An official website of the United States government
NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.
Poitras V, Wells C, Hutnik C, et al. Optimal Use of Minimally Invasive Glaucoma Surgery: A Health Technology Assessment [Internet]. Ottawa (ON): Canadian Agency for Drugs and Technologies in Health; 2019 Jan. (CADTH Optimal Use Report, No. 8.1b.)
A review of the literature was conducted to identify published economic evaluations and costing studies in patients with glaucoma that may be applicable to the Canadian setting. No identified studies were conducted in a Canadian setting. Most of the studies focused on non-MIGS comparisons, such as pharmacotherapy, laser therapy, and surgical therapies.100–103 While studies were identified that considered separately either the outcomes of costs or QoL associated with MIGS, these did not incorporate the effects of treatment with costs together and, as such, were not fully realized economic evaluations.25,104
Identified studies did not answer the research question posed in this HTA as no published Economic Evaluation has been undertaken to compare MIGS with various glaucoma treatments. In some instances, these studies did provide insight by informing the selection of input parameter values for this Economic Evaluation. The studies that informed this Economic Evaluation are described in more detail in later sections of the report.
The objective of this study was to determine the cost-effectiveness of various types of MIGS procedures, with or without cataract surgery, compared with each other, pharmacotherapy, laser therapy, or filtration surgery, for the treatment of glaucoma in adults.
This analysis was conducted according to a protocol developed a priori,40 and incorporated the findings of a systematic clinical review on the relative efficacy and safety of MIGS, with or without cataract surgery, compared with the above listed comparators. The scope and analytical approach taken in this Economic Evaluation was therefore based on the availability of data identified from the Clinical Review.
According to the CADTH guidelines, the reference case of an Economic Evaluation should be cost-utility analysis with outcomes expressed as quality-adjusted life-years (QALYs).105 Although the clinical condition and its potential treatments have no mortality effects, cost-utility analysis was considered the most appropriate given the different morbidity impacts related to the progression of glaucoma on vision. The main outcome of the Economic Evaluation was incremental cost per QALY gained, reported as the incremental cost-utility ratio (ICUR).
As described in the previous section, it was concluded in the Clinical Review that it was not possible to determine relative effectiveness and safety among MIGS devices given the limited comparisons between alternative MIGS devices. Furthermore, the Clinical Review reported 24 unique pairwise comparisons and, given the heterogeneity between studies, indirect treatment comparison to evaluate the potential treatment effects of multiple interventions was deemed inappropriate. As such, the Economic Evaluation focused on comparing MIGS, as a group of interventions, with other glaucoma treatments to assess the potential cost-effectiveness range associated with MIGS. Furthermore, in alignment with the Clinical Review, only pairwise comparisons were considered as part of the Economic Evaluation. In particular, five broad comparisons by class of treatment were evaluated based on the availability of the clinical data:
Without cataract surgery (models 1 to 3):
MIGS versus pharmacotherapy
MIGS versus laser therapy
MIGS versus filtration surgery (i.e., GDDs or Trabeculectomy).
With cataract surgery (models 4 to 5):
MIGS + cataract surgery versus cataract surgery
MIGS + cataract surgery versus filtration surgery + cataract surgery.
The target population modelled was adults with acquired glaucoma, with an average age ranging from 64 to 72 years old, and reflected the characteristics of patients enrolled in the studies that were identified from the Clinical Review (Table 8).
Baseline Patient Characteristics Associated With Each Model.
As the potential use of MIGS within the treatment pathway for glaucoma is unclear, it is important to define disease severity at baseline to allow modelling patients’ disease progression within each model. While the clinical expert consulted on this project indicated that the staging of glaucoma is evolving, the commonly used Hodapp-Parrish-Anderson grading scale was employed to define disease severity. In this scale, VF scores correspond to the following severity: 1) mild: 0 to −6 dB; 2) moderate: −6.01 to −12 dB; 3) advanced: −12.01 to −20 dB; and 4) severe/blindness: < −20 dB.106
Baseline disease severity in each model was based on the baseline characteristics of patients recruited in the clinical studies. This approach was further validated and confirmed appropriate by the clinical expert. Patients in Models 2 (MIGS versus laser therapy; average baseline VF −5.74 dB)62 and 4 (MIGS + cataract surgery versus cataract surgery alone; weighted average baseline VF −4.13 dB)71,88 reflected those with mild-stage glaucoma and patients started in moderate-stage glaucoma in Model 1 (MIGS versus pharmacotherapy; average baseline VF −6.65 dB)36. For Model 3, as populations with both moderate- (Model 3a; average baseline VF −6.45 dB)64 and advanced-stage (Model 3b; average baseline VF −16.64 dB)63 glaucoma were identified in the clinical studies, both populations were examined separately. Finally, advanced-stage patients were considered in Model 5 (average baseline VF −13.49 dB),85 again reflecting the characteristics of patients recruited in these studies.
We conducted the analysis from the Canadian health care payer’s perspective. Accordingly, direct and indirect medical costs were captured, including device or drug costs, physician fees, operating room (OR), ophthalmologist visits, and tests, as well as complications. Although the original project protocol had noted interest in considering a societal perspective,40 this was omitted in the final report given that the average baseline age of patients in the economic analysis was between 64 to 72 years of age and most would be considered retired.
The time horizon for this analysis was lifetime (up to 95 years old), with a one-year time horizon assessed in sensitivity analysis (to validate against the majority of clinical studies that reported outcomes at one year). An annual discount rate of 1.5% was used per the CADTH guidelines for economic evaluations.105 Rates of 0% and 5% were considered in sensitivity analyses.
