Cost-Effectiveness and Budget Impact of Cardioform versus Amplatzer and Medical Therapy Alone for Patent Foramen Ovale Closure: A UK Payer Perspective

Abstract

Purpose

This study evaluated the cost-effectiveness and budget impact of two patent foramen ovale (PFO) closure devices and medical management for the prevention of recurrent PFO-associated strokes in the UK.

Patients and Methods

We used a Markov model to simulate 1000 individuals with a history of PFO-associated stroke over a five-year time horizon. Treatment options included GORE^® CARDIOFORM Septal Occluder (GSO Device), Amplatzer™ Talisman PFO Occluder (ATO Device), and Medical Therapy Alone. Effectiveness data were drawn from the REDUCE and RESPECT trials and prior cost-effectiveness analyses (CEAs). No head-to-head clinical trial comparing the ATO and GSO Device was identified; therefore, comparative clinical effectiveness estimates were derived from a matching-adjusted indirect comparison (MAIC). Costs were assessed from a UK payer perspective (2023 GBP). Outcomes included quality-adjusted life-years (QALYs), prevented strokes, incremental cost-effectiveness ratio (ICER), and budget impact.

Results

GSO Device was estimated to be cost-saving and generate 24.66 additional QALYs versus ATO Device (dominant ICER) over five years and in a cohort of 1000 patients. Compared with Medical Therapy Alone, it provided 406.66 additional QALYs at an ICER of £14,343/QALY. It also prevented 27.86 and 67.05 strokes versus ATO Device and Medical Therapy Alone, respectively.

Conclusion

The use of GSO Device is a cost-effective option for the prevention of secondary PFO-associated strokes, with favorable estimated clinical outcomes. Future studies providing direct comparative evidence between GSO and ATO could further inform the relative clinical and cost-effectiveness of available PFO closure devices.

Introduction

Each year, 100,000 strokes occur in the United Kingdom (UK),¹ resulting in 38,000 stroke-related deaths,^2,3 with ischemic stroke comprising 85% of these cases.⁴ Among 1.4 million survivors in the UK,¹ 26% will have another stroke within five years.³ In addition, stroke incidence in the UK is expected to rise 60% from 2015 to 2035, causing the economic burden to increase from £26 billion in 2015 to £75 billion in 2035.⁵

Approximately 25–40% of ischemic strokes are classified as cryptogenic, an entity largely consisting of embolic stroke of undetermined source (ESUS) cases.⁶ Patent foramen ovale (PFO), an incomplete closure of the opening between the atria, present in about 25% of the general population, has been frequently identified as a possible cause of ESUS, as the opening can potentially enable an embolus to enter the systemic circulation and subsequently reach the brain.^7–12 The PFO-Associated Stroke Causal Likelihood (PASCAL) risk stratification system and the Risk of Paradoxical Embolism (RoPE) score are used to estimate the likelihood that a stroke is causally related to a PFO (PFO-associated stroke).^13,14 PFO-associated stroke, therefore, represents a distinct clinical entity within the broader category of cryptogenic stroke.¹³ This study focuses specifically on patients with PFO-associated stroke.

PFO-associated stroke management usually involves Medical Treatment (MT) Alone or in combination with percutaneous PFO closure by a transcatheter device. For patients aged 18 to 60 with a history of ischemic cryptogenic stroke and PFO anatomy, the National Clinical Guideline for Stroke in the UK and Ireland recommends percutaneous PFO closure to prevent recurrent strokes in this patient population.¹⁵ Multiple meta-analyses further support the clinical effectiveness of PFO closure over MT Alone.^16,17 The current uptake of PFO closure in England was assessed by Willits et al,¹⁸ who reported 973 PFO closure procedures for the secondary prevention of stroke or TIA across 20 hospitals between October 2014 and August 2017. This demonstrates that PFO closure is routinely performed in the UK and highlights the relevance of assessing its clinical and economic implications.

The leading PFO closure devices available in the UK market are the GORE^® CARDIOFORM Septal Occluder (GSO Device) from W. L. Gore & Associates, Inc., Amplatzer™ Talisman PFO Occluder (ATO Device) from Abbott, and the Occlutech^® Figulla Flex II Occluder (OFO Device) from Occlutech GmbH. GSO Device and ATO Device are both FDA-approved and have demonstrated efficacy vs MT Alone in the REDUCE and RESPECT trials.^19,20

In contrast, the OFO Device is not FDA-approved and has been assessed primarily in real-world and observational studies until now.^21,22 The results of the ongoing OCCLUFLEX trial (NCT05069558)—which compares the OFO Device to GSO Device and ATO Device—are expected in 2026.²³ In the absence of a head-to-head comparison of GSO Device and ATO Device, a matching-adjusted indirect comparison (MAIC) on the REDUCE and RESPECT trial data provided insights into the relative clinical effectiveness of these devices. The MAIC showed a non-significant trend favoring the GSO Device due to lower recurrent stroke count and fewer serious adverse events (SAEs).²⁴ Specific attention was given to atrial fibrillation, which was found to occur less in patients treated with GSO Device compared to ATO Device (odds ratio (OR): 0.68), though this difference was not found to be statistically significant.²⁴

The cost-effectiveness of PFO closure (both in general and for GSO Device and ATO Device vs MT alone) is well-documented.^25–28 Previous studies comparing ATO and GSO Devices were manufacturer-supported.^24,27,29,30 In the US, GSO Device shows greater cost-effectiveness than the ATO, yielding better patient outcomes and cost savings.²⁹ However, similar data are absent in the UK context. Health economic data are crucial for decision-makers working within limited budgets to evaluate financial implications and guide resource allocation.

This study aims to address this gap by evaluating the comparative cost-effectiveness of GSO Device versus ATO Device, MT Alone, and the OFO Device for secondary PFO-associated stroke prevention in the UK. A secondary objective was to assess the budget impact of increased GSO Device use among eligible patients.

