Cost-effectiveness analysis of intravascular ultrasound-guided peripheral vascular interventions in patients with femoropopliteal peripheral artery disease

Introduction

Peripheral arterial disease (PAD) is a chronic atherosclerotic disease of the lower limbs, characterized by symptoms such as rest pain, intermittent claudication, and disability [1]. The global prevalence of PAD per 100,000 persons was estimated to be 1,401.8, and is expected to rise due to population aging [2]. In 2018, up to 2.3 million people in Germany were affected with PAD [3].

Endovascular interventions, such as percutaneous transluminal angioplasty (PTA), have emerged to be the treatment of choice in most patients with femoropopliteal (FP) occlusive lesions [4]. Conventional PTA for peripheral lesions is guided by angiography to provide 2-dimensional images of the affected vessel and its surroundings [5]. However, angiography has been associated with limitations, including visualization extent of anatomical features, which may impact the interpretation of angiographic images [6].

In recent years, use of adjunctive intravascular ultrasound (IVUS) during revascularization for PAD patients has been growing [7]. However, use of IVUS in peripheral interventions has remained suboptimal primarily due to low supporting evidence. However, a recently published randomized controlled trial showed the ability of IVUS to reduce target lesion revascularization in peripheral artery disease [13]. Observational data has suggested that the use of IVUS can improve long-term outcomes [7].

In addition, reimbursement, and financing mechanisms for IVUS vary amongst countries, and have posed challenges to its adoption. In Germany, reimbursement based on current diagnostic-related group (DRG) codes for peripheral interventions does not sufficiently cover the costs associated with adjunctive IVUS use. Economic evaluations are increasingly used by health technology agencies to quantify the value of health technologies to aid in healthcare decision-making. A recent review of the current literature revealed that there is a lack of model-based economic evaluations of IVUS-guided PTA in PAD patients. Furthermore, there have been only few published economic models studying the trajectory and outcomes of PTA for PAD [8, 9, 10].

This health economic analysis aimed to propose a plausible model to investigate the cost-effectiveness of endovascular interventions in PAD patients and establish the cost-utility of IVUS-guided PTA over angiography-guided PTA. The results may inform German clinicians, hospital administrations, and policy makers on the more cost-effective treatment approach in PAD patients with FP lesions.

Material and methods

Model schematics and analytical framework

A two-step model consisting of a one-year decision tree, followed by a lifetime semi-Markov model was developed to reflect the clinical trajectory of PAD patients with FP lesions PAD (Figure 1). The one-year decision tree reflected immediate events after the index PTA and captured the following mutually exclusive clinical outcomes: all-cause mortality, target lesion revascularization (TLR), amputation, and alive with no further event. Depending on the health state at the end of the first year, patients transitioned into subsequent health states in the lifetime semi-Markov model.

The semi-Markov model consisted of health states that a patient could occupy at any given timepoint and reflected the clinical trajectory of a PAD patient. State occupancies were defined by transition probabilities that were time-dependent over a set of discrete time periods (cycles) and associated with different costs and utility weights. Patients who were alive and had not experienced any events in the first year entered the “Alive” state. Patients who experienced a TLR in the first year entered the “Post-TLR” state, while those who had an amputation entered the “Amputation” state.

The semi-Markov model consisted of the following health states: alive, TLR, post-TLR, amputation, and death with a cycle length of one year:

• •
  “Alive”: Patients had undergone and survived the index PTA. They could remain in this state, or transition to “TLR”, “Amputation” or “Death” states.
• •
  “TLR”: Patients had undergone an index PTA and now required a TLR. Following which, patients could transition to the “post-TLR”, “Amputation” or “Death” states.
• •
  “Post-TLR”: Patients had undergone and survived a TLR following the index PTA. They could either remain in this state, or transition to either the “Amputation” or “Death” states.
• •
  “Amputation”: Patients in this state required an amputation that may occur at any time point. They could either remain in this state, or transition to the “Death” state.
• •
  “Death”: This was the absorbing state for the model.

