Cost-effectiveness of recombinant tissue-type plasminogen activator for acute ischaemic stroke with unknown time of onset: a Markov modelling analysis from the Chinese and US perspectives

Abstract

Objective

The effectiveness of MRI-guided intravenous recombinant tissue-type plasminogen activator (r-tPA) for acute ischaemic stroke (AIS) with an unknown time of onset has been demonstrated by the WAKE-UP Trial. We aim to evaluate its long-term cost-effectiveness from the perspective of Chinese and US healthcare payers.

Methods

A combination of decision tree and Markov model was built to project lifetime costs and quality-adjusted life-years (QALYs) associated with intravenous r-tPA or placebo treatment. Model inputs including the transition probabilities, costs and utilities were derived from the WAKE-UP Trial, similar cost-effectiveness studies and other published sources. To compare intravenous r-tPA to placebo, we calculated incremental costs, incremental QALYs and incremental cost-effectiveness ratio (ICER). One-way sensitivity, probabilistic sensitivity and subgroup analyses were performed to evaluate uncertainty in the results.

Results

In China, intravenous r-tPA gained an additional lifetime QALY of 0.293 with an additional cost of the Chinese Yuan (¥) of 7871 when compared with placebo, resulting in an ICER of ¥26 870 (US$3894)/QALY. In the USA, intravenous r-tPA yielded a higher QALY (difference: 0.430) and lower cost (difference: ¥−4563) when compared with placebo. In probabilistic sensitivity analyses, intravenous r-tPA had a 97.8% and 99.8% probability of being cost-effective or cost-saving in China and the USA, respectively. These findings remained robust under one-way sensitivity and subgroup analysis except for patients with a National Institute of Health Stroke Scale Score of less than 4, between 11 and 16, and over 16.

Conclusions

MRI-guided intravenous r-tPA for patients with AIS with an unknown time of onset is cost-effective in China and cost-saving in the USA.

Strengths and limitations of this study.

• The cost-effectiveness of MRI-guided intravenous recombinant tissue-type plasminogen activator for acute ischaemic stroke with an unknown time of onset was evaluated from the perspectives of Chinese and US healthcare.

• Subgroup analysis was conducted to validate the results.

• The input parameters were derived from multiple sources and inconsistency across these sources might lead to some bias.

• Patients with different levels of disability were assumed to have the same rate of recurrent stroke, which might not be the case.

• The study was conducted from the healthcare perspective and indirect costs were not considered.

Introduction

Stroke represents a major public health problem around the world, with more than 12 million new cases and 6.5 million deaths every year.¹ In China, an estimated 3.4 million individuals suffered from a new stroke, resulting in 1.5 million deaths in 2019.² The total cost of stroke outpatient and inpatient service has reached 20.71 billion Chinese Yuan (¥) ($3.0 billion) with an average annual growth rate of 24.96%.³ Stroke is a huge burden in the USA as well. Every year, more than 795 000 people have a stroke and about 610 000 of them are first or new strokes.⁴ The total cost of stroke in the USA was $103.5 billion in 2016 and 66% of it was indirect cost from underemployment and premature death.⁵

Intravenous thrombolysis with recombinant tissue-type plasminogen activator (r-tPA) is effective in reducing the disability of acute ischaemic stroke (AIS).^{6 7} Economic studies showed that intravenous r-tPA was cost-effective or even cost-saving in China and the USA.^8–10 However, the efficacy of r-tPA is limited by the time since stroke onset and thrombolytic treatment is only recommended for patients with AIS with an onset time of fewer than 4.5 hours.^{11 12} There are some categories of strokes, such as wake-up stroke and daytime-unwitnessed stroke, that have an unknown time of onset. These strokes accounted for approximately 14%–27% of all strokes.^{13 14} It is uncertain whether these patients would benefit from r-tPA. In a multicentre randomised trial (the WAKE-UP Trial), the researchers sought to determine whether patients with AIS with an unknown time of onset and features suggesting recent cerebral infarction on MRI would benefit from intravenous thrombolysis with r-tPA.¹⁵ The results showed that r-tPA was associated with a significantly better functional outcome and numerically more intracranial haemorrhage (ICH) than placebo at 90 days. Therefore, medical decision analysis is needed to evaluate the advantage or disadvantage of MRI-guided intravenous r-tPA for patients with AIS with an unknown time of onset. Considering that both China and the USA had a large population and heavy burden of stroke, we evaluated the long-term cost-effectiveness of MRI-guided intravenous r-tPA from the perspective of Chinese and US healthcare payers.