Materials and Methods

Comparator Inclusion

This study aimed to compare the performance of three PFO closure devices, which collectively hold the majority of market share: GSO Device, ATO Device, and OFO Device,¹⁸ as well as MT Alone. MT Alone consisted of antiplatelet therapy or anticoagulation therapy (warfarin), aligning with the REDUCE and RESPECT trials.^19,31 Data on GSO Device and ATO Device devices are frequently and well-reported, making their inclusion in a comparative economic evaluation feasible.^{19,24–27,30} To ensure inclusion of all three relevant devices currently available on the UK market, an additional targeted literature review (TLR) was conducted, focusing on OFO Device, to provide a comprehensive overview of available evidence. Methods and results of the TLR are available in Supplementary Material 1 and Supplementary Table S1. Due to the absence of published randomized controlled trial (RCT) data and missing safety data across other study types, it was not feasible to include OFO Device as a comparator in our economic evaluation.

Study Design

A cost-effectiveness analysis (CEA) was conducted in Microsoft Excel to compare the clinical and economic value of the GSO Device versus the ATO Device and of the GSO Device versus MT Alone for patients with PFO-associated strokes. The CEA compared the costs and health outcomes of the three strategies from a UK payer perspective. A Markov model was used to simulate a cohort of individuals with a history of PFO-associated stroke who were assumed to have comparable baseline characteristics to individuals registered in the REDUCE trial.¹⁹ Data from the REDUCE and RESPECT trials informed the effectiveness of the three strategies,^19,31 together with previous CEAs.^27,30 To minimize confounding from differences in trial populations, we used a published MAIC to assess the relative effectiveness and safety outcomes for the GSO and ATO devices,²⁴ as no direct head-to-head comparison exists. The ISPOR CHEERS 2022 checklist and NICE economic evaluation guidance were integrated into our model’s development and reporting methods.^32,33 The CHEERS 2022 checklist is available in Supplementary Material 2 and Supplementary Table S2.

Model Structure

We built an eight-state Markov model based on previously published CEAs on PFO closure.^25–27,30 The model structure is shown in Figure 1. After experiencing a PFO-associated stroke and receiving MT Alone, GSO Device plus MT, or ATO Device plus MT, patients enter the model in the stable post-index stroke health state and can transition to other states in subsequent model cycles. These include ischemic attack (TIA) and recurrent stroke, accompanied by corresponding post-event health states, as well as death. Recurrent strokes, in this context, refer to clinical ischemic strokes as defined in the REDUCE and RESPECT trials.^19,20 Transitions from TIA, post-TIA, recurrent stroke, and post-stroke to a stable state were not feasible; after a recurrent stroke or TIA, patients move to post-stroke or post-TIA states in the next cycle. Half-cycle correction was applied to account for transitions occurring throughout each model cycle. A portion of patients become disabled after a stroke, needing nursing home care. These patients enter a “tunnel” stroke state in the disabled post-stroke category, which leads either back to the same state or to death. Patients in the disabled post-stroke state cannot transition elsewhere except to the tunnel or death, the latter being an absorbing state.

Figure 1.

Model structure of the Markov model for the cost-effectiveness of PFO closure.

Notes: After having a PFO-associated index stroke, patients are treated with either Medical Therapy Alone, GSO Device in combination with Medical Therapy, or ATO Device in combination with Medical Therapy. Subsequently, patients enter the stable state of the Markov model. *In the model, recurrent strokes refer to clinical ischemic strokes as defined in the REDUCE and RESPECT trials.

Model Inputs

Clinical Inputs

All data sources used for constructing the CEA are detailed in Table 1. We used the results from the REDUCE trial and the MAIC, comparing the REDUCE and RESPECT studies, to inform the relative efficacy of the treatment arms in the model.²⁴ In the absence of TIA-specific data to scale TIA risk between ATO and GSO Devices, we used equivalent hazard ratios for both stroke and TIA, sourced from the MAIC (HR 0.46; 95% CI 0.15–1.43).

Table 1.

Base Case Clinical Input Parameters for the Cost-Effectiveness Model

Efficacy
Safety outcome*	Hazard/odds ratio (95% CI)		p-value			Reference
Risk of recurrent stroke	HR: 0.46** (0.15 to 1.43)		0.17			[24]
Any procedure/device-related SAE	OR: 0.685 (0.314 to 1.490)		0.457			[24]
Transition probabilities
Health state transition	GSO Device	ATO Device		Medical Therapy Alone		Reference
Risk of TIA	0.07%	0.15%**		0.18%		[24,30]
Risk of recurrent stroke	0.11%	0.24%**		0.43%		[24,28,30]
Post-stroke to recurrent stroke	1.14%	1.14%		1.14%		[34]
Risk of disability after stroke	2.30%	2.30%		2.30%		[24]
Stable to death*	0.00%	0.00%		0.00%		[19,20]
Recurrent stroke to death	0.14%	0.14%		0.14%		[34]
Quality of Life
Health state	Utility					Reference
Stable						[30]
Medical Therapy Alone	0.8
PFO closure	0.88
Recurrent stroke	0.578					[30]
Post-stroke	0.506					[30]
Disabled post-stroke	0.406					[35]
Caregiver disutility	−0.112					[36,37]
Transient ischemic attack	0.791					[30]
Caregiver disutility	−0.112					[36,37]
Post-transient ischemic attack	0.791					[30]
Death	0					Assumption

Notes: Efficacy: * The direction of effect of the hazard ratios and odds ratios: GSO Device relative to ATO Device. **We performed an internal analysis, which concluded that the proportional hazards assumption holds (Supplementary Material 6 and Figures S6-1 to S6-3). Transition probabilities: *Analysis of mortality data from the REDUCE and RESPECT trials revealed that reported deaths were unrelated to the device, procedure, or therapy, leading to the decision to exclude additional post-surgery mortality risk in the model to prevent inflation of mortality rates. **Transition probabilities relative to GSO calculated based on hazard ratio for risk of recurrent stroke from the matching-adjusted indirect comparison, also shown above.²⁴

Abbreviations: CI, confidence interval; SAE, serious adverse event; PFO, patent foramen ovale; TIA, transient ischemic attack.

