Cost-Utility Analysis of a Medication Adherence-Enhancing Educational Intervention for Glaucoma

Abstract

Objective:

To evaluate the cost utility of a glaucoma medication-enhancing intervention compared to standard of care over a lifetime from the United States Department of Veterans Affairs (VA) payer perspective.

Design:

Model-based cost-utility analysis of a glaucoma medication-enhancing intervention from a randomized clinical trial.

Subjects:

Veterans with glaucoma, or suspected glaucoma who were prescribed topical glaucoma medications, had their visual field assessed within the last 9 months, and endorsed poor glaucoma medication adherence.

Methods:

Veterans were randomized either to a behavioral intervention to promote adherence or to a standard of care (control) session about general eye health. A decision analytic model was developed to simulate lifelong costs and quality-adjusted life years (QALYs) for an intervention tested in a randomized clinical trial at a single VA eye clinic. Costs included direct medical costs that the VA payer would incur, as informed initially by the clinical trial and then by published estimates. Health-state quality of life was based on published utility values. Scenario analyses included addition of booster interventions, a 3% decline in chance of staying medication adherent annually, and the combination of the two. Analyses were also conducted in the following subgroups: those with companion versus not, and those with once-daily versus more than once-daily dosing frequency.

Main Outcome Measures:

Incremental cost-effectiveness ratio (ICER).

Results:

Compared to standard of care, the intervention dominated resulting in lower costs ($23 339.28 versus $23 504.02) and higher QALYs (11.62 versus 11.58). Among the 4 subgroups, the intervention dominated for 3 of them. In the fourth subgroup, those with more than once-daily dosing, the ICER was $2625/QALY. Compared to standard of care, an intervention with booster interventions led to an ICER of $3278/QALY. Assuming both a 3% annual loss in chance of continuing to be adherent and addition of booster interventions, the ICER increased to $71 371/QALY.

Conclusions:

From a VA payer perspective over a lifetime, the glaucoma medication-enhancing behavioral intervention dominated standard of care in terms of generating cost savings and greater QALYs.

Financial Disclosures:

Proprietary or commercial disclosure may be found after the references.

Glaucoma is a leading cause of blindness, and its prevalence is growing worldwide. In 2013, 64.3 million people were diagnosed with glaucoma, and this number is projected to double by 2040.¹ Glaucoma is also costly, incurring an estimated $2.86 billion loss each year in the United States through both direct costs and productivity losses.²

Glaucoma medications that lower intraocular pressure can slow the progression of disease.³ However, adherence to medication is often sub-optimal due to barriers such as cost,^4-6 inadequate patient engagement and understanding of the disease or medication,⁷ dosing frequency,⁸ and eye-drop administration technique.⁹ A recent survey also reported poor medication adherence due to the following specific reasons: forgetfulness, falling asleep before dosing time, difficulty administering the eye drops, pain or discomfort when taking the eye drops, and suffering from side effects when taking the eye drops.¹⁰

To improve medication nonadherence, an intervention that included glaucoma education, personalized disease management suggestions, and a smart bottle reminder aid was developed. This intervention was tested in a randomized trial of 200 Veterans with glaucoma and was compared to a control arm with a standard of care educational session on general eye health (without a smart bottle reminder aid). After 6 months, the intervention increased medication adherence compared to control.¹¹ Specifically, the mean proportion of patients who achieved adherence (defined as taking at least 80% of medication doses on time, a threshold that is associated with relative preservation of visual fields [VFs]¹²) was higher in intervention arm compared to the control (0.78 versus 0.40, P < 0.0001).¹³

A prior study found that adherence to glaucoma medications, compared to nonadherence, is a cost-effective strategy.¹⁴ A subsequent study evaluated the within-trial cost-effectiveness of the glaucoma medication adherence-enhancing intervention, but based on what was available from clinical trial data, was limited to a 6-month time horizon and measured cost-effectiveness in terms of cost per additional patient able to achieve medication adherence.¹³ The objective of this study is to build on prior literature and estimate the cost utility (measured as cost per quality-adjusted life year [QALY]) of a glaucoma medication adherence-enhancing intervention compared to standard of care from the Department of Veterans Affairs (VA) payer perspective over a lifetime horizon.