This simplified trajectory was considered reasonable for the model due to a lack of data for probabilities for downstream events and to maintain model parsimony. Half-cycle correction was applied for the semi-Markov model to account for events that may occur within each one-year cycle.

Model inputs

Transition probabilities

One-year decision tree

The probabilities for events occurring in the one-year decision tree were derived from Allan et al, a prospective, single-center randomized controlled trial (RCT) that investigated the effect of IVUS imaging during endovascular interventions for PAD patients with FP disease (Table I) [13]. The probability of not experiencing any event (i.e. alive with no further event) was the complement probability of the summation of all three events (i.e. alive with no further event, TLR, or amputation). The primary outcome was binary restenosis within 12 months of the index procedure. Secondary outcomes of interest included TLR, defined as a repeat procedure to treat originally treated lesion, amputation, and mortality. One-year rates of these three events were used in this model.

Table I Model parameters

Parameter	Value (Range*)	Distribution	Source
Notes. *Range of values used in sensitivity analyses. IVUS: intravascular ultrasound; PTA: percutaneous transluminal angioplasty; TLR: target lesion revascularization; UC: usual care (angiography-guided PTA).
Overall model assumptions
Age	70	NA	Malyar et al. [16]
Proportion, males	65.2%
Discount rate, cost	3.0%		IQWiG [11]
Discount rate, outcomes	3.0%
Clinical parameters year decision tree
Probability, TLR, IVUS	15.8% (12.6%–19.0%)	Beta	Allan et al. [13]
Probability, TLR, UC	17.6% (14.1%–21.1%)
Probability, amputation, IVUS	2.6% (2.1%–3.2%)
Probability, amputation, UC	2.7% (2.2%–3.2%)
Probability, all-cause mortality, IVUS	2.6% (2.1%–3.2%)
Probability, all-cause mortality, UC	5.4% (4.3%–6.5%)
Lifetime semi-Markov model (1 year)
Probability, TLR, IVUS	3.4% (2.8%–4.1%)	Beta	Iida et al. [17]
Probability, TLR, UC	11.6% (9.7%–13.7%)
Probability, amputation, IVUS	1.0% (0.6%–1.4%)
Probability, amputation, UC	0.6% (0.2%–1.0%)
HR, all-cause mortality	1.89 (1.55–2.30)	Lognormal	Diehm et al. [14]
Baseline mortality	Age- and gender-specific	NA	DeStatis [15]
Cost parameters
Cost, PTA	€3,867 (€2,993–€4,741)	Gamma	InEK [21]
Cost, TLR	€5,323 (€4,142–€6,504)
Cost, amputation	€6,424 (€5,187–€7,661)
Cost, IVUS	€1,020 (€916–€1,124)	Assumed fixed	Philips
Utility
Pre-intervention	0.603 (0.525–0.692)	Beta	Petersohn et al. [19]
Post-intervention	0.680 (0.522–0.849)
Amputation	0.540 (0.485–0.595)		Barshes and Belkin [20]

Table I Model parameters

View as image HTML

Semi-Markov model

The probability for all-cause mortality was modelled by applying a hazard ratio (HR) from Diehm et al. [14] onto age- and sex-specific mortality derived from German life tables by DeStatis [15]. The starting age of the modelled population, as well as the proportion of males and females, were derived from Malyar et al, who reported epidemiological data from the RECCORD registry, an observational, prospective, and multi-center registry operated by the German Society of Angiology for patients with PAD in Germany [16].

The probabilities for TLR and amputation were derived from Iida et al., a retrospective, propensity score analysis on the effect on IVUS use during endovascular therapy (EVT) among patients with Trans-Atlantic Inter-Society Consensus Document on Management of Peripheral Arterial Disease (TASC II) class A to C FP lesion [17]. Outcomes of interest in this model included reintervention, defined as any procedure performed when indicated clinically by symptom recurrence, and major amputation. Five-year rates for these two events were converted into annual rates for the model to accommodate for the one-year cycle length, with the assumption that the rates were constant across a five-year period.