We used annual transition probabilities reported by Amarenco et al to inform our model’s transitions.³⁴ Given our 3-month cycle length, we converted these annual probabilities to three-monthly probabilities (see Supplementary Material 3). Background mortality rates were applied across all health states, and we excluded additional post-surgery mortality risk from the model to prevent exaggerated mortality rates, since the REDUCE and RESPECT trials indicated that reported deaths were unrelated to the device, procedure, or MT. Utility values for all health states are listed in Table 1. For details about the origin of health-related quality of life data, see Supplementary Material 4.

The model applied stable health-state utility values of 0.80 and 0.88 for patients receiving MT Alone and PFO closure, respectively. The incremental utility benefit associated with PFO closure was informed by studies suggesting improved QoL following PFO closure, independently of recurrent stroke reduction.^26,38,39 Specifically, Mirzada et al³⁸ reported significantly better physical functioning, role limitations due to physical health, vitality, and general health among patients undergoing PFO closure compared with non-closure patients (p<0.05). Based on this evidence and following the methodology of Tirschwell et al,²⁶ a conservative 10% utility improvement after PFO closure was assumed.

We included SAEs that occurred after device placement and assigned them to patients at the end of the first cycle of the model. Patients who received GSO Device had an estimated 3.63% chance of an SAE according to the REDUCE trial, while those treated with the ATO device had a 4.21% chance of an SAE based on the RESPECT trial.^19,20 Additionally, we excluded SAEs related to MT, as MT was included across all treatment arms and thus their inclusion would not affect the comparative results. Detailed information about these SAEs and their associated costs can be found in Supplementary Material 5.

Cost Inputs

We included direct healthcare costs in our model, covering hospitalization, diagnostic imaging, MT, institutionalization, and PFO closure procedures to reflect a UK payer perspective. Data on PFO closure included device costs, device placement costs, and follow-up check costs. To accurately estimate the costs of diagnostic tests for PFO-associated stroke, we used NHS Reference Costs.^40,41 Additionally, we included costs of acute treatment and hospitalization, aggregating cost figures from multiple hospital units.^41,42 Post-cryptogenic ischemic stroke costs were extracted from a previous model.³⁰ We inflated all costs to 2023 UK£ using the GDP deflator provided by HM Treasury, as recommended by NHS England.⁴³ We assumed negligible costs for the post-TIA state, a model assumption we validated by discussions with key opinion leaders. A complete breakdown of all costs, including sources, is available in Supplementary Material 7 and Supplementary Table S7. The costs included in the base case analysis are presented in Table 2.

Table 2.

Cost Inputs per Health State

Direct costs – PFO closure
Cost specification	GSO Device		ATO Device		Medical Therapy Alone
PFO closure:	£6881		£6777		£0
- Device costs	£3300		£3168		£0
- Device placement costs	£3222		£3222		£0
- Procedure/device-related SAEs	£61		£88		£0
- Routine follow-up	£298		£298		£0
Direct costs – cost per health state
Cost specification	Cost per unit	Units per year		Units per cycle		Cost per cycle
Stable: £24
- GP visit	£94	1		0.25		£24
Recurrent stroke: £5772
- Diagnostic tests	£1300	-		1		£1300
- Hospitalization and acute treatment	£4472	-		1		£4472
Post-stroke: £1080
-	£4321	1		0.25		£1080
Transient ischemic attack: £4258
- Diagnostic tests	£1300	-		1		£1300
- Hospitalization and acute treatment	£2958	-		1		£2958
Post-transient ischemic attack: £0
-	£0	-		-		£0
Disabled post-stroke: £9475
- Institutionalization	£44,377	0.85		0.21		£9475
Death: £0.00
-	£0	1		1		£0

Notes: All costs are 2023 UK£ and numbers may not add up due to rounding. A full breakdown of all costs, including their original sources, is presented in Supplementary Material 7 and Supplementary Table S7.

Abbreviations: PFO, patent foramen ovale; SAE, serious adverse event; GP, general practitioner.

Analytical Methods

Base Case Model

The base case adopted a UK payer perspective and employed a five-year time horizon, consistent with previous evaluations of GSO, and to limit additional uncertainty associated with extrapolating treatment effects over longer periods.^27,30 We simulated a cohort of 1000 individuals with a starting age of 45 years from the UK with a history of PFO-associated stroke, aligning with previous models and clinical trials.^19,27,30 We discounted both costs and health outcomes with an annual rate of 3.5% and adopted a willingness-to-pay (WTP) threshold of £20,000 per quality-adjusted life year (QALY), as recommended by NICE.³³

Outcomes

Health outcomes were measured in QALYs, life years (LY), and strokes prevented. The incremental cost-effectiveness ratio (ICER) was calculated by combining QALY with cost data and LY outcomes with cost data for competing treatments. We deemed a treatment cost-effective compared to its comparator when the ICER was below the willingness-to-pay (WTP) threshold; lower ICERs indicated greater cost-effectiveness. Moreover, we calculated the incremental net monetary benefit (NMB) to express the value of an intervention in monetary terms using a known WTP per QALY. We included the NMB framework because it offers advantages, including transparent comparison of multiple strategies, the ability to consider various WTP thresholds, and greater stability compared to incremental cost-effectiveness ratios, especially when differences in effectiveness are minor. A positive NMB suggests that the ICER is below the WTP threshold, indicating good value for money. Higher NMB values were preferred because they indicated greater health improvements at lower cost.

Scenario and Sensitivity Analyses

To ensure the robustness of our model and estimate the certainty around our outcomes, we conducted both a probabilistic sensitivity analysis (PSA) and a deterministic sensitivity analysis (DSA). In the PSA, we simultaneously varied 100 model parameters across 1000 iterations. For the DSA, we independently varied 101 input parameters from their base-case values to compare GSO Device versus ATO Device and GSO Device versus MT Alone. In both sensitivity analyses, we varied model parameters within appropriate ranges derived from the standard errors of each variable. When measures of uncertainty were unavailable, we established the range estimates as ±20% of the point estimate for cost data, age data, and transition probabilities, and ±5% for utility values.^27,30

Additionally, we conducted scenario analyses to test critical model assumptions and assess the robustness of the base-case ICER to key parameters within the model. 21 scenarios were examined, ranging from adjusting a single parameter to simultaneously modifying multiple parameters. For example, we incorporated indirect costs to adopt a societal perspective, encompassing productivity losses from stroke or TIA (See Supplementary Material 8 and Supplementary Figure S8-1 to S8-3). We also analyzed the effect of different WTP thresholds or age-dependent stroke risk with a higher starting age on the NMB. An overview of all scenarios is available in Supplementary Material 9 (Supplementary Table S9-1).

