Within-Trial Cost-Effectiveness of an Adherence-Enhancing Educational Intervention for Glaucoma

Abstract

Purpose: To assess the within-trial cost-effectiveness of a behavioral intervention to improve glaucoma medication adherence.

Design: Prospective cost-effectiveness analysis of randomized, controlled trial data.

Methods: The study setting was a Veterans Affairs (VA) eye clinic. The patient population comprised veterans with medically treated glaucoma and self-reported poor adherence. Participants were randomized to a personalized educational session with a reminder bottle to promote medication adherence or to a control session on general eye health. Costs were assessed from the perspective of the VA payor at 6 months using the VA Managerial Cost Accounting System. Probabilistic sensitivity analyses were conducted using bootstrapped samples. The main outcome measures were the proportion of participants attaining ≥80% adherence as measured by electronic monitor, total intervention and medical resource costs, and incremental cost-effectiveness ratios comparing intervention to control at 6 months.

Results: Of 200 randomized participants, 95 of 100 assigned to the intervention and 97 of 100 assigned to the control had adherence outcomes at 6 months, and the proportion of adherent patients was higher in the intervention group compared to control (0.78 vs 0.40, P < .0001). All participants had costs at 6 months. The total cost at 6 months was $1,149,600 in the intervention group (n = 100) compared to $1,298,700 in the control group (n = 100). Thus, in a hypothetical cohort of 100 patients, the intervention was associated with cost savings (-$149,100) and resulted in 38 additional patients achieving medication adherence.

Conclusions: An adherence-enhancing behavioral intervention was effective and cost saving at 6 months.

FIGURE 1.

Cost-effectiveness scatter plot. Probabilistic sensitivity analysis showing incremental cost-effectiveness ratios (ICERs) from 1000 bootstrapped samples. The incremental effectiveness (y axis) represents the number of additional patients who achieved adherence in the intervention compared to control. The incremental cost (x axis) is the difference in total cost between the intervention and control groups. All points are greater than 0 on the y axis, indicating that the intervention resulted in more adherence than the control. Points plotted left of the $0 line on the x axis demonstrate that the intervention was cost saving compared to the control. Points plotted right of the $0 line on the x axis are cost-effective if the ICER is less than or equal to how much a payor is willing to pay for increased adherence (Figure 2).

FIGURE 2.

Cost-effectiveness acceptability curve. Cost-effectiveness curve plotted by willingness to pay threshold for each additional patient achieving adherence. For example, if the payor were willing to pay $5000 for 1 more patient to achieve 80% medication adherence, the intervention would be cost-effective 85% of the time in the first 6 months after starting the intervention.

FIGURE 3.

Cost-effectiveness scatter plot by companion status. The intervention was more effective than the control for all bootstrapped samples (ie, all points on the y axis are greater than zero). Each point left of the x = $0 line indicates cost savings. Almost all of the companion subgroup (light green) falls to the left of this line, demonstrating both better adherence outcomes and cost savings. For the no-companion subgroup (dark green), many points fall to the right of the x = $0 line, indicating that the intervention cost more than the control. Whether the intervention is cost-effective compared to control for patients with no companion thus depends on the willingness to pay threshold for increased adherence (Figure 4).

FIGURE 4.

Cost-effectiveness acceptability curve by companion status. Cost-effectiveness acceptability curve by companion and no-companion subgroups. Even at a willingness to pay threshold of $0, the intervention is almost 100% cost-effective for the companion subgroup, because it is cost saving. The cost-effectiveness of the no-companion subgroup is dependent on the willingness to pay threshold.

FIGURE 5.

Cost-effectiveness scatter plot for once-daily vs more than once-daily dosing. The intervention led to better adherence than the control for both the once-daily and more than once-daily subgroups (ie, all y axis values are greater than zero). The once-daily subgroup had a larger incremental effect than the more than once-daily subgroup (greater values on the y axis). Incremental cost was similar for both populations, with greater variability in the more than once-daily subgroup. With greater incremental effectiveness at a similar cost, the intervention is more cost-effective for the once-daily subgroup than for the more than once-daily subgroup.

FIGURE 6.

Cost-effectiveness acceptability curve for once-daily vs more than once-daily dosing. The once-daily subgroup has a higher probability of being cost-effective for a given willingness to pay threshold compared to the more than once-daily subgroup.