Methods

Study Design

A cost-utility analysis was conducted to compare costs and QALYs over a lifetime in the intervention arm versus standard of care (control) arm from a randomized controlled trial conducted at a VA eye clinic.¹¹ A decision analytic model using microsimulation was adapted from a previously published and validated glaucoma economic model.¹⁴

Clinical Trial

This single-site randomized controlled trial has been previously summarized¹¹ and registered at clinicaltrials.gov (NCT03052257), and the approval of the Institutional Review Board/Ethics Committee of the Durham VA Medical Center was obtained. The described research adhered to the tenets of the Declaration of Helsinki. Informed consent was obtained from all participants. Briefly, 200 veterans with medically treated open-angle glaucoma were randomized 1:1 to either the intervention or control group, and randomization was stratified based on whether the prescribed medication was dosed once daily or more frequently as well as whether a companion was enrolled with the participant. Enrolled veterans were older (mean age of 67 vs. 68 for intervention vs. control), mostly male (96% vs. 91%), and majority Black (67% vs. 78%; Table 1). Visual field severity was spread relatively similarly across mild, moderate, and severe. A little over half were using medications that were dosed more than once daily, and only a fifth had a companion at randomization.

Table 1.

Veteran Characteristics

	Intervention (n = 100)	Control (n = 100)
Age, Mean (SD)	67.9 (8.6)	67.1 (8.1)
Male, N (%)	96 (96)	91 (91)
Race, N (%)
Black	67 (67)	78 (78)
White	30 (30)	21 (21)
Other	3 (3)	1 (1)
Hispanic, N (%)	10 (10)	4 (4)
Employment Status, N (%)
Employed	21 (21)	22 (22)
Unemployed	6 (6)	6 (6)
Disabled	19 (19)	20 (20)
Retired	51 (51)	49 (49)
Other	3 (3)	3 (3)
Marital Status, N (%)
Married/Living Together/Committed	63 (63)	70 (70)
Relationship
Divorced/Separated/Widowed/Single/Missing	37 (37)	30 (30)
Financial Status, N (%)
Less Than 30K	25 (25)	23 (23)
30–70K	45 (45)	48 (48)
Above 70K	17 (17)	19 (19)
Missing	13 (13)	10 (10)
Visual Field Severity, N (%)
Mild	31 (31)	23 (23)
Moderate	28 (28)	29 (29)
Severe	35 (35)	37 (37)
Indeterminate/NA	6 (6)	11 (11)
History of Glaucoma, N (%)
6 mos to less than 1 yr	2 (2)	4 (4)
1 to 2 yrs	12 (12)	7 (7)
More than 2 yrs to less than 5 yrs	30 (30)	19 (19)
5 yrsor more	51 (51)	63 (63)
Don’t Know	5 (5)	7 (7)
Number of Glaucoma Medications, Mean (SD)	1.77 (0.76)	1.81 (0.77)
Doses Per Day, N (%)
One Dose Per Day	44 (44)	43 (43)
More Than One Dose Per Day	56 (56)	57 (57)
Companion Status, N (%)
Companion at Randomization	20 (20)	21 (21)
No Companion at Randomization	80 (80)	79 (79)
NA = not available; SD = standard deviation.

The intervention consisted of an ophthalmic technician engaging the participant in a discussion around glaucoma, addressing participant-specific barriers to adherence, observing eye-drop administration technique and recommending drop aids when needed, and providing a written schedule of drops. Finally, participants in the intervention arm received a smart bottle with reminders (e.g., chime, flash, text message, or call depending on user preference) within 2 hours of the prescribed dosing time if the bottle had not been opened.