Budget Impact Analysis

Scenarios

Budget impact analyses (BIAs) assessed healthcare expenditure changes when increasing GSO use for preventing recurrent stroke in patients with index PFO-associated stroke from a UK payer perspective. Two BIAs compared different scenarios, considering population size, eligibility, market share, and accumulating annual costs for new patients.

The first analysis evaluated the financial impact of shifting market shares by increasing the use of GSO device in combination with MT, while reducing the use of ATO Device. This analysis compared a scenario in which GSO Device in combination with MT is used in 23.7% and ATO Device in 13.0% of the population (new environment) to a scenario reflecting the current environment, where ATO Device is utilized in 23.7% and GSO Device in 13.0% of cases.

The second analysis estimated the budgetary impact of using the GSO as the PFO-closure device in combination with MT. This analysis compared a scenario in which only GSO Device is used for PFO closure (new environment), covering 36.7% of the total population, with the current climate, in which both GSO Device and ATO Device are used, covering 13.0% and 23.7% of the population, respectively. In both scenarios, a similar proportion (63.3%) of the total population is treated with MT Alone.

Input Data

To represent the current treatment landscape, we used the most recent market shares for treatments for PFO-associated stroke in the UK. The market shares of GSO Device and ATO Device were reweighted to encompass the entire market share of PFO closures, as these treatments have sufficient data available for modelling.¹⁸ In the current environment, 63.3% of patients were treated with MT Alone, 13.0% with GSO Device, and 23.7% with ATO Device.

Model Structure

To generate the BIA results, we employed the Markov structure of the CEA. We modeled a real-life scenario using a population of 4811 patients. This size was determined by combining several factors: the annual PFO-associated stroke incidence in the UK (100,000 cases),¹ the percentage of index strokes (71%),⁴⁴ the percentage of ischemic strokes (85%),¹ the percentage of patients younger than 60 years and thus eligible for PFO-closure (25%),⁴⁵ the percentage of cryptogenic strokes in patients younger than 60 (48%),⁴⁶ and the PFO-attributable fraction in patients younger than 65 (66.43%).^14,47 In our BIA, we assumed a consistent yearly inflow of 4811 patients.

Outcomes

The primary outcome of this analysis was the budget impact from a payer perspective, assessed annually and cumulatively over 1 to 5 years. This impact was estimated by calculating the reimbursement the payer paid for medical services provided. Additionally, the number of strokes was reported as a secondary outcome for the compared scenarios.

Results

GSO Device versus ATO Device

Key results for GSO Device versus both comparators are shown in Figure 2A and B, Table 3 and Supplementary Material 10 (Supplementary Table S10). In a simulated cohort of 1000 patients over a five-year time horizon, GSO Device was estimated to be cost-saving and associated with 24.66 additional QALYs versus the ATO Device (dominant ICER). The NMB was £850,415, indicating that the GSO Device was cost-effective at a WTP threshold of £20,000.

Figure 2.

Impact of change in time horizon on the key model results for GSO Device versus ATO Device (A) and GSO Device versus Medical Therapy Alone (B).

Abbreviations: ICER, incremental cost-effectiveness ratio; NMB, net monetary benefit; QALY, Quality-adjusted life year.

Table 3.

Base Case Results of GSO Device versus ATO Device and GSO Device versus Medical Therapy Alone

Specification	GSO	ATO	Medical Therapy Alone	Δ GSO vs ATO	Δ GSO vs Medical Therapy Alone
Incremental Cost-Effectiveness Ratio
ICER (Cost/QALY)				Dominant	£14,343
ICER (Cost/LY)				Dominant	£4,893,772
Cost outcomes
Total costs (in millions)	£7.7	£8.1	£1.9	-£0.36	£5.8
Health outcomes
Total QALYs	4068.63	4043.98	3661.98	24.66	406.66
Total LYs	4647.03	4646.54	4645.84	0.49	1.19
Total number of strokes*	23.9	51.83	91.02	−27.86	−67.05
Total number of days in hospital*	163.59	353.72	603.24	−190.13	−603.24

Notes: All costs are 2023 UK£ and expressed in millions. *Undiscounted.

Abbreviations: ICER, Incremental Cost-Effectiveness Ratio; LY, life year QALY, quality-adjusted life year; TIA, transient ischemic attack.

Several secondary outcomes were assessed in the base case, including LYs saved, strokes prevented, hospital days prevented, and associated costs. Use of GSO Device resulted in 0.49 LYs saved and 27.86 strokes prevented at lower costs (dominant). Consequently, the number needed to treat (NNT) to prevent one stroke with a GSO Device compared to an ATO Device was 36. Patients treated with the GSO Device are expected to spend 190 fewer days in the hospital than those treated with the ATO Device, while also saving costs (dominant). The NNT to prevent one hospital day using the GSO device was six. NNT results for additional time horizons are presented in Supplementary Material 11 and Supplementary Table S11.

GSO Device versus Medical Therapy Alone

In terms of secondary outcomes, GSO Device saved 1.19 LYs and prevented 67.05 strokes at £85,576 per stroke prevented. The NNT to prevent one stroke using GSO Device compared to MT Alone was 15. Patients who received GSO Device were estimated to spend 603 fewer days in the hospital than those receiving MT Alone. Associated costs were £13,051 per hospital day prevented. The NNT to prevent one hospital day using GSO Device was three. NNT results for additional time horizons are presented in Supplementary Material 11 and Supplementary Table S11. Base case results of GSO Device versus ATO Device and GSO Device versus Medical Therapy Alone are presented in Table 3.