A Markov model was developed to estimate the cost-effectiveness of MIGS versus other treatments in adult patients with glaucoma. The cycle length of the model was one month. As previously noted, the model categorized patients into disease severity based on the Hodapp-Parrish-Anderson score. This allowed consideration of patients entering the model at different severities of the disease. Furthermore, as the clinical management and treatment of glaucoma is dependent on the extent of glaucomatous damage, these categories allowed modelling of potential changes in the clinical care pathway of glaucoma over time with respect to vision-related QoL and associated resource use. Health states in the model were defined according to Hodapp-Parrish-Anderson staging106 and were as follows: 1) Mild Stage: 0 to −6 dB; 2) Moderate Stage: −6.01 to −12 dB; 3) Advanced Stage: 12.01 to −20 dB and 4) Severe/Blindness < −20 dB, with an absorbing death state.
A graphical representation of model structure is shown in Figure 8. It was assumed that patients without treatment progressed annually at a rate of −0.6 dB, according to the Early Manifest Glaucoma Trial.107 Once patients progressed to the next stage of severity, they could not reverse in disease severity to a more proximal glaucoma health state. When patients reached advanced-stage glaucoma (< −12dB), Trabeculectomy was performed. For patients who started in the advanced disease stage (models 3b and 5), it was assumed that filtration surgery was the last option, and as such no subsequent treatment was modelled. A summary of subsequent treatments for each model is summarized in Table 9.
Outline of Model Structure and Health States. dB = decibels; IOP = intraocular pressure; MIGS = minimally invasive glaucoma surgery; vs. = versus.
Subsequent Treatments.
As glaucoma and its treatment have long-term consequences, a lifetime model was considered appropriate.105 However, the included clinical studies all reported on surrogate outcomes over a short time period (typically a one-year study duration). Modelling of disease progression and treatment was necessary to project long-term costs and consequences. A one-year model was also conducted in a non–reference-case analysis to explore this.
Mortality rates were obtained from the Canadian Life Tables 2014 to 2016108 and were influenced by age; mortality was assumed to be the same for all VF stages (no increased risk of death with more severe VF deficit). When reported, the average baseline VF was informed by the clinical studies that were used to inform the model (Table 10). The starting age was calculated based on the weighted average from included clinical studies for each model category.
Relative Efficacy (Probability Distribution: Normal) of Reference-Case Models.
Relative treatment efficacy in the economic model was based on the most commonly reported outcomes from the identified studies of the Clinical Review: IOP reduction and medication reduction.
To estimate the rate of glaucoma progression defined by VF, from change in IOP, modelling was necessary. The following approach was taken to derive the relationship between rate of progression and change in IOP. In the Early Manifest Glaucoma Trial, the rate of progression of glaucoma in untreated patients was reported to be −0.05 dB per month (converted to −0.6 dB per year).107 Treatment in this trial (i.e., laser therapy with medication) resulted in an IOP reduction of 5.1 mm Hg with a corresponding reduction in the rate of VF progression from −0.05 dB (baseline) to −0.03 dB per month; this change in IOP corresponded to a reduction factor of 0.6 dB for VF progression (i.e., −0.03 dB and −0.05 dB). The standardized reduction per unit of IOP reduction was then calculated as follows:
Using this association, the change in disease progression from treatment was estimated using the IOP reduction that was reported from the clinical studies. For example, in a study comparing two iStent Injects (2nd generation) versus pharmacotherapy (i.e., Latanoprost + Timolol),36 the annual IOP reduction was 12.2 mm Hg and 11.6 mm Hg, respectively. As such, the annual rate of disease progression with treatment was calculated using the following equation:
Of note, other observational studies have reported on the association between change in IOP and change in rate of VF progression (i.e., standardized reduction per unit of IOP reduction of 0.840, calculated using changes in IOP and VF observed in the Canadian Glaucoma Study);109 this was used to inform sensitivity analysis where the change in IOP resulted in slower disease progression than the reference-case analysis. In another scenario analysis, a faster rate of progression was modelled based on the reported decline in VF in untreated patients (i.e., −0.92 dB per year110).
The transition probabilities in each monthly cycle were estimated as the inverse of the number of months needed for a patient to transition from one health state to the next. For example, the average baseline VF in Model 1 was −6.65 dB (Table 8) and the numbers of months needed to transit from a moderate glaucoma stage to an advanced stage for two iStent Injects would therefore be calculated as:
As such, the transition probability from moderate-to-advanced stage for iStent was 0.28% per month (the inverse of 364 months) or 3.3% per year.
These studies were selected as the reference case in each category of comparison based on the following criteria: a) when meta-analysis was available, the pooled clinical measure was used (Model 4); b) when a statistically significant difference (least conservative estimate) was observed in IOP reduction (Model 3a) or c) when 12-month data were reported (models 3b and 5). For Model 1, Fea et al. 2014 was selected as the reference case with the medication strategy assumed to entail two medications (i.e., average costs of one and three medication therapy).36 Only one study was available for Model 2.62 IOP reduction from the study was applied to the baseline rate of change (−0.6 dB per year and a standardized reduction of 0.905 for every unit of IOP) for each strategy. The remainder of the clinical studies were used to inform the range of relative efficacy that were examined in probabilistic scenario analyses. Relative efficacy from different studies was standardized to 12-month rates to inform model inputs.
To account for the likelihood of attenuating relative efficacy over time, a 10% decline per year in the treatment effect of IOP reduction was assumed after the trial follow-up period for all interventions. Scenario analysis was undertaken assuming no treatment effect attenuation after trial follow-up period.
Relative medication reduction between two treatments was modelled, and the difference in medication use after 12-month follow-up was also assumed to decline 10% per year. For example, in Model 2, where the relative medication reduction at 12 months for Hydrus Microstent versus laser therapy was reported to be 1.4 versus 0.5 (i.e., 0.9 less for laser therapy), an incremental medication cost of 0.9 units was added to the laser therapy group.62
Drug adherence was only considered in Model 1. Adherence rate was not reported in either of the clinical studies available for medication versus MIGS.36,58 As such, the reference case evaluated a range of adherence rates (20% to 95%) suggested by Newman-Casey et al.111 The definition of “adherence” on treatment effects was based on a Canadian RCT that evaluated an educational intervention on glaucoma drug adherence and defined drug adherence as persistence of at least 75% of prescribed medication in one year.112 Assuming similar definition of adherence (i.e., nonadherent patients are less than 75% adherent to their medication), and assuming there is a direct correlation between adherence and IOP reduction (i.e., nonadherent patients achieve 75% IOP reduction), the rate of disease progression for patients not adherent to medication was 25% faster than patients who demonstrated “complete” adherence. Furthermore, in a patient that is nonadherent, it was assumed that 75% of drug use (and cost) would be incurred.112 Note that in the RCTs that inform relative efficacy, medication adherence was not reported; it was assumed that medication adherence in the RCTs represented “100%” adherence (reference case).
