days through 1, 3, 6, and 12 months of treatment,
FIGURE 4. Intraocular pressure change from baseline in pa- although adherence was less than ideal. Furthermore,
tients with open-angle glaucoma or ocular hypertension significantly greater percentages of patients achieved
receiving fixed-combination travoprost 0.004%/timolol 0.5% adherence at cumulative thresholds of 80%–95% with
or unfixed travoprost 0.004% D timolol 0.5%. Results are TTFC compared with TRAVþTIM throughout the study.
presented as mean and standard deviation. Mean intraocular Notably, between-group differences in adherence were
pressure changes are indicated within bottom of bars; group sizes observed in the first month of treatment with regard to
are indicated within top of bars. both the percentage of days that patients were adherent
with their treatment regimen and the percentage of
patients achieving higher adherence thresholds. These
syncope (TTFC group, n ¼ 1; severe), hip arthroplasty results were consistent with demonstrations that adherence
(TRAVþTIM group, n ¼ 1; severe), and anemia generally decreases with increasing treatment
(TRAVþTIM group, n ¼ 1; moderate). No patients complexity.11,12,15,16
discontinued the study because of a serious AE. One pa- Fixed-combination therapies simplify treatment admin-
tient in the TTFC group discontinued the study because istration for patients requiring multiple IOP-lowering med-
of 3 moderate treatment-related AEs (eye irritation, ocular ications, potentially increasing adherence with dosing
hyperemia, and photophobia), and 1 patient in the regimens.22 In the current study, adherence declined over
TRAVþTIM group discontinued because of 1 AE unre- time in both treatment groups, consistent with a previous
lated to treatment (alopecia; mild). 6-month assessment of adherence monitored using dosing
We identified no safety issues or trends in either treat- aids,8 but appeared to decrease at a more rapid rate with
ment group based on changes from baseline in visual field TRAVþTIM than with the fixed-combination medica-
tests, gonioscopy evaluations, BCVA assessments, ocular tion. This suggests that although fixed-combination thera-
signs, or fundus parameters. The overall safety profiles of pies with once-daily dosing may promote better adherence
TTFC and TRAVþTIM were similar. over longer periods of time compared with separate instil-
lation of 2 medications at different times of day,12 fixed-
combination dosing does not address all factors that may
play a role in treatment adherence. In addition to treat-
DISCUSSION ment complexity, other factors that may influence adher-
ence include age, income, race, education, health
PATIENT ADHERENCE WITH IOP-LOWERING THERAPY FOR awareness, health literacy, lifestyle, and number of
glaucoma or ocular hypertension is a difficult challenge concomitant eye diseases.23–25 Several studies have
for health care providers.9 Many individuals requiring med- demonstrated that race is an important determinant of
ical therapy for glaucoma are already on multiple systemic treatment adherence.23,26,27 In the current study, most
prescription medications.21 This becomes more complex if patients in both treatment groups were white, and
daily nutraceuticals and other over-the-counter medica- patient race was not significantly different between
tions are considered. Therapies requiring separate instilla- groups. Patient age was also similar between groups.
tion of multiple individual medications increase Nevertheless, important differences in treatment
treatment complexity, which has been associated with adherence were observed with TTFC compared with
decreased treatment adherence. Adherence with treatment TRAVþTIM.
regimens is multifactorial; however, the effect of treatment The IOP-lowering efficacy of TTFC and TRAVþTIM
complexity as a factor in adherence can be minimized with has been demonstrated previously.28–31 IOP change from
use of fixed-combination therapies that provide multiple baseline was not a primary outcome of the current study;
medications in a single formulation. The purpose of this however, both treatment regimens effectively reduced
study was to assess patient adherence with daily glaucoma IOP from baseline. Both treatments were well tolerated,
FUNDING/SUPPORT: THIS STUDY WAS SPONSORED BY ALCON RESEARCH, LTD, FORT WORTH, TEXAS, USA. THE STUDY SPONSOR
participated in study design and data analysis, provided feedback on manuscript drafts, and agreed with the decision to submit the manuscript for publi-
cation. Financial Disclosures: Howard S. Barnebey is a consultant to Alcon, Allergan, Aerie, and Biolight. Alan L. Robin is a consultant to Aerie Phar-
maceuticals, ForeSight Visions, Glaukos, Biolight, and Clearside, and is on the board of the Aravind Eye Foundation. All authors attest that they meet the
current ICMJE criteria for authorship.
Medical writing support was provided by Heather D. Starkey, PhD, of CHC Group, LLC (Chadds Ford, Pennsylvania, USA) and was funded by Alcon.
REFERENCES 12. Robin AL, Novack GD, Covert DW, Crockett RS,
Marcic TS. Adherence in glaucoma: objective measurements
1. American Academy of Ophthalmology Glaucoma Panel. of once-daily and adjunctive medication use. Am J
Preferred Practice PatternÒ Guidelines. Primary Open- Ophthalmol 2007;144(4):533–540.