Scenario Analyses

21 scenario analyses were performed to assess how changes in key assumptions and parameters affect the model results. When comparing GSO Device and ATO device, these analyses consistently supported the original findings: GSO Device remained cost-effective relative to ATO Device across all scenarios. Compared to MT Alone, two of the 21 scenarios resulted in a negative NMB, indicating that the GSO Device was not cost-effective at a WTP of £20,000 per QALY when the model assumed a shorter time horizon and equal utility values for stable health states between MT alone and PFO closure.

The scenario analyses covered various assumptions, but here we focus on two—10-year and lifetime (until patients reach 100 years of age; 55-year) horizons—that illustrate their impact on cost-effectiveness and reflect real-world situations. These scenarios demonstrated a substantial increase in NMB for GSO Device, as shown in Figures 2A and B. Specifically, the NMB for GSO Device versus ATO Device increased to £2,951,086 at ten years and £20,060,396 over a lifetime horizon. Similarly, for GSO Device versus MT Alone, the NMB increased to £12,222,909 at ten years and £60,480,163 over a lifetime horizon. These results highlight the significant economic advantage of adopting GSO Device treatment over extended periods. Moreover, when applying extended time horizons alongside the increased risk of stroke and mortality with age, GSO Device’s cost-effectiveness further increased. An overview of all scenario analysis results is available in Supplementary Material 9 (Supplementary Table S9-2).

Sensitivity Analyses

Probabilistic Sensitivity Analysis

A PSA was conducted to assess model uncertainty by varying parameters randomly within their respective boundaries. At a WTP threshold of £20,000 per QALY, GSO Device was the cost-effective treatment option in 81.2% of simulations compared to the ATO Device. Against MT Alone, GSO Device was the cost-effective treatment option in 77.7% of iterations with an average ICER of £14,343 per QALY gained.

The results of the PSA are visualized in cost-effectiveness planes (Figures 3A and B) and the cost-effectiveness acceptability curve (CEAC) (Figure 4). The cost-effectiveness planes provide a graphical representation of the incremental costs and QALYs associated with GSO Device compared to ATO Device (Figure 3A) and MT Alone (Figure 3B). Both plots show that most points fall below the dotted line, which represents the WTP threshold of £20,000 per QALY, illustrating the cost-effectiveness of GSO Device against both comparators. The CEAC demonstrates the probability that GSO Device is the most cost-effective option compared to ATO Device and MT Alone across different WTP thresholds. The CEAC indicates that at WTP thresholds below £15,000 per QALY, MT alone is most likely to be cost-effective, and at higher WTP thresholds, GSO Device is most likely to be cost-effective. At the no-WTP threshold, MT Alone is most likely the most cost-effective option.

Figure 3.

Cost-effectiveness plane of GSO Device versus ATO Device (A) and GSO Device versus Medical Therapy Alone (B).

Note: WTP threshold = £20,000.

Abbreviations: PSA, probabilistic sensitivity analysis; QALY, Quality-adjusted life year; WTP, willingness to pay.

Figure 4.

Cost-effectiveness acceptability curve of GSO Device versus ATO Device and Medical Therapy Alone for a 1000-patient cohort over a 5-year time horizon.

Notes: blue = GSO Device; orange = ATO Device; yellow = Medical Therapy Alone.

Deterministic Sensitivity Analysis

The DSA results indicated that in the comparison between GSO Device and ATO Device, key drivers of the model were the hazard ratio (HR) of stroke risk for patients in the ATO Device arm that resulted from the MAIC, device costs of both devices, time horizon, and the HR of TIA risk for patients in the ATO device arm.²⁴ In the comparison between GSO Device and MT Alone, the time horizon, utility value assigned to patients in the stable state in both the MT Alone and PFO closure arm, GSO Device costs, and device placement costs significantly affected the NMB. These results are visualized in tornado diagrams (Figures 5A and B). In the diagrams, a negative NMB corresponds to an ICER above £20,000 per QALY gained and a positive value to an ICER below this threshold.

Figure 5.

Univariate deterministic sensitivity analysis of GSO Device versus ATO Device (A) and GSO Device versus Medical Therapy Alone (B).

Notes: blue = lower bound; red = upper bound.

Abbreviations: PFO, patent foramen ovale; TIA, transient ischemic attack Impact of change in time horizon on the key model results for GSO Device versus Medical Therapy Alone.

Budget Impact

Switching Market Shares of GSO Device and ATO Device

The first BIA assessed the financial implications for UK payers over a five-year time horizon, comparing two scenarios: (1) a new environment, in which more patients are treated with GSO Device than ATO Device, and (2) the current environment, where ATO Device is used more frequently than GSO Device. In both scenarios, 63.3% of the population received MT Alone. The budget impact of shifting market shares is presented in Table 4.

Table 4.

Key Results Budget Impact Analysis

Scenario 1	Total costs	PFO closure	Diagnostics	Hospital and treatment	Long-term care
Current environment (13.0% GSO, 23.7% ATO)	£82.04	£60.18	£1.85	£5.70	£14.30
New environment (23.7% GSO, 13.0% ATO)	£81.66	£60.45	£1.76	£5.45	£14.00
Budget impact *	- £0.37	£0.27	- £0.083	- £0.25	- £0.30
Scenario 2	Total costs	PFO closure	Diagnostics	Hospital and treatment	Long-term care
Current environment (13.0% GSO, 23.7% ATO)	£82.04	£60.18	£1.85	£5.70	£14.30
New environment (36.7% GSO, 0% ATO)	£81.18	£60.74	£1.66	£5.15	£13.63
Budget impact *	- £0.86	£0.56	- £0.19	- £0.56	- £0.68

Notes: All costs are presented in millions. Numbers may not add up due to rounding. *Negative figures represent a favorable outcome due to hospital reimbursements exceeding costs, positive figures represent a net loss for hospitals.

Abbreviation: PFO, patent foramen ovale.