AEs from MIGS or surgical interventions were included in the model by applying a one-time AE-related cost and, for major complications or those necessitating a secondary surgical intervention, a disutility within the model’s cycle in which the AEs were expected to occur. Prevalence of AEs was obtained from clinical studies (Table 11).
Adverse Events in Reference-Case Models, Approach to Manage Different Types of Complications and Rates (> 2%).
The baseline utility values for patients with glaucoma were derived from the formula developed by Van Gestel et al. in their discrete event simulation model:113
The coefficients were derived from cross-sectional survey data collected on 531 Dutch patients with ocular hypertension or primary OAG and mapped the impact of VF loss, presence of cataracts, and development of side effects on utility values.113 Utility estimates were based on the Health Utilities Index mark 3 using tariffs for the Canadian population. To estimate state-specific utility values, the midpoint VF in each health state (−3 dB for mild, −9 dB for moderate, −16 dB for advanced, and −26 dB for severe/blindness) was selected. Utility values were further specific to whether patients had cataract or no cataract (Table 12).
Utility Values Per Year for Health States.
Disutility from side effects was assumed to be the same across all treatments in the model and was applied to one cycle within the model.
In the reference case, no disutility was applied to patients on medications. However, the Patients’ Perspectives and Experiences Review has noted that patients find eye drops highly disruptive. As such, a disutility from medication was applied in a sensitivity analysis to explore how this may impact the cost-effectiveness of MIGS versus medication. In this sensitivity analysis, the side-effect coefficient (−0.101) from the previously used equation was applied to patients on medications.
All costs were reported in Canadian dollars and, where appropriate, were inflated to 2018 costs using the Consumer Price Index for all items in Canada.114
MIGS device costs were obtained from a Canadian costing study19 comparing MIGS with medications. For other device costs that were not listed in the literature, the clinical expert was consulted on this review to provide an estimate on these costs. Per the Implementation Issues Analysis review that noted start-up costs for MIGS are generally minimal or are covered by the manufacturers, it was assumed to be negligible.
Medication costs were updated using 2018 prices from Ontario115 and Alberta116 formularies (additional details are provided in Appendix 15). In Model 1, where MIGS was compared with medications, patients were assumed to be on two medications at baseline until subsequent treatment occured. The annual cost of one medication (Alberta: $96) was treated as the unit cost to calculate the cost of relative medication reduction for all models.
Surgeons’ fees and OR costs were respectively obtained from the Schedule of Benefits and Ontario Case Costing Initiative (OCCI) (2016/2017 day surgery)117 in Ontario, and Alberta Medical Association and Interactive Health Data Application118 in Alberta to allow exploration of the potential variability in costs across Canada (and recognizing that specific billing fees for MIGS do not exist in many jurisdictions). Details on physician billing codes for each procedure are document in Table 47, Appendix 14.
In terms of OR costs for Trabeculectomy, it was assumed to be the same as those of a “major eye intervention” (OCCI 2016/17 day surgery117) given that, when converting the procedure code for Trabeculectomy to the Comprehensive Ambulatory Classification System grouper on the OCCI, it was equivalent to a major intervention. OR costs were not available for MIGS; these costs were estimated with reference to the OR costs of cataract surgery, and more specifically for phacoemulsification. The approach to estimate the OR cost were based on separating the proportion of phacoemulsification costs that represented fixed and variable costs. In particular, it has been suggested that OR for phacoemulsification consist of 51% fixed and 49% variable costs.119 Fixed costs were assumed to be identical across all ophthalmological procedures while variable costs were adjusted according the time required to perform the procedure relative to the time required to perform phacoemulsification (approximately 20 minutes). The clinical expert consulted on this review provided insight to the expected procedure durations. Details of the calculation are presented in Table 13. To estimate the OR costs of combined surgeries (i.e., MIGS + cataract surgery or Trabeculectomy + cataract surgery), the variable cost for the second procedure was added to the overall OR cost of cataract surgery.
Detailed Calculation to Determine Operating Room Costs for MIGS or MIGS + Other Surgery.
All secondary surgical interventions (i.e., subsequent Trabeculectomy) were assumed to require the same resource utilization.
Alberta costs (physician fees, OR, and medication) were used in the reference case analysis. Scenario analyses with Ontario costs were performed to explore the impact of a different province’s health care service costs on the results.
Health state costs and utilization, including ophthalmologist consultations and set of tests including vision field test, optic disc imaging, and IOP measurement, were derived from Canadian Glaucoma Guidelines,3 expert opinion, and Canadian sources (Table 14).120,121 Ophthalmologist consultations are recommended at least every four to 12 months, depending on the stage of glaucoma severity. According to the Canadian guidelines, 1.5 sets of tests per year are further recommended for mild state, while two sets of tests are recommended for more severe diseases. Of note, in some jurisdictions, a maximum of two optic disc imagings per year are allowed for billing. Furthermore, one-time costs on low-vision aids (i.e., canes) were assumed to patients in the advanced stage.122,123 Ongoing cost of low-vision services, including low-vision care specialist visits, non-Humphrey Visual Field testing and physical rehabilitation services were also applied to 25% patients in the severe/blind stage (Table 14).121,124
Cost Parameters Used in the Model (2018 Canadian Dollars).
In terms of costing for AEs, the model assumed two additional ophthalmologist consultations for any AEs, and 10% of major complications would require surgical intervention, which was assumed to be equivalent to the cost of minor eye intervention and a physician fee equivalent to paracentesis.