In the control group, participants attended a slide presentation similar in duration to the intervention visit but on general eye health instead of being focused on glaucoma. Participants were also provided a smart bottle but without the reminder function active.

The primary outcome of this trial was medication adherence, measured 6 months after randomization. Seventy-eight percent of the intervention arm had a mean medication adherence of at least 80% (i.e., “medication-adherent”), while 40% of the standard of care control group were medication-adherent (Table 2).¹⁵

Table 2.

Model Inputs.

Parameter	Value	References
Effectiveness: Proportion who were medication-adherent 6 mos after randomization
Intervention	78%	15
Standard of care (control group)	40%	15
Costs
Intervention and standard of care costs
Intervention	$526	15
Standard of care (control group)	$28	15
Office Visits and Procedures
Level IV Office Visit, New Patient (CPT 99204)	$169.98	47
Gonioscopy (CPT 92020)	$28.33	47
Pachymetry (CPT 76514)	$11.60	47
Visual Field Testing (CPT 92083)	$65.60	47
Fundus Photography (CPT 92250)	$128.88	47
Optical Coherence Tomography (CPT 92134)	$338.30	47
Level IV Office Visit, Return Patient (CPT 99214)	$132.90	47
Level III Office Visit, Return Patient (CPT 99213)	$81.33	47
Low Vision Services, New Patient (CPT 99205)	$212.31	47
Occupational Therapy, Initial Evaluation (CPT 97166)	$321.18	47
Occupational Therapy Home Visits (CPT 97530 & CPT 97535)	$75.92	47
Trabeculectomy with Anti-metabolite (CPT 66170)	$1157.44	47
Trabeculectomy Revision (CPT 66185)	Not available, assumed same as above based on equal facility charges	48
Tube Shunt Implant with graft (CPT 66180)	$3523.36	47
Laser Trabeculoplasty (CPT 65855)	$348.37	47
Facility Fees for Glaucoma Surgery		48
Laser Trabeculoplasty (CPT 65855)	$2793.70	48
Trabeculectomy (CPT 66170, 66185)	$8075.24	48
Tube Shunt Implant (CPT 66180)	$14 650.17	47
Treatment of complications from trabeculectomy, tube shunt implant, or laser trabeculoplasty (CPT 66250)	$841.94	47
Medication Costs
Latanoprost 2.5 mL	$1.49	49
Timolol 0.5% 10 mL	$11.90	49
Brimonidine 0.2% 10 mL	$1.46	49
Dorzolamide 2% 10 mL	$3.99	49
Moxifloxacin 3 mL	$8.50	49
Prednisolone 10 mL × 2 bottles	$26.94	49
Ketorolac tromethamine 0.5% 5 mL × 2 bottles	$9.50	49
Atropine 2 mL	$15.40	49
Additional Low Vision Care Costs
Low Vision Rehabilitation for those with ≥1 Eye Blind	$1969/yr	35, 36
Utilities
Mild Glaucoma	0.92	50, 51
Moderate Glaucoma	0.89	50, 51
Severe Glaucoma	0.86	50, 51
Single-Eye Blindness	0.47	52
Bilateral Blindness	0.26	52
CPT = current procedural terminology.

Model

In the model, a hypothetical cohort of patients was randomly assigned to either the intervention or control arms to mimic the clinical trial (Fig 1). Based on their baseline disease severity (as informed by the clinical trial), these patients entered a Markov model at different health states, such as mild, moderate, or severe glaucoma, which assumed: mild disease patients were taking 1 medication (prostaglandin analogue) or 2 medications (beta blocker as the second medication); moderate disease patients were taking 3 (alpha agonist as the third medication) or 4 medications (carbonic anhydrase inhibitor as the fourth medication); and severe disease patients were to undergo surgery followed by ongoing medication management. Aligned with the clinical trial, patients entered the model at age 67 and were followed until either death or age 100, whichever came first. Similarly to the original model and in accordance with best practices,^16,17 this model accounted for probability of death at a given age based on United States census life tables.¹⁸

Figure 1.