Over five years, assuming an annual incidence of 4,811 new PFO-associated strokes in patients younger than 60 years, the total costs for UK payers were estimated at £82.04 million under the current environment and £81.66 million when treating more patients with GSO Device than ATO Device, resulting in a net budget impact of £0.37 million. The analysis indicated that the higher device costs associated with PFO closure were the only cost category contributing negatively to the budget impact in the new environment. However, this scenario was cost-saving across all other categories, including diagnostics, hospital and treatment costs, and long-term care.

From a clinical perspective, the current market share distribution was estimated to result in a total of 1037 recurrent strokes over five years. In contrast, increasing the use of GSO Device in combination with MT while reducing the use of ATO Device was estimated to reduce recurrent strokes to 995, preventing approximately 42 recurrent strokes over the study period, along with the associated healthcare resource utilization. The budget impact stratified by year is detailed in Table 5. In the first two years, the net budget impact remained positive due to the one-time costs associated with the GSO Device. However, in the subsequent years, cost savings emerged, leading to an overall reduction in healthcare expenditure over the five-year period.

Table 5.

Budget Impact Stratified by Year

Scenario 1	Year 1	Year 2	Year 3	Year 4	Year 5
Current environment (13.0% GSO, 23.7% ATO)	£13.03	£14.40	£16.09	£18.10	£20.41
New environment (23.7% GSO, 13.0% ATO)	£13.06	£14.39	£16.03	£17.97	£20.21
Budget impact *	£0.030	£0.0096	- £0.062	- £0.13	- £0.20
Scenario 2	Year 1	Year 2	Year 3	Year 4	Year 5
Current environment (13.0% GSO, 23.7% ATO)	£13.03	£14.40	£16.09	£18.10	£20.41
New environment (36.7% GSO, 0% ATO)	£13.09	£14.37	£15.95	£17.81	£19.95
Budget impact *	£0.059	- £0.028	- £0.14	- £0.29	- £0.46

Notes: All costs are presented in millions. Numbers may not add up due to rounding. *Negative figures represent a favorable outcome due to hospital reimbursements exceeding costs, positive figures represent a net loss for hospitals.

Using GSO Device for All PFO Closures

The second BIA evaluated the difference in expenditure for UK payers over a five-year time horizon, comparing two scenarios: 1) using only GSO Device as the PFO-closure device in combination with MT (new environment), and 2) using both GSO Device and ATO Device for PFO closures (current environment). In both scenarios, 63.3% of the population received MT Alone. Table 4 presents the key results of the BIA. Over a five-year time horizon with an annual incidence of 4811 new PFO-associated strokes in patients younger than 60 years, the total costs for UK payers were £82.04 million in the current environment and £81.18 million when only using GSO Device plus MT for PFO closures, resulting in a net budget impact of -£0.86 million. PFO closure was the only cost category negatively impacting the budget, but it remained cost-saving in diagnostics, hospital and treatment, and long-term care.

Over a five-year period, treating patients according to current practice was estimated to result in a total of 1037 recurrent strokes. In contrast, when we modeled all patients eligible for PFO closure to be treated with GSO Device in combination with MT, the number of recurrent strokes was estimated to be 943. Therefore, more frequent use of GSO for PFO closure could prevent nearly 94 recurrent strokes over the five years, along with associated resource use.

Discussion

From the UK payer perspective, percutaneous PFO closure using GSO Device for the prevention of recurrent ischemic stroke in patients with PFO-associated stroke proved more cost-effective in comparison to ATO Device and MT Alone. The base case analysis demonstrated GSO Device to be dominant over ATO Device, with an additional 24.66 QALYs at lower costs. Compared to MT Alone, GSO Device was cost-effective with an ICER of £14,343 per QALY. Based on the PSA, there was an 81.2% probability of GSO Device remaining cost-effective over ATO Device and a 77.7% probability of being cost-effective over MT alone at a WTP threshold of £20,000/QALY.

Cost-Effectiveness Analysis

The CEAC from the PSA was cut off at £150,000, as the curve stabilized around a WTP threshold of £100,000, indicating that GSO Device was almost certainly the cost-effective treatment option. The PSA demonstrated GSO Device as the preferred option over ATO Device across a wide range of thresholds, supporting GSO Device as a valuable PFO closure device. ATO Device was not cost-effective compared to the other interventions, primarily because of differences in efficacy rather than costs: GSO devices were more expensive, and device placement costs were equal between GSO Device and ATO Device. Additionally, the CEAC indicated that at WTP thresholds below £15,000 per QALY, MT Alone is most likely the most cost-effective option, based on a cohort of 1000 patients over a five-year time horizon. The high upfront costs of PFO closure mainly drive this result. Nevertheless, GSO Device resulted in significant health improvements by preventing strokes, which outweighed the additional costs at higher WTP thresholds and longer time horizons.

GSO Device was projected to save LYs, prevent strokes, and reduce hospital days compared to both ATO Device and MT Alone. Consequently, the quality of life for patients who received GSO Device, as well as their caregivers, is generally higher. These results were driven by the MAIC-derived hazard ratio for recurrent stroke (HR 0.46; 95% CI 0.15–1.43), which indicated a numerical reduction in recurrent stroke events but did not demonstrate a statistically significant difference. These benefits are further reflected in cost savings, as fewer patients will require subsequent treatment for recurrent strokes or SAEs, including atrial fibrillation, related to the PFO closure procedure or device. However, while our model incorporates the 25 device- and procedure-related SAEs from the RESPECT trial,^20,26 it excludes rare clinical complications such as cardiac erosion, and factors beyond the model’s scope may influence the cost-effectiveness of any of our comparators. Furthermore, even though we used a comprehensive SAE analysis, models inherently simplify reality and cannot account for every scenario.

The cost-effectiveness results were sensitive to the assumed long-term QoL benefit associated with PFO closure. While available evidence suggests potential improvements in QoL following PFO closure,^38,39 the magnitude and durability of this benefit remain uncertain. Sensitivity analyses demonstrated that assuming equal utility values between PFO closure and MT Alone substantially reduced cost-effectiveness, highlighting that the incremental QALY benefit in the model was significantly driven by this utility assumption rather than differences in survival.