The reference case reflects the probabilistic results based on 1,000 Monte Carlo simulations. The probabilistic results characterize the extent to which parameter uncertainty affects the cost-effectiveness estimates in the model. Standard distributional forms were taken to describe the probability distribution functions relating to input parameters: relative efficacy (relative IOP reduction and relative medication reduction) were characterized by normal distributions, utility and complication rates were characterized by beta distribution, and costs were characterized by gamma distributions. Details on the probabilistic values within each model can be found in Table 49. Probabilistic scenario analyses were further performed to explore uncertainties in specific model inputs or to address known variability in clinical practices. The ranges of plausible values for model parameters were tested in the base-case probabilistic models, and distributions were informed by the reference case or meta-analysis for that comparison (additional details in Appendix 15). Cost-effectiveness acceptability curves demonstrating the probability that a modality would be considered optimal at a given WTP threshold were also presented.
As noted, the specific MIGS device selected for the reference case differed by model based on the available clinical evidence from the Clinical Review (Table 10). Specifically, for models 3a and 4, in which more than two different MIGS devices have been studied and in which the necessary clinical outcome data were available, non–reference-case analyses were conducted for the other MIGS devices that had not been selected for the reference-case. This involved adjusting the effectiveness on IOP and applying the specific device cost to explore the plausible range of ICURs associated with different MIGS devices for that specific comparison.
Across all five models, the following sensitivity analyses were performed:
Comparison of Total Intervention Cost Between Alberta and Ontario (2018 Canadian Dollar).
Specific to Model 1 only, the following additional sensitivity analyses were performed:
The model structure and data inputs were presented to the clinical experts to ensure that the model, its parameters, and its assumptions reflect clinical practice and the available body of literature (i.e., face validity). Internal validity was assessed through a peer review process to ensure the mathematical calculations were performed correctly and were consistent with the model specification.
The reference-case analysis was developed according to the following key assumptions:
The results of the probabilistic reference case for each model are presented in Table 16; demonstrating lifetime expected costs and QALYs, incremental costs and QALYs, as well as the ICUR. Disaggregated lifetime costs from a deterministic analysis are further presented in Table 17. Of note, each model considered a different set of comparators, and may consider different patient populations (Table 8). As such, comparisons of results between models are not appropriate.
Lifetime Probabilistic Analysis: Reference Case.
Disaggregated Lifetime Costs, by Cost Categories (Deterministic Results).
Nonetheless, several overarching findings are present across all five models. First, the lifetime total cost per patient for glaucoma and treatment ranged between $8,431 and $14,621, depending on the treatment strategy and patient’s baseline disease severity. By cost categories, the costs of disease management made up the largest amount of costs (Table 17), with intervention costs being the next most costly. Across the five models, the incremental QALYs between comparators ranged between 0.023 and 0.070, which equated to a difference of approximately eight to 25 days of additional “perfect” health over a lifetime.
The detailed results and sensitivity analyses by each model are subsequently presented. Further, detailed results of the probabilistic results on the incremental cost-effectiveness plane can be found in Appendix 16.
As the Clinical Review only found clinical studies that have compared iStent Inject (2nd generation) to pharmacotherapy, the findings below specifically address the cost-effectiveness of iStent compared with pharmacotherapy. The potential cost-effectiveness of other MIGS devices to a pharmacotherapy strategy remains unclear and could not be explored in the Economic Evaluation.
Over a lifetime horizon in patients with moderate glaucoma, MIGS was associated with an additional $741 in costs, but provided an additional 0.039 QALYs, leading to an ICUR of $18,808 per QALY compared with a pharmacotherapy-based treatment strategy (Table 16). Intervention costs were similar between strategies ($2,380 versus $1,886), although the timing in which costs would be incurred differed. In the MIGS strategy, treatment-related costs occurred at the start of the model due to the high up-front costs associated with surgery, whereas treatment-related costs were constantly incurred in the pharmacotherapy strategy through the model’s time horizon. At lower WTP thresholds, pharmacotherapy was favoured; however, at willingness-to pay thresholds of $18,808 per QALY and above, MIGS was favoured (Figure 9). At a WTP threshold of $50,000 per QALY or $100,000 per QALY, the probability that MIGS was the most likely preferable strategy compared with pharmacotherapy was approximately 60% and 65%, respectively (similar from the Ontario setting, Figure 10). This underscores the uncertainty inherent in this comparison.
Cost-Effectiveness Acceptability Curve for Model 1 (iStent Inject Versus Pharmacotherapy): Reference Case. meds = medications.
Cost-Effectiveness Acceptability Curve for Model 1 (iStent Inject Versus Pharmacotherapy): Ontario Setting Assuming Combined Billing. meds = medications.
In sensitivity analyses, Model 1 was highly sensitive to changes that impacted clinical effectiveness estimates (Table 18). For instance, if assuming that there was no difference in clinical effectiveness between MIGS and pharmacotherapy after 12 months or in adopting a one-year time frame, the ICURs ranged between approximately $291,000 and $9.4 million per QALY gained. The instability in the ICUR was largely due to the very small difference in QALYs that would be estimated in these scenarios as the incremental treatment benefits would accrue only in the first year of treatment.
Sensitivity Analyses, Probabilistic (Model 1: MIGS Versus Pharmacotherapy).
Adherence with medication, as well as disutility to medication use, also impacted the results. The larger the proportion of patients who were nonadherent, the more favourable MIGS procedure appeared as patients who were nonadherent were assumed to have worse clinical outcomes. The disutility associated with medication use, as estimated by Van Gestel, also had a significant impact on effectiveness, with MIGS appearing more attractive, although from a face validity perspective, the disutility from this source (−0.10 [i.e., assumes patients are willing to forego 10% of their time alive to avoid medication use]) was very large.