Transition probabilities to the “Amputation” state were assumed to be similar for all states. It was also assumed that patients experienced a single amputation; amputations were assumed to be done on the offending limb and curative in nature. Age- and sex-specific mortality rates were also assumed to be similar for the “Alive”, “TLR”, “Post-TLR”, and “Amputation” states. This represented a conservative approach as a higher risk of mortality was not assumed for those in the “Amputation” state despite evidence suggesting otherwise [18].

Utility inputs

Utility values for pre- and post-intervention were obtained from an observational study by Petersohn et al, in which health-related quality of life (HRQoL) was documented in newly diagnosed PAD patients who were followed up over two years [19]. The HRQoL, measured using EQ-5D-3L with Dutch tariffs, at pre-intervention was found to be 0.603 (95% Confidence Interval [CI] 0.525, 0.692). Following a successful intervention, HRQoL was shown to improve to 0.680 (95% CI 0.522–0.849).

The utility of patients who had undergone an amputation was derived from Barshes and Belkin, who compiled HRQoL estimates for major amputation from various studies using the EQ-5D and calculated a weighted mean of 0.54 [20].

Cost inputs

Costs associated with resource utilization in the different health states were included in the model using a German payer perspective. Costs of PTA, TLR, and amputation were derived using relevant German DRG codes from the INeK database [21]. The cost of TLR was assumed to be higher than the cost of PTA to reflect a higher clinical severity warranting revascularization after an index PTA.

The cost incurred during IVUS-guided PTA was assumed to be an additional one-off cost of the IVUS catheter on top of the existing cost of PTA. This was based on the assumption that the only additional device needed for IVUS-guided PTA was the IVUS catheter, and that all other consumables and devices were assumed to be the same as the angiography-guided arm. The cost of the IVUS catheter was obtained from internal sources.

Sensitivity analyses

To reflect inherent parameter uncertainties within the model, one-way sensitivity analyses (OWSA) were performed by varying values for all inputs to the lower and upper bounds of its 95% CI, or at 10% of its deterministic value when the former was unavailable. Tornado diagrams were constructed to display the influence of the individual variables on the overall model results.

Probabilistic sensitivity analysis (PSA), by means of a Monte Carlo simulation with 1,000 iterations using the variability around the parameter inputs, was conducted to explore joint parameter uncertainty. Transition probabilities and utilities were varied using beta distributions, while hazard ratios and costs were varied using lognormal and gamma distributions respectively. Both cost and effectiveness (QALYs gained) were computed in each iteration, and the results of the 1,000 iterations were plotted on the cost-effectiveness (CE) plane. The probability of treatment acceptance, defined as the probability at which the technology is cost-effective compared to the alternative, was calculated, and plotted against various willingness-to-pay (WTP) thresholds to generate the cost-effectiveness acceptability curve (CEAC).

Results

Base-case results

In the base-case scenario with a one-year horizon, IVUS-guided PTA was associated with an incremental QALY of 0.02 at an increased cost of €919 per patient, resulting in an ICER of €45,195 per QALY gained (Table II). Over the lifetime horizon IVUS-guided PTA outperformed angiography-guided PTA with an incremental QALY of 0.22 at a lower cost of €46 per patient.

Table II Base-case results

	One-year horizon		Lifetime horizon
	IVUS-guided	Angiography-guided	IVUS-guided	Angiography-guided
Notes. ICER: incremental cost-effectiveness ratio; IVUS: intravascular ultrasound; LY: life-year; QALY: quality-adjusted life-year.
Outcomes
Total QALYs	0.65	0.63	6.35	6.13
Total LYs	0.97	0.95	9.48	9.21
Total costs, €	5,897	4,977	7,392	7,438
Incremental results
Incremental costs, €	919	–	–46	–
Incremental QALYs	0.02	–	0.22	–
Incremental LYs	0.03	–	0.27	–
ICER (€/QALY)	45,195	–	Dominant	–
ICER (€/LY)	30,633	–	Dominant	–
Net monetary benefit, €	98	–	11,066	–

Table II Base-case results

View as image HTML

Use of adjunctive IVUS during PTA resulted in lower TLR events compared to angiography-guided PTA in both the one-year and lifetime horizon (see electronic supplementary material [ESM]). While there were fewer amputations in the IVUS-guided PTA group in the one-year horizon, there were marginally more amputations in the lifetime horizon.