Model structure. Intervention and control arms entered the disease progression model based on disease severity (e.g., mild, moderate, and severe). Mild disease entered at 1 or 2 medications, moderate disease entered at 3 or 4 medications, and severe disease entered at surgery. At each health state, medication adherence and whether disease was stable or worsened was tracked in annual cycles. The figure on the right represents the disease progression model, in which each bubble represents a distinct health state. Straight-line arrows represent unidirectional state transitions and looped arrows represent being able to remain in a particular health state for later years in the model. RCT = randomized clinical trial.

Throughout the model, treatment, treatment adherence, disease progression, resource utilization, and outcomes of patients in each health state were tracked annually and were based on Monte Carlo microsimulations using 10 000 iterations per strategy. In the first 6 months, the likelihood of being adherent to medication was based on clinical trial estimates. Thereafter, similar to the original model,¹⁴ if patients were adherent to their medication, they were assumed to continue to be adherent and have similar outcomes as those in the treatment group in the UK Glaucoma Treatment Study based on empirical studies.^5,19-24

Similar to the original model, the classification of health state was based on the mean deviation in the worse seeing eye and glaucoma states were defined according to a modified Hodapp–Anderson–Parrish classification. Disease severity was based on Humphrey VF test values–mild (<−6 dB mean deviation), moderate (−6 to −12 dB mean deviation), severe (−12 to −20 dB mean deviation), and blindness (>−20 dB mean deviation).²⁵ The odds of disease progression determined the decibel reduction in each eye for a given patient in each year of the model. Based on the UK Glaucoma Treatment Study, the annual progression was a median of −0.8 dB/year²⁶ among those whose disease progressed and −0.1 dB/year among those who were stable and had mild and moderate glaucoma.²⁷ Once an individual transitioned to severe glaucoma, a −1.6 dB/year²⁸ rate was assumed. Beta blockers²⁹ and alpha agonists³⁰ were assumed to have similar effectiveness to prostaglandin analogues, whereas carbonic anhydrase inhibitors were 70% effective.³¹ Data from the Glaucoma Laser Trial,³² the Primary Tube versus Trabeculectomy Study,³³ and the Tube versus Trabeculectomy Study³³ were used to inform probabilities of worsening disease among treated patients. The same utilities for each health state as from the original model were used (Table 2). The original model was also validated in terms of incidence of unilateral and bilateral blindness as compared to the population-based study in Olmsted County³⁴ and in the Advanced Glaucoma Intervention Study.²⁵

Costs

Given the VA payer perspective, the costs included in this model were focused on costs that the VA payer would incur, such as direct medical costs and intervention costs (Table 2). The intervention and standard of care educational session costs included labor costs and indirect costs to the VA to provide the intervention and standard of care educational sessions, as well as the cost of the smart bottle (as priced for a VA research program), eye-drop aid, and eye drops for demonstration. Office visits and procedure costs were estimated using VA community care fee schedules. Medication costs were based on Federal Supply Schedule costs. The VA provides a variety of low vision rehabilitation care services. We assumed that Veterans with at least 1 blind eye would receive low vision rehabilitation, with cost estimates from a previously published clinical trial³⁵ that were inflation-adjusted using the Consumer Price Index for medical care to 2021 U.S. dollars to match all other costs.³⁶