Arguably, an alternate reference case would be to assume no difference in IOP between comparators given that no statistical analysis was conducted to compare between treatment groups among the included clinical studies identified from the Clinical Review. This non-reference-case analysis assumed identical efficacy between MIGS and pharmacotherapy with the safety profile driving the clinical differences observed between these two interventions. The incremental QALYs between these two treatment strategies were therefore smaller in this scenario, leading to ICURs that were slightly higher than the reference case (ICUR of $27,770 versus $18,808 respectively) (Table 18).
In addition, the model was sensitive to treatment-related costs. When medication costs included markup (7%) and dispensing fees ($12 per bottle), over a lifetime time horizon, MIGS appeared to be cost savings compared with pharmacotherapy. If Ontario costs were used, with a proposed $400 physician billing fee to perform a MIGS procedure (lower than the physician billing costs from Alberta that informed the reference case), the ICUR lowered to $5,173 per QALY gained. MIGS also became the dominant strategy (i.e., MIGS was less costly and more effective) when markup (8%) and dispensing fees (approximately $9 per bottle) were added to the medication costs under an Ontario setting (Table 18).
There are significant areas of uncertainty that preclude definitive conclusions on the optimal treatment strategy for MIGS versus pharmacotherapy. As in all models, the limitations of the underlying clinical efficacy data (surrogate outcomes, short-term follow-up, small number of studies) should be considered. As noted, for this pairwise comparison, only iStent and iStent Inject have been studied in a clinical trial setting, and the economic analysis specifically focused on iStent Inject. The costs associated with MIGS were estimated given that detailed Canadian micro-costing data are lacking, as are specific physician billing codes in many Canadian jurisdictions. Sensitivity analyses highlighted the fact that the cost of MIGS plays a key role in determining its potential cost-effectiveness, and further assessment of actual costs would be valuable as plausible scenarios suggest that there are conditions where MIGS may be cost neutral or even cost saving. Finally, the side effects of medication and adherence are important factors to consider. The reference case assumed that no medication-specific disutilities and incorporated Canadian-reported adherence rates for medication use. However, sensitivity analyses highlighted the fact that, if more conservative assumptions were selected for the pharmacotherapy strategy (e.g., lower drug adherence or disutility applied to being on medication), MIGS may be an attractive treatment option in such populations whereby adherence to medication is expected to be low or if there are considerable side effects experienced on medication.
Similar to Model 1, only one type of MIGS has been directly studied clinically in this comparison between MIGS and laser therapy. Specifically, the Clinical Review identified one clinical study62 that compared Hydrus Microstent with laser therapy (i.e., SLT). As such, the findings below specifically address the cost-effectiveness of Hydrus Microstent compared with SLT and the potential cost-effectiveness of other MIGS devices remains unclear and could not be further explored given the lack of comparative clinical data.
In the reference case, MIGS was associated with additional costs of $1,726, but fewer QALYs (−0.023) than laser therapy. MIGS was therefore dominated by laser therapy (i.e., laser therapy was more effective and less costly) (Table 16). These findings were due to the fact that no significant difference in IOP reduction was noted between Hydrus Microstent and laser therapy groups within the clinical study (see Clinical Review) and, while health state costs were similar between the two comparators (Table 17), the treatment-related costs for MIGS were considerably larger than for laser therapy ($2,380 versus $563) and were not offset by cost savings from reduced medication. The cost-effectiveness acceptability curve (Figure 11) shows that laser therapy was preferred across all WTP thresholds and, at a WTP threshold of $50,000 per QALY, the probability that laser therapy was preferred to MIGS was between 60% and 65%.
Cost-Effectiveness Acceptability Curve for Model 2 (Hydrus Microstent Versus Laser Therapy): Reference Case. Hydrus = Hydrus Microsent.
Across the range of sensitivity analyses performed, the model remained robust as laser therapy remained the dominant strategy (Table 19). Although there were differences in incremental QALYs observed across sensitivity analyses, a strategy consisting of MIGS always remained more costly than laser therapy. As the only clinical study identified in the review did not reveal a statistically significant difference in change in IOP, a non-reference-case analysis was conducted that assumed no difference in IOP between the two comparators. Under this analysis, the expected QALYs observed were similar between strategies with a QALY difference of less than 0.0004.
Sensitivity Analysis, Probabilistic (Model 2: MIGS Versus Laser Therapy).
Cost-Effectiveness Acceptability Curve for Model 2 (Hydrus Microstent Versus Laser Therapy): Ontario Setting Assuming Combined Billing. Hydrus = Hydrus Microstent.
While there is uncertainty regarding the true relative effectiveness of MIGS and laser therapy, the currently available data from one study62 suggested that there was no difference in clinical efficacy with respect to IOP reduction. At current procedure cost estimates, laser therapy strategy was overall less costly even when accounting for the additional medication costs that would be required for laser therapy, reflecting the difference in medication use reported in the clinical study. This finding did not change even when lower cost estimates for MIGS were used based on Ontario billing codes (sensitivity analysis 5). Laser therapy was found to be the preferred strategy across all sensitivity analyses performed.
Models 3a and 3b examined a cohort of patients with either moderate- or advanced-stage glaucoma. MIGS was less costly than filtration surgery; however, it was also less effective. In other words, filtration surgery was found to be more expensive and more effective over a patient’s lifetime and the ICURs of filtration surgery compared with MIGS are shown in Table 16. In both submodels, the primary driver in cost between strategies was the intervention-related costs (Table 17).
The reference case for model 3a (patients with moderate-stage disease) was based on the clinical study by Jea et al64 that compared Trabectome with Trabeculectomy with mitomycin C. Based on this study, the incremental cost of filtration surgery was found to be $703 and the additional benefits was 0.070 QALYs compared with MIGS, leading to an ICUR of $10,093 per QALY gained for filtration surgery versus MIGS (Table 16). The cost-effectiveness acceptability curve indicated that filtration surgery was preferred except when the WTP threshold was less than $10,000 per QALY. Above this value, the probability that filtration surgery was preferred ranged from 50% to 60% and, similar to Model 1, this highlighted the considerable parameter uncertainty to this analysis (Figure 13).