Methods

Model overview

Ethical approval from the institutional review board was not required as no human subjects were involved in the study. We conducted this study according to the Consolidated Health Economic Evaluation Reporting Standards reporting guidelines.¹⁶ A combination of decision tree and Markov model was constructed using decision-analytical software (TreeAge Pro, 2020, Williamstown, Massachusetts, USA) to evaluate the cost-effectiveness of MRI-guided intravenous r-tPA versus placebo for AIS with an unknown time of onset. The schematic structure of the model is provided in figure 1.

Figure 1.

Schematic structure of the model. Patients with acute ischaemic stroke with an unknown time of onset entered the model to receive either r-tPA or placebo. In the initial 3 months, patients would transition to one of the seven possible mRS health states. After the initial 3 months, patients would either stay in the same state, experience a recurrent stroke or die from other causes during each cycle. mRS, modified Rankin Scale; r-tPA, recombinant tissue-type plasminogen activator.

A short-run decision-analytical model was created to analyse data on costs and clinical outcomes within the initial 3 months. The target population was analogous to that of the WAKE-UP Trial.¹⁵ Patients were 65 years on average. They were assumed to have AIS with an unknown time of onset and a mismatch between the presence of an abnormal signal on MRI diffusion-weighted imaging and no visible signal change on fluid-attenuated inversion recovery in the region of the acute stroke. These patients could be treated within 4.5 hours after symptom recognition. Patients entered the model to receive either r-tPA or placebo and afterwards transition to one of the seven possible health states according to the degree of disability as assessed by the modified Rankin Scale (mRS) Score. The occurrence of ICH was incorporated into the model with additional costs and disutility.

After the initial 3 months, a long-run Markov state-transition model was used to evaluate the lifetime costs and outcomes. We modelled a lifetime horizon of 30 years and the cycle length was 3 months. During each cycle, patients would either stay in the same state, experience a recurrent stroke or die from other causes. The absorbing state was death (mRS 6) due to stroke or other causes.

Transition probabilities

The input parameters of this model were obtained from the recently published literature (table 1). The mRS health state distribution of the initial 3 months was extracted directly from the WAKE-UP Trial.¹⁵ The annual recurrence rate of stroke after the initial 3 months was 0.118 for the Chinese population⁸ and 0.05 for the US population.¹⁷ A constant recurrent rate was assumed in this study.9 10 18 19 The death rate after recurrent stroke was 0.21 and 0.19 in China⁸ and the USA,²⁰ respectively. According to the previous studies, patients who remained alive after recurrent stroke were assumed to be reallocated equally among health states of equal and greater disability.^{18 19}

Table 1.

Input parameters

Model input	Base-case value	Range	Distribution	Source
Outcomes at initial 3 months with r-tPA
mRS 0	0.213	0–1	Dirichlet	Thomalla et al15
mRS 1	0.323
mRS 2	0.213
mRS 3	0.122
mRS 4	0.071
mRS 5	0.020
mRS 6	0.039
Outcomes at initial 3 months with placebo
mRS 0	0.149	0–1	Dirichlet	Thomalla et al15
mRS 1	0.269
mRS 2	0.229
mRS 3	0.169
mRS 4	0.133
mRS 5	0.040
mRS 6	0.012
Probability of ICH with r-tPA	0.020	0.009–0.046	Beta	Thomalla et al15
Probability of ICH with placebo	0.004	0.001–0.023	Beta	Thomalla et al15
Probability of recurrent stroke in China	0.118	0.112–0.124	Beta	Pan et al8
Probability of death after recurrent stroke in China	0.210	0.189–0.232	Beta	Pan et al8
Probability of recurrent stroke in the USA	0.050	0.040–0.060	Beta	Hong et al17
Probability of death after recurrent stroke in the USA	0.190	0.100–0.300	Beta	Fagan et al20
Death HRs
mRS 0	1	1–1.2	Lognormal	Samsa et al23
mRS 1	1	1–1.2
mRS 2	1.11	0.89–1.3
mRS 3	1.27	1.02–1.52
mRS 4	1.71	1.37–2.05
mRS 5	2.37	1.90–2.84
Cost in the US ($)
MRI	816	± 25%	Gamma	26
Additional cost of r-tPA treatment	8619	4309–12928	Gamma	Leppert et al9
Acute stroke (mRS 0–3)	9268	4633–13901	Gamma	Earnshaw et al27
Acute stroke (mRS 4–5)	14 115	7057–21171	Gamma	Earnshaw et al27
Acute stroke (death)	16 457	8228–24685	Gamma	Earnshaw et al27
ICH	3399	2719–4079	Gamma	Earnshaw et al27
Annual posthospitalisation (mRS 0–3)	8157	4078–12 235	Gamma	Earnshaw et al27
Annual posthospitalisation (mRS 4–5)	22 139	11 070–33209	Gamma	Earnshaw et al27
Recurrent stroke	25 143	12572–37 715	Gamma	Leppert et al9
Cost in China (¥)
MRI	600	±25%	Gamma	Chen et al25
Additional cost of r-tPA treatment	13 886	10 751–16 194	Gamma	Pan et al8
Acute stroke (mRS 0–1)	12 214	7055–15 379	Gamma	Pan et al8
Acute stroke (mRS 2–5)	16 149	8875–21 177	Gamma	Pan et al8
Acute stroke (death)	13 840	6503–18 293	Gamma	Pan et al8
ICH	2949	641–6155	Gamma	Pan et al8
Annual posthospitalisation (mRS 0–1)	8684	2600–11 077	Gamma	Pan et al8
Annual posthospitalisation (mRS 2–5)	13 213	3323–16616	Gamma	Pan et al8
Recurrent stroke	18 000	± 25%	Gamma	Institutional database
Utility
mRS 0	0.80	0.64–1	Beta	Samsa et al23
mRS 1	0.80	0.64–1
mRS 2	0.65	0.52–0.78
mRS 3	0.50	0.4–0.6
mRS 4	0.35	0.28–0.42
mRS 5	0.20	0.16–0.24
Death	0.00	NA	NA	NA
Disutility of ICH	0.38	0.30–0.46	Normal	Christensen et al30
Discount rate	0.03	0–0.08	Beta	Sanders et al31