Sensitivity analyses

One-way sensitivity analyses

One-way sensitivity analyses performed for the one-year horizon showed that IVUS-guided PTA remained cost-effective regardless of changes to individual parameters within their respective plausible range (Figure 2). The utility value for post-TLR, as well as the probabilities of death for both arms had the greatest impact on the ICER. One-way sensitivity analysis performed for the lifetime horizon showed that costs of TLR, as well as the probability of TLR and amputation for both groups influenced the ICER most. Altering the discounting rate for costs and outcomes between 0% and 5% did not yield significant changes to the ICER observed in both time horizons.

Probabilistic sensitivity analyses

The mean incremental costs of all iterations for the one-year horizon were €916, while the mean incremental effects were 0.02, resulting in a mean ICER of €20,228/QALY that is below the €50,000/QALY WTP threshold. The probability of IVUS-guided PTA being cost-effective compared with angiography-guided PTA at a WTP threshold of €50,000/QALY was 50.4% in the one-year horizon (Figure 3).

In contrast, IVUS-guided PTA outperformed angiography-guided PTA with mean incremental costs of all iterations for the lifetime horizon at -€32 and the mean incremental effects at 0.22, resulting in a mean ICER of -€513/QALY. The probability of IVUS-guided PTA being cost-effective compared with angiography-guided PTA at a WTP threshold of €50,000/QALY was 85.9% in the lifetime horizon (Figure 4).

Discussion

This study sought to investigate the cost-effectiveness of IVUS-guided PTA over angiography-guided PTA among PAD patients with FP lesions in Germany. We found that IVUS-guided endovascular intervention was cost-effective over angiography-guided endovascular intervention in a one-year horizon at an arbitrary WTP threshold of €50,000/QALY. Using a lifetime horizon, IVUS-guided endovascular intervention became a dominant strategy compared to conventional angiography-guided interventions due to the higher QALYs accrued at a lower cost. These results remained robust to a number of sensitivity analyses.

Our findings add to the growing body of evidence recommending the use of IVUS in endovascular interventions. This economic evaluation, to the best of our knowledge, is the first of its kind to investigate the cost-effectiveness of IVUS-guided PTA using a model-based approach. The model was conceptualised to capture events that may have significant cost and health impact across the clinical trajectory of a PAD patient, which were validated with clinical experts to ensure relevance to both German payers and healthcare providers for decision-making purposes. The model parameters were obtained from contemporary sources that best reflect the current evidence on IVUS use in PTA and the consequent implications on patient care.

Our findings are largely attributed to the lower rate of TLRs observed in the IVUS group. Revascularization remains a major contributor to overall healthcare costs for PAD patients as it often necessitates a hospitalization episode with considerable length of stay [22]. Chronic total occlusion crossing for example could be challenging, therefore adding IVUS to a best crossing algorithm could be helpful to increase technical and procedural success [30]. By offering higher accuracy during lumen area assessment and detailed evaluation of plaque morphology to determine a best revascularization strategy, use of IVUS may improve patency rates and reduce risk for subsequent TLRs [29, 31]. Fewer TLRs will, in turn, translate to fewer hospitalization episodes and associated healthcare utilization, thus lowering the overall costs of care. Furthermore, ensuring patency and preventing revascularizations can positively impact HRQoL, as the need for revascularization is typically preceded by a loss of patency and the experience of symptoms, both of which have been associated with a loss in quality of life [23].