Cost-Effectiveness Acceptability Curve for Model 3a (MIGS Versus Surgery; Moderate Stage): Reference Case. MIGS = minimally invasive glaucoma surgery.
Other clinical studies were identified from the Clinical Review on other MIGS devices for this pairwise comparison. The relative efficacy and safety inputs reported in these studies and the device-specific costs were used to inform the possible range in cost-effectiveness of different MIGS devices. The analysis highlighted the wide uncertainty associated with the economic findings for this pairwise comparison. Although MIGS produced fewer QALYs over a lifetime compared with filtration surgery (in alignment with the Clinical Review findings), the cost difference of MIGS ranged from being more to less costly compared with filtration surgery. As such, the potential cost-effectiveness of different MIGS devices ranged from dominated to being cost-effective compared with filtration surgery depending on the expected cost difference between these treatment strategies (Table 20).
Range of Cost-Effectiveness of MIGS Versus Filtration Surgery, by Different MIGS Devices.
The cost-effectiveness acceptability curves showed that Trabeculectomy was the preferred strategy when compared with iStent Inject (Figure 14). In contrast, XEN45 was the preferred strategy when compared with Trabeculectomy (Figure 15). Of note, the studies reporting the clinical outcomes for alternate MIGS devices that informed the non-reference cases did not provide measures of distribution (e.g., 95% CI, variance). As such, the cost-effectiveness acceptability curves may not truly reflect the level of parameter uncertainty within the analysis as no parameter distribution characterized the clinical efficacy inputs in the model.
![Figure 14. Cost-Effectiveness Acceptability Curve for Model 3a (MIGS [iStent Inject] Versus Surgery; Moderate Stage).](/books/NBK543897/bin/ch6f7.gif)
Cost-Effectiveness Acceptability Curve for Model 3a (MIGS [iStent Inject] Versus Surgery; Moderate Stage). MIGS = minimally invasive glaucoma surgery.
![Figure 15. Cost-Effectiveness Acceptability Curve for Model 3a (MIGS [XEN45] Versus Surgery; Moderate Stage).](/books/NBK543897/bin/ch6f8.gif)
Cost-Effectiveness Acceptability Curve for Model 3a (MIGS [XEN45] Versus Surgery; Moderate Stage). MIGS = minimally invasive glaucoma surgery; XEN = XEN45.
In patients with advanced-stage disease (Model 3b), there was an incremental cost of $3,267 and an incremental benefit of 0.027 QALYs, leading to an ICUR of $121,959 per QALY gained for filtration surgery versus MIGS (Table 16). The cost-effectiveness acceptability curves were similar to Model 3a; however, the WTP threshold at which filtration surgery would be preferred shifted to $100,000 per QALY (Figure 16).
Cost-Effectiveness Acceptability Curve for Model 3b (MIGS Versus Surgery; Advanced Stage): Reference Case. ECP = endoscopic cyclophotocoagulation; MIGS = minimally invasive glaucoma surgery.
For patients with moderate-stage glaucoma (Model 3a), the sensitivity analyses on the reference-case comparison (i.e., Trabectome versus Trabeculectomy with MMC) indicated that filtration surgery remained more expensive than MIGS in all scenarios examined. However, the incremental differences in QALYs did change in some scenarios, leading to fairly dramatic changes in the ICUR from $3.3 million per QALY gained with filtration surgery to filtration surgery being dominant in some cases (Table 21). The instability in the ICUR may be attributed to the very small differences in incremental QALYs observed between treatment strategies.
Sensitivity Analysis, Probabilistic (Model 3a: MIGS Versus Filtration Surgery, Moderate Stage).
Similar to the reference-case analysis, the sensitivity analyses highlight the fact that the model is highly sensitive to the treatment-related costs. From an Ontario setting (Table 15), whereby the total surgical cost associated with filtration surgery was lower than those costs in Alberta, MIGS was overall more costly than filtration surgery. In such cases, MIGS was dominated by filtration surgery (Table 21). Trabeculectomy was therefore the preferred strategy across all WTP thresholds (Figure 17).
Cost-Effectiveness Acceptability Curve for Model 3a (MIGS Versus Surgery; Moderate Stage): Ontario Setting Assuming Combined Billing. MIGS = minimally invasive glaucoma surgery.
For patients with late-stage glaucoma (Model 3b), sensitivity analyses where the incremental effectiveness was reduced — such as using a one-year time horizon or assuming no difference in effectiveness after 12 months — resulted in smaller incremental QALYs estimated and, thereby, the ICUR increased dramatically (Table 22).
Sensitivity Analysis, Probabilistic (Model 3b: MIGS Versus Filtration Surgery, Advanced Stage).
Similar to Model 3a, MIGS became more costly than filtration surgery when Ontario costs were used, as the costs of filtration surgery in Ontario were lower than those in Alberta (Table 10). This resulted in MIGS being dominated (more expensive and less effective compared with filtration surgery) (Table 22).
Model 3b (MIGS Versus Surgery; Advanced Stage): Ontario Setting Assuming Combined Billing. ECP = endoscopic cyclophotocoagulation; MIGS = minimally invasive glaucoma surgery.
This comparison is different than others seen so far. MIGS was less costly but also less effective than the comparator treatment of filtration surgery, which placed MIGS in a different quadrant of the cost-effectiveness plane. As such, the economic results were interpreted differently as the ICUR was calculated for filtration surgery. As with other comparisons, relative efficacy should be considered with caution. Four of the five clinical studies from the Clinical Review (that considered the short-term surrogate outcome of IOP) were not statistically significant or were not compared statistically. This does not establish noninferiority and, from the probabilistic analyses, MIGS generally was found to be less effective than filtration surgery. In addition, both set of models were highly sensitive to treatment-related costs. As such, future studies are required to provide higher quality clinical evidence on MIGS compared with filtration surgery and detailed costing studies are needed to better inform the Economic Evaluation.