Angle Glaucoma. San Francisco, CA: American Academy 13. Tsai JC, McClure CA, Ramos SE, Schlundt DG, Pichert JW.
of Ophthalmology; 2010. Compliance barriers in glaucoma: a systematic classification.
2. Peters D, Bengtsson B, Heijl A. Factors associated with life- J Glaucoma 2003;12(5):393–398.
time risk of open-angle glaucoma blindness. Acta Ophthalmol 14. Schwartz G, Burk C, Bennett T, Patel VD. Adherence and
2014;92(5):421–425. persistence with glaucoma therapy: brimonidine/timolol
3. Stewart WC, Kolker AE, Sharpe ED, et al. Long-term pro- versus dorzolamide/timolol and various two-bottle combina-
gression at individual mean intraocular pressure levels in pri- tions. J Clin Exp Ophthalmol 2012;3(8):1–6.
mary open-angle and exfoliative glaucoma. Eur J Ophthalmol 15. Robin AL, Covert D. Does adjunctive glaucoma therapy
2008;18(5):765–770. affect adherence to the initial primary therapy?
4. Heijl A, Leske MC, Bengtsson B, Hyman L, Hussein M. Early Ophthalmology 2005;112(5):863–868.
Manifest Glaucoma Trial Group. Reduction of intraocular 16. Higginbotham EJ, Hansen J, Davis EJ, Walt JG, Guckian A.
pressure and glaucoma progression: results from the Early Glaucoma medication persistence with a fixed combination
Manifest Glaucoma Trial. Arch Ophthalmol 2002;120(10): versus multiple bottles. Curr Med Res Opin 2009;25(10):
1268–1279. 2543–2547.
5. Kass MA, Heuer DK, Higginbotham EJ, et al. The Ocular 17. Rossi GC, Pasinetti GM, Scudeller L, Radaelli R, Bianchi PE.
Hypertension Treatment Study: a randomized trial deter- Do adherence rates and glaucomatous visual field progression
mines that topical ocular hypotensive medication delays or correlate? Eur J Ophthalmol 2011;21(4):410–414.
prevents the onset of primary open-angle glaucoma. Arch 18. Sleath B, Blalock S, Covert D, et al. The relationship
Ophthalmol 2002;120(6):701–713. between glaucoma medication adherence, eye drop tech-
6. Lichter PR, Musch DC, Gillespie BW, et al. Interim clinical nique, and visual field defect severity. Ophthalmology 2011;
outcomes in the Collaborative Initial Glaucoma Treatment 118(12):2398–2402.
Study comparing initial treatment randomized to medications 19. Okeke CO, Quigley HA, Jampel HD, et al. Adherence with
or surgery. Ophthalmology 2001;108(11):1943–1953. topical glaucoma medication monitored electronically: the
7. Miglior S, Zeyen T, Pfeiffer N, et al. Results of the European Travatan Dosing Aid study. Ophthalmology 2009;116(2):
Glaucoma Prevention Study. Ophthalmology 2005;112(3): 191–199.
366–375. 20. Stone JL, Robin AL, Novack GD, Covert DW, Cagle GD. An
8. Rossi GC, Pasinetti GM, Scudeller L, Tinelli C, Milano G, objective evaluation of eyedrop instillation in patients with
Bianchi PE. Monitoring adherence rates in glaucoma patients glaucoma. Arch Ophthalmol 2009;127(6):732–736.
using the Travatan Dosing Aid. A 6-month study comparing 21. Covert D, Robin AL, Novack GD. Systemic medications and
patients on travoprost 0.004% and patients on travoprost glaucoma patients. Ophthalmology 2005;112(10):1849.
0.004%/timolol 0.5% fixed combination. Expert Opin 22. Hollo G, Topouzis F, Fechtner RD. Fixed-combination intra-
Pharmacother 2010;11(4):499–504. ocular pressure-lowering therapy for glaucoma and ocular
9. Reardon G, Kotak S, Schwartz GF. Objective assessment of hypertension: advantages in clinical practice. Expert Opin
compliance and persistence among patients treated for glau- Pharmacother 2014;15(12):1737–1747.
coma and ocular hypertension: a systematic review. Patient 23. Dreer LE, Girkin C, Mansberger SL. Determinants of medica-
Prefer Adherence 2011;5:441–463. tion adherence to topical glaucoma therapy. J Glaucoma
10. Waterman H, Evans JR, Gray TA, Henson D, Harper R. 2012;21(4):234–240.
Interventions for improving adherence to ocular hypotensive 24. Muir KW, Santiago-Turla C, Stinnett SS, et al. Health liter-
therapy. Cochrane Database Syst Rev 2013;4:CD006132. acy and adherence to glaucoma therapy. Am J Ophthalmol
11. Djafari F, Lesk MR, Harasymowycz PJ, Desjardins D, 2006;142(2):223–226.
Lachaine J. Determinants of adherence to glaucoma medical 25. Muir KW, Ventura A, Stinnett SS, Enfiedjian A,
therapy in a long-term patient population. J Glaucoma 2009; Allingham RR, Lee PP. The influence of health literacy level
18(3):238–243. on an educational intervention to improve glaucoma