Recurrent TLRs were not modelled in this study due to a lack of high-quality data, as well as to preserve model parsimony. However, we postulate that the costs incurred for angiography-guided PTA would be higher considering a higher rate of TLRs anticipated in this group. Therefore, the incremental costs (IVUS-guided PTA vs angiography-guided PTA) from this study may be overestimated, further suggesting an advantage for IVUS over angiography during PTA from a cost-effectiveness standpoint.

Despite the growing clinical interest in the use of adjunctive-IVUS during PTA, a lack of studies evaluating its cost-effectiveness may have contributed to its current underutilization. This limits the comparability and understanding against relevant evaluations. Of the limited studies that have investigated cost related to IVUS use, Panaich et al. [24] highlighted that the use of IVUS led to higher in-hospital costs. However, IVUS utilization was also associated with decreased peri-procedural complications and a lower rate of amputation. The nonsignificant increase in hospital costs associated with IVUS should be weighed against a potential reduction in repeat revascularizations and rehospitalizations associated with the decreased complications.

In a similar vein, it is imperative to consider the total costs over a sufficiently long horizon to ascertain the full potential economic value of the technology. We observed that cost savings from averted TLRs over a longer time horizon were sufficient to negate the higher initial costs, where the incremental cost in the lifetime horizon was lower than that in the one-year horizon. We had included a once-off acquisition cost of the IVUS catheter IVUS group similar to cost-effectiveness studies of the use of IVUS in coronary interventions [25]. This was deemed a conservative approach. Consequently, we postulate that the true ICERs may be lower, further justifying the value of incorporating IVUS guidance into current practice.

As of present, there is no official WTP threshold in Germany to guide decision-making. Instead, an efficiency frontier based on findings from economic evaluations may be drawn to ascertain the most efficient interventions among available comparators [11]. Ultrasound (with or without Doppler imaging) and carbon dioxide-angiography have been proposed as alternatives to conventional angiography guidance for PTA, in particular for patients with renal insufficiency, despite a lack data on long-term outcomes [26, 27]. However, there are currently no economic evaluations that have compared these technologies against angiography-guided PTA. As a result, we were unable to compare our findings against such alternative approaches.

There were several limitations to this study. First, there is a general lack of high-quality long-term data surrounding the use of IVUS-guided PTA. Nevertheless, the inclusion of clinical parameters from Allan et al. [13] and Iida et al. [17] offered contemporary insights into the economic and cost-effectiveness implications of a technology that has seen few investigations.

Second, this model investigated IVUS-guided PTA among PAD patients with FP lesions, and thus the results may not be transferable to similar interventions for PAD patients with lesions in a different vessel segment and/or treated with additional endovascular devices (e.g. atherectomy, drug coated devices). This is largely due to a lack of robust data for PAD patients with ex-FP lesions. Additionally, PTA may not be the intervention of choice particularly in patients with infrapopliteal occlusions due to demonstrated suboptimal outcomes [28].

Third, as per all model-based studies, inherent uncertainty exists due to the need for assumptions and data extrapolations. We sought expert opinion establish the clinical relevance of the model. Additionally, we attempted to address this by performing extensive one-way and probabilistic sensitive analyses, where the results were robust and consistent. When the more conservative bound of the point estimates were used in the OWSAs, IVUS-guided PTA was still cost-effective compared against angiography-guided PTA, further strengthening the base-case findings.

Conclusions

In this model, IVUS-guided PTA is a dominant strategy for PAD patients with FP lesions when compared against angiography-guided PTA using a lifetime horizon and is cost-effective in a one-year horizon at a WTP threshold of €50,000 / QALY gained from a German healthcare perspective. IVUS guided intervention was associated with lower TLR, thus improving outcomes, and reducing costs in the long run, which highlights IVUS as an overall cost saving strategy. Considering the current evidence, our analysis supports IVUS guidance during PTA from a cost-effectiveness perspective. Our findings merit further investigation in future prospective multi-centre randomized controlled trials.

Electronic supplementary material