As noted, the reference case was based on a meta-analysis conducted as part of the Clinical Review on the Hydrus Microstent device (Appendix 14). In the reference case, MIGS with cataract surgery was more expensive than cataract surgery alone, resulting in incremental costs of $1,641 over a lifetime time horizon (Table 16). MIGS with cataract surgery was also associated with 0.026 additional QALYs, producing an ICUR of $63,626 per QALY gained. The cost category accounting for the greatest difference in costs between strategies was intervention costs (Table 17), which were estimated to be $3,219 and $1,499 for MIGS + cataract and cataract alone, respectively. The cost-effectiveness acceptability curve (Figure 19) indicated that MIGS + cataract surgery became the preferred strategy above a WTP threshold greater than $65,000 per QALY gained.
Cost-Effectiveness Acceptability Curve for Model 4 (MIGS + Cataract Surgery Versus Cataract Surgery): Reference Case. Hydrus = Hydrus Microstent; MIGS = minimally invasive glaucoma surgery; Phaco = phacoemulsification.
As other clinical studies on other MIGS devices exist for this pairwise comparison, analyses were conducted on alternative devices to inform the plausible range in cost-effectiveness for MIGS. MIGS with cataract surgery remained more effective than cataract surgery alone. However, caution is required for the clinical effectiveness estimates for ECP and CyPass Micro-Stent as these were from single studies. The estimated QALY difference was larger when clinical efficacy came from single studies than when the clinical efficacy was based on a meta-analysis of multiple studies (i.e., Hydrus Microstent and iStent). Furthermore, as the incremental costs were higher for MIGS with cataract surgery compared with cataract surgery alone, this resulted in a range of ICURs for MIGS from $5,984 to $108,934 per QALY gained (Table 23). The cost-effectiveness acceptability curves highlight the variability in the cost-effectiveness findings between different MIGS devices across a range of WTP thresholds (Figure 20 and Figure 21).
Range of Cost-Effectiveness of MIGS With Cataract Surgery Versus Cataract Surgery Alone, by Different MIGS Devices.
![Figure 20. Cost-Effectiveness Acceptability Curve for Model 4 (MIGS [iStent Inject] + Cataract Surgery Versus Cataract Surgery).](/books/NBK543897/bin/ch6f13.gif)
Cost-Effectiveness Acceptability Curve for Model 4 (MIGS [iStent Inject] + Cataract Surgery Versus Cataract Surgery). MIGS = minimally invasive glaucoma surgery; Phaco = phacoemulsification.
![Figure 21. Cost-Effectiveness Acceptability Curve for Model 4 (MIGS [ECP] + Cataract Surgery Versus Cataract Surgery).](/books/NBK543897/bin/ch6f14.gif)
Cost-Effectiveness Acceptability Curve for Model 4 (MIGS [ECP] + Cataract Surgery Versus Cataract Surgery). ECP = endoscopic cyclophotocoagulation; MIGS = minimally invasive glaucoma surgery; Phaco = phacoemulsification.
![Figure 22. Cost-Effectiveness Acceptability Curve for Model 4 (MIGS [CyPass Micro-Stent] + Cataract Surgery Versus Cataract Surgery).](/books/NBK543897/bin/ch6f15.gif)
Cost-Effectiveness Acceptability Curve for Model 4 (MIGS [CyPass Micro-Stent] + Cataract Surgery Versus Cataract Surgery). MIGS = minimally invasive glaucoma surgery; Phaco = phacoemulsification.
For the majority of the sensitivity analyses, the model’s overall findings remained robust (Table 24). Similar to the other models, the model was most sensitive to changes that impacted the clinical effectiveness estimates. In sensitivity analyses where there was no difference in effectiveness between the two strategies after 12 months or where either a one-year time frame was considered, the ICURs ranged between approximately $377,804 and $3,384,115 per QALY gained. The analyses from an Ontario setting were found to be similar to the reference case, regardless of the approach to billing (Table 24); the cost-effectiveness acceptability curve (Figure 23) indicated that MIGS + cataract surgery would be the preferred strategy above a WTP threshold greater than $70,000 per QALY gained.
Sensitivity Analyses, Probabilistic (Model 4: MIGS + Cataract Surgery Versus Cataract Surgery Alone).
Cost-Effectiveness Acceptability Curve for Model 4 (MIGS + Cataract Surgery Versus Cataract Surgery): Ontario Setting Assuming Combined Billings. Hydrus = Hydrus Microstent; MIGS = minimally invasive glaucoma surgery; Phaco = phacoemulsification.
In this analysis, it is clear that treatment-related costs should be greater as an additional procedure (MIGS) is being performed in addition to cataract surgery. There was more clinical evidence available to inform this analysis and, despite its limitations that are common across the other models (e.g., short-term studies using a surrogate outcome); the model was found to be robust to the majority of the sensitivity analyses. The ICUR of MIGS, as a class, added to cataract surgery compared with cataract surgery alone fell within a range between $5,984 and $108,934 per QALY. A limitation with this analysis, that is common across all models, is that it only considered two comparators. Clinical interpretation of these findings must therefore consider the place in therapy between these two comparators along the trajectory of the disease (for example, should filtration surgery with cataract surgery be the most relevant comparator, the findings from Model 5 would be more relevant).
Filtration surgery with cataract surgery was more expensive, but also more effective than MIGS with cataract surgery (Table 16). With incremental costs of $473 and incremental benefits of 0.032 QALYs, this led to an ICUR of $14,968 per QALY gained for filtration surgery + cataract surgery compared with MIGS + cataract surgery. Similar to other comparisons, health state costs formed the largest proportion by cost category although incremental differences within this cost category were small (Table 17) with the major cost driver between strategies being related to intervention costs.
The cost-effectiveness acceptability curve indicated that filtration surgery with cataract surgery was the preferred therapy except when the WTP threshold was approximately below $15,000 per QALY. Above this WTP value, filtration surgery with cataract surgery was preferred in approximately 55% of the iterations (Figure 24).