GSO Device is estimated to significantly reduce hospital stays compared to both ATO Device and MT Alone, thereby enhancing patient health outcomes and optimizing healthcare resource utilization. This reduction in hospital days could be beneficial for the UK healthcare system, which is under pressure from limited resources and high demand.⁴⁸ Consequently, our outcomes provide valuable insights to aid treatment decisions and ensure effective resource allocation, ultimately alleviating the substantial burden on healthcare providers.

The impact of using GSO Device is expected to be noticeable in the short term, given the favorable NNT values. Specifically, over a five-year time horizon, the NNT to prevent one stroke with GSO Device is 36 compared to ATO Device and 15 compared to MT Alone, and the NNT to prevent one hospital day using GSO Device is six compared to ATO Device and three compared to MT Alone. For the BIA, we estimated the total annual number of candidates for PFO closure in the UK to be 4811.^1,14,44–47

Differences in device design may contribute to differences in procedural characteristics and potential long-term outcomes between PFO closure devices. The GSO Device uses a PTFE-covered structure with two discs designed to facilitate septal apposition and tissue integration, whereas the ATO Device uses a nitinol wire-mesh framework. These differences have been proposed to influence factors such as endothelialization, closure rates, and device-related complications.^49,50 In contrast, wire mesh frame devices are generally stiffer and rely on endothelialization for closure. Wire-mesh frame devices have also been associated with cardiac erosion.⁵¹ The GSO Device is reported to release less nickel compared to ATO,⁵² potentially benefiting nickel-sensitive patients.⁵³ These studies highlight how the differences in device architecture can result in greater clinical outcomes. Potential differences in rare device-related complications, such as cardiac erosion, were not included in the current model due to limited comparative data and may represent an area for future research.

Patient characteristics in the model were aligned with those in the REDUCE trial, but in clinical practice, stroke risk varies substantially by factors such as RoPE score, age, shunt size, and atrial septal aneurysm.⁴⁷ This means that stroke risk may be higher or lower in specific subgroups. For example, in a sample of 402 patients, recurrent stroke or TIA occurred more frequently in patients with a RoPE score <7 compared to those with a score ≥ 7 (HR 3.21, p=0.03).⁵⁴

Sensitivity and Scenario Analyses

We conducted a one-way DSA to understand the impact of individual parameter uncertainties on the NMB estimates. When comparing GSO Device to ATO Device, the analysis revealed that the effectiveness of GSO Device compared to ATO Device had a major impact on the model results, as it translates into favorable quality of life values for patients. Additionally, device costs were an important factor in the model. The results highlight the importance of the selected time horizon, a point further underscored by scenario analyses using different time horizons and their influence on cost-effectiveness outcomes. When comparing GSO Device to MT Alone, the DSA highlighted the strong influence of the utility value for stable patients on model outcomes. Additionally, it accentuated the importance of upfront PFO closure costs. This indicates that improved health outcomes after GSO Device compared to MT, as well as high upfront device costs, significantly influenced the model results.

21 scenario analyses strengthened the robustness of the model findings, as GSO remained cost-effective in all simulated scenarios compared to ATO and in 19 of 21 compared to MT Alone. The scenarios that resulted in a negative NMB accentuate the impact of a shorter time horizon and utility values on the model outcomes, as reflected in the DSA results.

While the base case used a five-year horizon, longer horizons demonstrated that long-term savings outweigh the higher initial cost of GSO. Extending the horizon from five to ten years increased NMB from £2.3 million to £12.2 million for GSO versus MT, while the NNT to prevent one stroke decreased from 15 to 7, and the NNT to prevent one hospital day decreased from 3 to 1. Compared with ATO, the NNT to prevent one stroke declined from 36 to 17, and the NNT to prevent one hospital day declined from 6 to 3.

Budget-Impact Analyses

This study assessed the financial impact of increasing the use of GSO Device for PFO closure through two BIAs. The first examined the effect of switching market share between GSO Device and ATO Device for secondary stroke prevention in patients eligible for PFO closure over five years. In the current treatment landscape, 63.3% of patients receive MT alone, while 36.7% undergo PFO closure (35.5% receive GSO Device and 64.5% receive ATO Device). Adjusting market shares to favor GSO Device resulted in net savings of £370,000, primarily driven by preventing approximately 42 recurrent strokes and their associated costs in a simulated cohort of 1000 patients. While performing a PFO closure procedure with GSO Device has higher upfront costs, its improved safety profile and recurrent stroke prevention offset these initial expenses, yielding long-term economic benefits.

The second BIA explored a scenario in which all patients undergoing PFO closure (36.7%) were treated exclusively with GSO Device, leading to an even greater budget impact saving of £860,000 for UK payers over five years. The only cost increase in this scenario was the one-time expense of PFO closure in the first year, including the device, placement procedure, and related SAEs. However, GSO Device was associated with lower costs across all other categories—diagnostics, hospital and treatment costs, and long-term care—resulting in substantial cost savings over time. These savings were driven by its superior effectiveness in preventing recurrent strokes compared to ATO Device, with an estimated total of 94 strokes prevented over five years, ultimately reducing hospitalizations due to recurrent stroke and TIA and long-term care requirements. These findings highlight the potential to improve PFO closure strategies for better clinical and economic results, emphasizing the need for further research on real-world cost-effectiveness and long-term benefits.

Future Research and Implications

As the use of PFO closure devices continues to expand, future research should address remaining uncertainties, including the lack of direct comparative evidence between devices and limited data on long-term clinical outcomes, safety, and HRQoL following PFO closure. Additional real-world evidence and direct comparative studies between GSO and ATO Devices will be valuable to further inform their relative clinical and economic value.

Although this study did not focus on cardiac erosion, a recent real-world evaluation of GSO Devices using the MAUDE database reported no cardiac erosion incidents from product launch through April 2024. In contrast, rare instances of erosion have been reported with other occluders, such as ATO Device.⁵⁵ There are also a few case reports documenting erosion events with OFO Devices, including a late-onset cardiac tamponade caused by an erosion 5 years after OFO Device implantation,⁵⁶ and a separate series reporting cardiac erosions after closure with the same device.⁵⁷ Further collection and validation of long-term safety data are needed to determine whether device-related complications should be incorporated into future cost-effectiveness analyses.