Sensitivity and Subgroup Analyses

Several deterministic sensitivity and subgroup analyses were conducted. In addition to examining a lifetime horizon, we also examined 5-year and 10-year time horizons. In threshold analyses, we varied the time horizon to determine the points in time at which the intervention became cost-effective based on $100 000/QALY and $150 000/QALY thresholds as well as the point in time at which the intervention became cost-saving. In one-way sensitivity analyses, we varied the parameter values for 7 different model parameters, including key effectiveness parameters (i.e., the proportion of the intervention group and the proportion of the control group who were medication-adherent at 6 months after randomization), key cost parameters (i.e., the intervention and standard of care costs), and key utility parameters (i.e., mild, moderate, and severe glaucoma). In scenario analyses, we first varied the assumption that once adherent to a medication, a patient would forever be adherent to a medication and instead applied a 3% decrease in annual chance of staying adherent in the medication-adherent group. The 3% decrease was determined based on extrapolating from clinical trial data in the first 6 months, so could be an overestimate in terms of predicting longer-term medication adherence, especially given a prior 4-year study showing that those who were adherent over the first year were likely to be adherent over the subsequent 4 years.⁵ Nonetheless, we included this as a scenario analysis to provide us with a conservative cost-utility estimate. Next, we added a booster intervention so that whenever a patient progressed to an additional medication, if they were non-adherent to their medications, then they received another 6-month intervention with the same cost and effectiveness assumed as the initial intervention. Lastly, as a third scenario analysis, we combined the 3% decrease in annual chance of staying adherent and a booster intervention, as a worst-case scenario.

The clinical trial a priori stratified randomization by whether a companion was present (companion versus no companion) and by dosing frequency of the medication (once daily versus more than once daily). Therefore, in subgroup analyses, we also evaluated the cost utility of intervention versus control within these subgroups.

TreeAge Pro 2018 R2.0 (Williamstown, MA) was used to build the decision analytic model and perform analyses.

Results

Over a lifetime, the intervention dominated the control group resulting in lower costs ($23 339.28 versus $23 504.02) and higher QALYs (11.62 versus 11.58; Table 3). When reducing the time horizon from over a lifetime to 5 and 10 years, the incremental cost-effectiveness ratio (ICER) increased to $513 330/QALY and $98 375/QALY, respectively. Threshold analyses found that between years 8 and 9, the ICER crossed the $150 000/QALY threshold ($155 126/QALY in year 8 and $120 325/QALY in year 9). Between years 9 and 10, the ICER crossed the $100 000/QALY threshold ($98 375/QALY in year 10). Between years 15 and 16, the ICER crossed from being positive to negative (“dominating”), indicating that the intervention became cost-saving ($1409/QALY in year 15 and -$5604/QALY in year 16).

Table 3.

Cost Utility of Glaucoma Medication Adherence-Enhancing Intervention Compared to Control.

	Intervention Group Costs, $	Control Group Costs, $	Intervention Group QALYs	Control Group QALYs	ICER ($/QALY)
Base Case (lifetime horizon)	23 339.28	23 504.02	11.62	11.58	Dominated
5 yrs	9558.20	8992.95	4.04	4.04	513 330
10 yrs	14 204.95	13 567.51	7.12	7.12	98 375
Booster Intervention	23 184.17	23 094.12	11.62	11.59	3278
3% loss in likelihood of staying adherent	24 222.56	23 953.52	11.58	11.56	16 686
3% loss in likelihood of staying adherent + Booster Intervention	24 072.17	23 579.54	11.54	11.55	71 371
Companion	23 408.54	23 769.26	11.62	11.57	Dominated
No Companion	23 373.50	23 575.46	11.62	11.58	Dominated
Once-daily dosing	23 033.22	23 560.27	11.62	11.59	Dominated
More than once-daily dosing	23 679.95	23 627.46	11.60	11.58	2625
ICER = incremental cost-effectiveness ratio; QALY = quality-adjusted life year.

Figure 2 shows that the proportion of patients in more progressive disease states was greater in the control group (panel B) versus in the intervention group (panel A). In particular, patients more quickly progressed to moderate disease (orange line in Fig 2) in the control group versus the intervention group, with slight differences seen as early as year 4 (33% versus 32% with moderate disease in the control versus intervention groups). The greater accumulation of patients with moderate disease then led to a larger proportion of patients more quickly progressing to severe disease (gray line in Fig 2). By year 12, 38% of patients in the control group had progressed to severe disease, as compared to 32% of patients in the intervention group.

Figure 2.