NA: not available; mRS: modified Rankin Scale; ICH, intracranial haemorrhage; r-tPA, recombinant tissue-type plasminogen activator.

The age-specific death rate was drawn from the most recently published census of China²¹ and the US Life Table in the USA.²² Disabled patients tend to have increased mortality compared with non-disabled ones and we adjusted the death rate according to the HRs for each mRS health state.²³ We converted the annual transition probabilities to 3-month probabilities according to the standard formula if necessary.²⁴

Costs

This study was conducted from the perspective of healthcare payers in China and the USA and only direct costs were considered. In China, the additional costs of intravenous thrombolysis, acute stroke treatment cost for different mRS health states, cost of ICH treatment and posthospitalisation cost for different mRS health states were extracted directly from another cost-effectiveness study in which the authors derived these costs from the database of Thrombolysis Implementation and Monitor of Acute Ischaemic Stroke in China and the China National Stroke Registry.⁸ The cost of MRI in China was obtained from previous literature.²⁵ The cost of recurrent stroke was assumed to be the same in the two intervention arms because it is unlikely to predict the type and severity of a recurrent stroke. The cost of recurrent stroke in China was not reported in previous literature and we estimated it from our institutional database as the mean expected cost to treat an average stroke without thrombolysis or thrombectomy. In the USA, the cost of MRI was from the Centers for Medicare & Medicaid Services (CPT code 70557).²⁶ Acute stroke treatment cost for different mRS health states, cost of ICH treatment and posthospitalisation cost for different mRS health states were derived from a previous study in which these costs were calculated based on multiple resources.²⁷ These costs were validated and used by some other similar cost-effectiveness studies.^{9 10 18} The additional cost of intravenous thrombolysis and the cost of recurrent stroke were also reported previously.⁹ To account for the uncertainty, a range of ±25% of the base-case value was used for the costs. All costs were converted to the 2020 price according to the medical care component of the consumer price index if necessary.^{28 29}

Utility

Health-related quality of life values (utility scores) was assigned to all health states. Quality-adjusted life-years (QALYs) was calculated, which was defined as the length of period the patient spent in a particular state multiplied by the corresponding utility score. Currently, the Chinese population-specific utility scores for mRS scores of 0, 1, 2, 3, 4 and 5 were not reported in the previous literature. We used the same utility scores in China and the USA. Utility scores for different mRS health states were obtained from a previous validation study²³ and other similar cost-effectiveness studies also used the same utility scores.^{9 18} Patients with ICH were assumed to have a disutility of 0.38.³⁰ All costs and utilities were discounted by 3% per year.³¹

Statistical analysis

The primary measure was the incremental cost-effectiveness ratio (ICER) defined as the incremental cost per additional QALYs gained. One treatment was deemed cost-effective when compared with another if the ICER was below the willingness-to-pay (WTP) threshold. In China, according to the recommendation of the WHO, the WTP threshold was chosen as 3 × gross domestic product per capita.³² This WTP threshold corresponded to ¥217 341 ($31 499)/QALY in 2020.³³ We used a WTP threshold of $100 000/QALY for the USA.³⁴

The base-case analysis was performed with the mean value of all input parameters. One-way sensitivity analyses were performed to identify key parameters related to the robustness of the results by varying one parameter while keeping others fixed. Probabilistic sensitivity analysis was performed with all parameters varying simultaneously in their assigned distributions. A total of 10 000 iterations was carried out to evaluate the impact of uncertainty.