Cost-Effectiveness Acceptability Curve for Model 5 (MIGS + Cataract Surgery Versus Surgery + Cataract Surgery): Reference Case. MIGS = minimally invasive glaucoma surgery; Phaco = phacoemulsification.
Filtration surgery + cataract surgery was more expensive than MIGS with cataract surgery in most scenarios. As there was no statistically significant difference in IOP reduction reported in the clinical studies, if no difference in IOP reduction was assumed, the incremental QALYs became very small between the two comparators. The differences in costs as well as effectiveness observed in sensitivity analysis led to quite varied results with a wide range of ICURs for filtration surgery with cataract surgery.
When Ontario costs for both surgeries were used, filtration surgery with cataract surgery was less costly as surgical costs were lower in Ontario compared with Alberta. As such, across all WTP thresholds, filtration surgery with cataract surgery was the preferred strategy in the Ontario setting (Figure 25).
Cost-Effectiveness Acceptability Curve for Model 5 (MIGS + Cataract Surgery Versus Surgery + Cataract Surgery): Ontario Setting. MIGS = minimally invasive glaucoma surgery; Phaco = phacoemulsification.
Sensitivity Analyses, Probabilistic (Model 5: MIGS + Cataract Surgery Versus Filtration Surgery + Cataract Surgery).
As with Model 3, MIGS (with cataract surgery) was less costly but also less effective than filtration surgery with cataract surgery, which placed MIGS in a different quadrant on the cost-effectiveness plane. As such, the ICUR was calculated for filtration surgery (standard of care) compared with MIGS.
The sensitivity analyses highlight the significant uncertainty in the clinical effectiveness due to the quality of underlying studies (see Clinical Review). While in general, MIGS with cataract surgery was found to be a less costly strategy, further evidence demonstrating equivalent or superior clinical outcomes are needed to better ascertain the likely cost-effectiveness of MIGS with cataract surgery compared with filtration surgery with cataract surgery.
While the Clinical Review identified many clinical studies that compared the efficacy and safety among MIGS and comparators, there was a lack of high-quality data, and studies were highly heterogeneous, and as such the evidence was not definitive. Pairwise comparisons were used in the reference case as network meta-analysis was not available for any models, and there was high heterogeneity in MIGS devices and patient populations. Specific criteria were used to select the reference case, with sensitivity analysis conducted on the potential range of cost-effectiveness.
Among all models, the incremental difference in QALYs was relatively small over a lifetime time horizon. In the reference-case models, the difference in QALYs among comparators equated to between eight and 25 additional days of perfect health. This can lead to instability of the ICUR if the denominator becomes quite small, as it did in many sensitivity analyses; for example, when equal clinical efficacy was assumed for comparisons that were not statistically significant or in which statistical comparisons were not conducted.
It is notable that the underlying clinical data and evidence that informed difference in effectiveness (and subsequent QALYs) were generally of poor quality, as noted in the Clinical Review. Many of the studies did not demonstrate statistically significant differences in the surrogate outcome of IOP reduction. Further, there is uncertainty in the precise relationship between changes in IOP and its ultimate impact on VF status and vision-related QoL over time. It is notable that many of the differences in incremental QALYs occur over a long time period — when only a one-year time horizon was used, the incremental QALYs were very small among all models. Most of the clinical studies considered outcomes at one year only, and the long-term relative efficacy of alternative treatment strategies is unknown. As such, estimated differences in QALYs should be interpreted with caution as the incremental QALYs were generated over the extrapolation of a lifetime. Ideally, adequately powered studies using clinically important outcome measures should be conducted.
The incremental differences in costs over a lifetime time horizon were also relatively small. In the reference-case model, the differences ranged between approximately $473 and $3,267 per patient. Unlike QALYs, these incremental costs tended to occur relatively early (with the exception of medication costs), largely due to the initial costs of the operation, procedure, and device that occur within the first year. Costs were largely driven by the intervention costs as well as costs due to patients being in different health states, although the incremental differences for health state costs tended to be low between comparators. Other cost categories, including complications and medical costs in general, were relatively small compared with the intervention costs. The intervention costs for filtration surgeries in Ontario were relatively lower compared with those in Alberta (Table 15); thus leading to differences in cost-effectiveness of MIGS. The differences and range of costs reflect both uncertainty (for example, in the true cost of MIGS, or the physician fee that will be used in jurisdictions where one does not exist) as well as variability in costs that may occur between settings and jurisdictions. As such, the attractiveness of MIGS compared with alternative strategies from a cost perspective may differ depending on the cost of the device, the procedure, as well as the physician’s claims. Detailed micro-costing of MIGS procedures may allow greater certainty in the true absolute and incremental costs of MIGS.
In summary, this cost-effectiveness analysis considered MIGS versus alternative therapies in patients with varying stages of glaucoma within the disease trajectory. Definitive conclusions on the attractiveness of MIGS from a cost-effectiveness perspective are precluded given the uncertainty in relative efficacy and cost. However, there are some scenarios where MIGS may be attractive, should emerging evidence be supportive. Complexity remains with respect to determining what patients are candidates for all procedures; for example, there may be subgroups of patients with very aggressive glaucoma who may not be clinically appropriate for all treatments. Key areas that may assist the determination of cost-effectiveness include conduct of detailed micro-costing studies of MIGS and comparator interventions, assessment of the impact of medication adherence on disease progression and relative effectiveness (Model 1), and determination of relative effectiveness using clinically important and relevant outcomes. Finally, this analysis examined pairwise comparisons, but did not examine scenarios where there may be multiple treatment options for a patient; this may require further clarification and inquiry prior to recommendations for the optimal use of therapies in patients with glaucoma.
Copyright © CADTH. This report may be reproduced for non-commercial purposes only and provided that appropriate credit is given to CADTH.
Except where otherwise noted, this work is distributed under the terms of a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International licence (CC BY-NC-ND), a copy of which is available at http://creativecommons.org/licenses/by-nc-nd/4.0/
Your browsing activity is empty.
Activity recording is turned off.
See more...