Model-based disease severity over time among intervention versus control patients. A, Health-state transitions for intervention subjects. B, Health-state transitions for control subjects.

Across all one-way sensitivity analyses, we found that the ICER was “dominating,” indicating that our results were robust to variation in model input parameter values. In scenario analyses, if a booster intervention were provided for non-adherent patients when they progressed to needing an additional medication, the ICER for the intervention group compared to the control group was $3278/QALY (Table 3). Assuming a 3% annual loss in chance of continuing to be adherent, the ICER increased to $16 686/QALY. Assuming both a 3% annual loss in chance of continuing to be adherent and a booster intervention, the ICER increased to $71 371/QALY.

Across 3 of the 4 subgroups (i.e., those with a companion, those without a companion, and those with once-daily dosing), the intervention group dominated the control group and led to reduced costs as well as increased QALYs over a lifetime (Table 3). In the fourth subgroup, those with more than once-daily dosing, the ICER was $2625/QALY.

Discussion

This study found that over a lifetime, this adherence-enhancing glaucoma intervention was a dominating strategy compared to standard of care, meaning that the intervention both increased QALYs as well as reduced costs. This is an important finding because often health care interventions and technologies cost extra and thus, healthcare payers must decide whether they are willing to pay the extra cost for the extra clinical benefit. However, in our case, the upfront costs of the intervention were offset by downstream cost savings, which led to an overall reduction in lifetime costs compared to standard of care. Although the cost savings ($165 per patient) were less than 1% of the mean total medical costs over a lifetime for a glaucoma patient in the intervention group ($23 339 per patient), this was also approximately 30% of the upfront intervention costs ($526 per patient) and, importantly, represents savings. Similarly, the average lifetime difference in QALYs between intervention and control group patients was only 0.04 QALYs gained; however, this can represent a myriad of scenarios: (1) approximately 2 additional weeks of perfect health, (2) a longer period of time in less-than-perfect health (such as approximately 8 weeks for a person with single-eye blindness), or (3) the same amount of time but at better quality of life. It is also important to note that both the cost savings and QALY gains represent average values across the patient population, so bigger differences in costs and QALYs can be seen by individuals, and the average values result from a smaller proportion progressing to more severe disease stages in the intervention versus control groups. In fact, much of the gains in QALYs occurred later in life because the decrease in quality of life is greater as glaucomatous damage and vision loss progresses to blindness, which explains the higher ICERs when examining shorter time horizons, such as 5 and 10 years.^37,38 In essence, the upfront investment in terms of medications and treatments in earlier disease stages pays out in the long-term when there is avoidance of, and less time spent in, lower-quality health states alongside cost savings.

Although the 6-month clinical trial did not provide booster interventions when patients became medication non-adherent, we simulated booster interventions in a scenario analysis because it is possible that over a lifetime, patients would undergo another intervention to increase the chance of becoming medication-adherent. Booster sessions have been used to augment educational interventions for adherence to antiviral medications³⁹ and exercise therapy, and these follow-up encounters could have a potential role to promote adherence to glaucoma medication.⁴⁰ Assuming that the VA payer were willing to pay the more conservative $50 000 per QALY, the intervention with booster interventions would be considered cost-effective. However, the VA payer would need to be willing to pay $71 371 per QALY in order to decide to cover an intervention with booster interventions in a worst-case scenario where 3% of medication-adherent patients became non-adherent annually. In this worst-case scenario, the combination of both needing more frequent intervention and losing some effectiveness when not everyone is adherent (with the latter as the primary driver) is what leads to a less cost-effective intervention. However, there has been discussion and an increasing trend of updating the $50 000 per QALY threshold to $100 000 per QALY or $150 000 per QALY,^41,42 and given the lower of these thresholds (e.g., $100 000 per QALY), the intervention would still be cost-effective.