Moreover, subgroup analysis was performed in the prespecified subgroups as defined in the WAKE-UP Trial. In the subgroup analysis, the distribution of mRS health states at the initial 3 months for different subgroups was calculated according to the reported proportion of favourable outcomes (defined as mRS Scores of 0 and 1). During the calculation, the ratio of proportion for different mRS Scores (mRS 0/mRS 1 and mRS 2/mRS 3/mRS 4/mRS 5/mRS 6) in the subgroups was assumed to be as same as that in the overall population. The mRS health states distributions for different subgroups were provided in online supplemental table 1.

Patient and public involvement

None.

Results

Base-case analysis

The cost, QALY and ICER of r-tPA versus placebo in patients with AIS with an unknown time of onset are summarised in table 2. In China, r-tPA was associated with an additional cost of ¥7871 with an additional QALY of 0.293 over 30 years when compared with placebo, and the ICER was ¥26 870 ($3894)/QALY. In the USA, r-tPA was associated with a higher QALY (0.430) and a lower cost (¥−4563) over 30 years when compared with placebo.

Table 2.

Base-case analysis

	China			US
	r-tPA	Placebo	Difference	r-tPA	Placebo	Difference
Costs	¥152 686	¥144 815	¥7871	$159 326	$163 889	$−4563
QALY	5.305	5.012	0.293	7.297	6.867	0.430
ICER			26 870			Negative

ICER, incremental cost-effectiveness ratio; QALY, quality-adjusted life-years; r-tPA, recombinant tissue-type plasminogen activator.

Sensitivity analysis

One-way sensitivity analyses were performed to evaluate the impact of the uncertainty of different parameters on the final ICER and the results were presented in the tornado diagram (figure 2). Overall, the results were more sensitive to the posthospitalisation cost for different mRS health states, additional cost of r-tPA treatment and discount rate. However, when these parameters varied in their ranges, the ICERs remained under the corresponding WTP thresholds both in China and the USA, indicating that the results were robust.

Figure 2.

Tornado diagram of one-way sensitivity analyses in China. The plot indicates how varying one parameter to its limits at a time affects the incremental cost-effectiveness ratio. ICH, intracranial haemorrhage; mRS, modified Rankin Scale; QALY, quality-adjusted life-years; r-tPA, recombinant tissue-type plasminogen activator.

The results of probabilistic sensitivity analyses are presented in figure 3. Among 10 000 iterations, r-tPA had a 97.8% probability of being cost-effective at the WTP threshold of ¥217 341/QALY in China and a 99.8% probability of being cost-effective at the WTP threshold of $100 000/QALY in the USA.

Figure 3.

Tornado diagram of one-way sensitivity analyses in the USA. mRS, modified Rankin Scale; QALY, quality-adjusted life-years; r-tPA, recombinant tissue-type plasminogen activator.

Subgroup analysis

Subgroup analysis was conducted according to the different distributions of mRS health states at the initial 3 months among different subgroups (online supplemental table 1). In China, all the ICERs were below the WTP threshold of ¥217 341/QALY except for the populations with a National Institute of Health Stroke Scale (NIHSS) Score of less than 4, between 11 and 16, and over 16 (figure 4). In the USA, r-tPA is dominant when compared with r-tPA except for the populations with an NIHSS Score of between 11 and 16 and over 16. For the subgroup with an NIHSS Score of less than 4, r-tPA had an ICER of $57 826/QALY when compared with placebo, and it was below the WTP threshold of $100 000/QALY in the USA (figure 5).

Figure 4.

Results of the probabilistic sensitivity analysis in China. The dots that lie to the right of the willingness-to-pay (WTP, ¥217 341) line mean the cases where intravenous r-tPA is cost-effective when compared with placebo. r-tPA, recombinant tissue-type plasminogen activator.

Figure 5.

Results of the probabilistic sensitivity analysis in the USA. The dots that lie to the right of the willingness-to-pay (WTP, $100000) line mean the cases where intravenous r-tPA is cost-effective when compared with placebo. r-tPA, recombinant tissue-type plasminogen activator.