Our findings may help inform decision-making by UK payers and healthcare providers regarding the potential value of PFO closure strategies for patients with PFO-associated stroke. While uncertainty remains around specific model inputs, these results provide additional evidence to support considerations of the clinical and economic value of PFO closure and should be interpreted alongside current clinical guidelines, local expertise, patient selection criteria, and emerging comparative evidence when informing commissioning and treatment decisions.

Strengths and Limitations

Regarding our CEA, novelty is among its main strengths, as this is the first UK-based study to compare two PFO closure devices and to include new scenarios.^27,30 These encompassed a societal perspective, the application of similar utility values for stroke patients treated with PFO closure versus MT alone, the increasing risk of stroke and mortality with age, and the use of unadjusted transition probabilities for the ATO Device arm instead of adjusted probabilities calculated from the comparative evidence in the MAIC.²⁴ Lastly, the model incorporated insights from previous CEAs comparing the economic value of GSO Device or ATO Device to MT Alone, along with evaluations from national HTA bodies to enhance its robustness.^27,30 These included, for instance, stroke severity as it can lead to disability, caregiver disutilities, and long-term institutionalization in a nursing home. Additionally, robust estimates of SAE-related expenses after PFO closure in a UK context were derived from a previous CEA, based on the SAEs reported in the RESPECT study.²⁰

The CEA has six main limitations. Given that the incidence of TIA, stroke, stroke-related mortality, and SAEs was limited during the REDUCE trial, it was not possible to utilize individual patient data to calculate more robust values for transition probabilities and the incidence of safety outcomes. To address this, we relied on transition probabilities and incidence data derived from available models and literature.^{19,20,24,28,30,34}

Another limitation is our reliance on a published matched-adjusted indirect treatment comparison, given the absence of RCTs directly comparing ATO Device and GSO Device. This MAIC used the REDUCE and RESPECT trial data and objectively compared each of the devices to MT Alone. Thus, the limitations outlined by Kasner et al²⁴ may, by extension, apply to this study and the comparative evidence used in this model should be interpreted with caution. This included reducing the sample size of the REDUCE trial cohort after matching it to the RESPECT trial. Future studies, directly comparing the ATO and GSO device would be desirable to confirm the findings of the MAIC. Nevertheless, this study addressed this uncertainty in the PSA, mitigating its impact on the results.

Furthermore, the favorable base-case results versus ATO are dependent on the MAIC-derived point estimate for recurrent stroke reduction, which did not demonstrate a statistically significant difference, assumptions regarding long-term utility improvements following PFO closure compared to MT Alone. Therefore, the findings should be interpreted in the context of uncertainty surrounding the comparative effectiveness estimates and QoL assumptions, rather than as evidence of substantial differences in survival between treatment strategies.

Another limitation relates to the time horizon of five years and the starting age of 45, which may not fully capture the long-term consequences of PFO closure and medical therapy over patients’ lifetimes. These base-case assumptions were selected to align with previous evaluations of the GSO Device, the characteristics of patients enrolled in the REDUCE and RESPECT trials, and the available clinical follow-up data. The potential impact of longer time horizons was explored in scenario analyses and resulted in more cost-effective results, indicating that a five-year time horizon may potentially underestimate the value of GSO Device.

Lastly, the economic model developed in this study used data from controlled clinical settings and focuses on a sub-population that specifically fits the criteria for PFO-associated stroke in the UK. While using these tailored data strengthens the internal validity of the model, it also limits generalizability. The results may not fully reflect outcomes in broader, real-world populations—either outside the UK or in patients with differing demographics, comorbidities, or stroke characteristics—and rehospitalization rates or other outcomes may vary accordingly.

Despite these limitations, the study offers a robust model incorporating accurate information on the clinical practice of recurrent stroke prevention in patients who were diagnosed with a PFO. In the absence of head-to-head studies comparing GSO Device and ATO Device, this model offers the best available evidence by enabling a comparative CEA between different treatment options for preventing recurrent strokes in patients with a PFO.

To develop a comprehensive and transparent model, we aimed to include the three main PFO closure devices in this study; however, due to the lack of appropriate data for OFO Device, it could not be incorporated into the analysis. The absence of available information on this device, with neither RCTs performed nor safety data for the device being publicly available, not only precluded OFO Device from being included in indirect treatment comparisons but also could raise valid concerns for healthcare providers and patients about the use of this device.

The inclusion of a BIA shows the full potential of increasing GSO use. Nevertheless, it’s not without its limitations. First, multiple sources were used to estimate the population eligible for PFO closure, which may have affected the precision of our population estimates. However, using multiple sources helped mitigate potential biases from individual sources. Second, due to limited data, it was not possible to include all available treatments on the market. Nevertheless, by employing reweighted market shares and a Markov model, we were able to provide a thorough assessment of the financial implications of adopting GSO exclusively in combination with MT. This methodological approach ensured that our analysis is robust and reflective of real-world conditions.

Conclusion

The results of the comparative economic evaluation suggest that the use of GSO Device may represent a cost-effective option for the prevention of secondary PFO-associated strokes compared with ATO Device and MT Alone, while also demonstrating a favorable incremental budget impact. The CEA and BIA demonstrated the upfront costs associated with PFO closure and suggested potential long-term resource savings associated with preventing recurrent strokes.

The analyses were based on the best available comparative evidence and incorporated uncertainty around key clinical inputs. Future direct comparative evidence will further inform the relative clinical and cost-effectiveness of available PFO closure strategies and support decision-making regarding their use in clinical practice.

Funding Statement

The analyses and associated manuscript were conducted by Asc Academics and funded by W. L. Gore and Associates.

Disclosure

JB received grant support from W. L. Gore & Associates. JN, TL, and LFW are employees of Asc Academics. Asc Academics has received consultancy fees for this project from W.L. Gore & Associates. EL, JL, and MN are Associates at W.L. Gore & Associates. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or material discussed in the manuscript apart from those disclosed.