Among clinically important subgroups based on companion status and dosing frequency, the intervention led to cost savings and greater QALYs over a lifetime compared to standard of care or was highly cost-effective, with an ICER far below the conservative threshold of $50 000 per QALY. Cost savings and QALY gains were greatest in the subgroup with a companion over a lifetime model, as they were in the 6-month clinical trial.¹⁵ Having a companion at clinical visits is associated with greater patient comfort and facilitates patient expression, which are features that may increase engagement and efficacy of the behavioral adherence intervention.⁴³ Compared to standard of care, the intervention led to cost savings for all subgroups except those with more than once-daily dosing, largely due to a slightly lower intervention effectiveness.¹⁵ This is not unexpected, as more complicated regimens, such as those with more than once-daily dosing, are well-known barriers to medication adherence.^9,44-46 Nonetheless, in this subgroup, the intervention was still considered highly cost-effective, given that the ICER ($2625 per QALY) was far below the conservative $50 000 per QALY.

A consideration of patient barriers to medication adherence that were not addressed in this intervention is also important. For example, a recent study found that patients with glaucoma were more likely to report cost-related medication non-adherence than those without glaucoma.⁴ Medication costs and healthcare expenses are likely unique for the VA population compared to other Americans with glaucoma who have other forms of healthcare coverage. Thus, in addition to educating patients of the importance of adhering to their glaucoma medications and equipping them with skills to appropriately instill their eye drops, consideration of other cost-related factors is important. Helping to address cost as a barrier to adherence by choosing an affordable medication on each person’s prescription insurance plan formulary is important to optimizing medication adherence.

Limitations

Our results rely on our model assumptions as well as the model input parameter estimates. We relied heavily on both the clinical trial estimates as well as published literature to inform various model input parameter estimates. The model was also previously validated against published longitudinal studies to assess how well the model predicted disease progression over time.^25,34 Results could change if the analysis was able to differentially assess the cost-effectiveness of the intervention among both slow and fast progressors. However, we were unable to add this layer of complexity to the overall model given limited data on the proportion of the population who were fast progressors and whether the effectiveness of the intervention would vary for fast progressors. Assuming similar intervention effectiveness, we would hypothesize that the intervention would be even more cost-effective in a fast-progressor sub-cohort because the cost-effectiveness of the intervention improved as we moved from a 5-year to a lifetime horizon, as cohort members were given the time to progress to more severe disease stages. Future research delineating the proportion of the population of patients with glaucoma who are fast progressors could help provide the evidence that would enable studies to compare cost-effectiveness of various interventions in a more realistic population that included both fast and slow progressors.

One key assumption related to the intervention effect in the base-case model was that once patients were adherent to their medications from the intervention at 6 months, we assumed that these patients would continue to be adherent to their medications thereafter. While adherence persistence has been supported by a prior study⁵ and mimics the assumption from the original model,¹⁴ we also included a scenario in which the chance that medication-adherent patients continued to be adherent declined by 3% each year as a conservative estimate based on 6-month trends seen in the clinical trial. It is certainly possible that it could take some time to become used to taking eye drops every day and so early 6-month medication adherence trends seen in the clinical trial would not be expected to continue for many years later. We therefore included the 3% decline as a conservative estimate to model waning adherence over time.

Given the selective trial population and focus on the VA, our study results are limited in generalizability to similar patients using this glaucoma medication-enhancing intervention within a VA context. Since the mean age of our trial participants was 67, we aligned the hypothetical cohort for our model to have a starting age of 67. However, had we started the age of the cohort earlier (such as 40 in the original model¹⁴), it is likely that we would have seen even larger differences in the costs saved and QALYs gained between the intervention and standard of care arms. Thus, our approach was more conservative.

In conclusion, from a VA payer perspective over a lifetime, the glaucoma medication-enhancing intervention dominated standard of care in terms of generating cost savings and greater QALYs. This finding was consistent across clinically important subgroups such as those with companion versus not and those with once-daily dosing frequency.

Abbreviations and Acronyms:

ICER

incremental cost-effectiveness ratio

QALY

quality-adjusted life year

VA

Veterans Affairs

VF

visual field