Discussion

For patients with AIS with an unknown time of onset, MRI-guided intravenous r-tPA increased life expectancy by 0.293 and 0.430 over a lifetime in China and the USA, near 3.6 months and 5.2 months of perfect health at excellent value. Intravenous r-tPA gained an additional cost of ¥7871 with an ICER of ¥26 870/QALY and it was cost-effective under the current Chinese WTP threshold. In the USA, intravenous r-tPA was associated with a lower cost when compared with placebo and was deemed cost-saving in the long run. The robustness of our overall conclusion that intravenous r-tPA was cost-effective or even cost-saving was supported by the sensitivity analyses. In the one-way sensitivity analyses, all the ICERs were below the WTP threshold when the parameters varied in their ranges. In the probabilistic sensitivity analyses, intravenous r-tPA had a 97.8% and 99.8% probability of being cost-effective or cost-saving in China and the USA, respectively. Moreover, in the subgroup analysis, intravenous r-tPA was cost-effective or cost-saving in all subgroups except for patients with an NIHSS Score of less than 4, between 11 and 16, and over 16, which was due to the similar treatment effects for r-tPA and placebo or a small number of participants in these subgroups.

The cost-effectiveness of intravenous r-tPA for treating AIS has been extensively studied before. However, all these studies are limited to the time window of 0–3 hours, 3–4.5 hours or 0–6 hours.^{35 36} In all, the literature generally indicated that intravenous r-tPA was a cost-effective or dominant strategy compared with the traditional treatment for patients with AIS. Our results and conclusions were comparable to the studies from the Chinese and US perspective. For example, Pan et al investigated the cost-effectiveness of thrombolysis with r-tPA within 4.5 hours of AIS in China. In this study, r-tPA yielded an ICER of ¥15 500/QALY in 30 years when compared with placebo, and its cost and effectiveness were ¥120 880 and 4.993 QALYs, respectively.⁸ In the USA, Boudreau et al studied the cost-effectiveness of intravenous r-tPA within 3 hours of AIS and found that r-tPA resulted in an additional gain of 0.39 QALY and a lifetime cost saving of $25 000. In probabilistic sensitivity analysis, intravenous r-tPA was dominant compared with no r-tPA in nearly 100% of probability.¹⁰

We noticed a study that modelled the cost-effectiveness of MRI-guided treatment of acute wake-up stroke.³⁷ The authors created a microsimulation model to simulate the wake-up stroke patient population with an average age of 65 years and 60% of them being male. According to the results, the MRI-based treatment strategy for this population was cost-effective. However, the conclusion was sensitive to several factors such as sleep duration, hospital travel and door-to-needle times. The stroke onset time was decided by the MRI and the mRS health state distributions, which we believe were the major limitations of this study. Our study has the advantage of using the latest effectiveness data from the multicentre randomised WAKE-UP Trial.¹⁵ This study focused on outcomes based on the mismatch between diffusion-weighted imaging and fluid-attenuated inversion recovery findings on MRI rather than on a time-dependent effect of intravenous r-tPA. Therefore, the sensitivity and specificity of MRI to determine stroke onset time were not incorporated into our model as they were unnecessary. Our study also has the advantage of analysing from the perspective of China and the USA, both of which are inflicted by the heavy burden of stroke. Even though MRI-guided intravenous r-tPA for stroke with an unknown time of onset has not been suggested in China,¹² the 2019 Guideline from the American Heart Association/American Stroke Association for the early management of AIS suggests that in patients with AIS with unknown time of onset, MRI can be useful to select those who can benefit from intravenous r-tPA within 4.5 hours of stroke symptom recognition (Class of recommendation IIa, level of evidence B).¹¹

The limitations of our study should be noted. First, our model was based on required assumptions such as the average age and time window of treatment. Therefore, the results and conclusions are specific to those assumptions. Generalising the conclusions to other situations should be done with caution. Second, the data on input parameters were from numerous published studies including clinical trials. We were not able to estimate the costs and effectiveness after AIS with one data source. The inconsistency among these published studies might cause some bias towards our study. Third, our results were based on the efficacy findings of the WAKE-UP Trial that was conducted in Europe. It is unknown whether similar treatment effects would occur if the participants were restricted to Chinese or US patients. What’s more, the utility scores were not Chinese-population specific. However, the sensitivity analyses have accounted for these uncertainties and this is not unprecedented in other cost-effectiveness studies.

Conclusions

In summary, MRI-guided intravenous r-tPA for patients with AIS with an unknown time of onset is cost-effective in China and cost-saving in the USA. However, further studies are needed to evaluate the benefit of intravenous r-tPA for treating specific subgroups of patients.

Data availability statement

Data are available upon reasonable request.

Ethics statements

Patient consent for publication

Not applicable.

Ethics approval

Not applicable.