Cost-effectiveness analysis of tissue plasminogen activator in acute ischemic stroke in Iran

Abstract

Aims: Tissue plasminogen activator (tPA) is used to treat acute ischemic stroke up to 4.5 h after symptom onset. Its cost-effectiveness in developing countries is not specified yet. This study aimed to study cost-effectiveness of tPA in Iran.

Methods: This is a cost-effectiveness analysis from the perspective of the third party payer to compare IV tPA with no tPA of ischemic stroke. A Markov model with a lifetime horizon was used to analyze the costs and outcomes. Cost data were extracted from the 94 patients admitted in two hospitals in Iran. All costs were calculated based on US dollars in 2016. Quality-adjusted life years (QALY) were extracted from previously published literature. Cost-effectiveness was determined by calculating ICER by TreeAge Pro 2011 software.

Results: Lifetime costs of no tPA strategy were higher than tPA ($10,718 in the no tPA group compared with $8,796 in the tPA group). The tPA arm gained 0.20 QALY compared with no tPA. ICER was $8,471 per QALY. ICER value suggests that tPA is cost-effective compared with no tPA.

Limitations: The limitations of the present study are the reliance on calculated QALY value of other countries and difficulty in accessing patients treated with tPA.

Conclusions: The balance of hospitalization and rehabilitation costs and QALYs support the conclusion that treatment with intravenous tPA in the 4.5-h time window is cost-effective from the perspectives of the third party payer and inclusion of tPA in the insurance benefit package being reasonable.

Keywords:

• Cost
• quality adjusted life year
• tissue plasminogen activator
• ischemic stroke
• cost-effectiveness

View correction statement:

Erratum

Introduction

According to the World Health Organization’s definition, stroke happens by an interruption of blood flow to the brain, which is usually due to the rupture (hemorrhagic) or blockage of a blood vessel by a clot (ischemic). Each of these interrupt the flow of oxygen and nutrients to the brain, which causes damage to brain tissue¹. About 87% of strokes are ischemic². The mortality rate of stroke in the first 30 days of onset is 10%, and the other 90% often result in disabilities. In general, 70% of people who have a stroke will not be able to return to work, and 30% of them will need an aid for walking³. This creates a psychological, social, and economic pressure on society.

Stroke was the second cause of death in the world in 2012 (6.7 million deaths). A third of these deaths occurred in the developing countries³. According to an estimated global burden of stroke by Feigin et al.⁴, the incidence of stroke in Iran is 252–336 per 100,000 persons. In 1996, the NINDS rtPA Stroke Trial, in which 624 patients with ischemic stroke were treated with placebo or intravenous rtPA within 3 h of symptom onset, was conducted⁵, which led to the tissue plasminogen activator (tPA) being approved as the sole drug to treat acute ischemic stroke within 3 h of symptoms onset by the US Food and Drug Administration⁶. In 2008, the ECASS III Group performed a randomized trial and demonstrated the efficacy of tPA administered in patients with acute ischemic stroke which can be safely extended to a time window of 3–4.5 h after the onset of stroke symptoms⁷.

However, further trials and subsequent studies increased the thrombolytic therapy time to 6 h for some eligible patient⁸. This drug acts by dissolving the blood clot that has caused the stroke. It improves the functional outcomes and decreasing disability within 6 months. One of the important side-effects of tPA is intracranial haemorrhage, which could be fatal in the first 7–10 days after the injection⁹.

Jung et al.¹⁰ conducted a systematic review on the overall cost-effectiveness of tPA in comparison with conventional treatment in the absence of the tPA. Eight similar studies in six countries, including the US, UK, Canada, Australia, Spain, and Denmark were conducted mostly from the perspectives of the healthcare system or society, and, in all of them, tPA treatment was the most cost-effective treatment. Other studies in Brazil, the US, Australia, and China were conducted from the perspective of the community or the healthcare system, and in all of them similar results were obtained, and all accepted tPA as a the treatment of choice^11–16.

It has been over 20 years since this drug was introduced, but in developing countries, including Iran, there are still many limitations for the use of tPA, including doubt on the safety of the drug, and a lack of the required equipment in case of complications after the injection. Also, in Iran, the high cost of tPA is a reason for third party payers not to cover this treatment. This problem motivated the authors to conduct a study to examine the cost-effectiveness of tissue plasminogen activator (tPA) in the treatment of acute ischemic stroke in Iran.

Method

This is a tPA cost-effectiveness analysis (CEA) study which was conducted based on pharmaco-economics guidelines of Fox-Rushby and Cairns¹⁷ and WHO¹⁸. The present study was carried out from the perspective of a third party payer compared to a no tPA strategy of ischemic stroke in lifetime horizon. The no tPA method includes maintenance therapy and continuous monitoring, and is performed according to standard guidelines developed in Iran which are prepared based on international guidelines for ischemic stroke patients who were eligible for tPA, but, due to the delay in referral to the hospital, did not receive tPA^6,19,20. This study retrospectively recruited 47 patients with ischemic stroke from an Iranian hospital who received the tPA treatment (intervention group) and, because the dominant treatment in the first hospital was tPA, we had to choose 47 patients with ischemic stroke from another Hospital in the same city for the no tPA group (comparator group) during 2015–2016. Additionally, it is worth mentioning that medical education in Iran is centralized, such that the Ministry of Health and Medical Education develops unified curriculums to be performed in all universities of medical education. This minimizes variation in medical practice in different hospitals. We measured the costs associated with these two treatment strategies in studied patients, and the outcome data was taken from Fagan et al.²¹ and Davis et al.²².

Decision analysis model

The Three-State Markov model with Treeage pro 2011 software was used to measure the cost-effectiveness of tPA. The Markov model includes gradual analysis of the state for a specific disease. The model in this study had three states which were determining the possible state of people after ischemic stroke according to the severity of disability or modified Rankin Scale (mRS)²³. This scale shows the degree of disability or dependency in the daily activities after a stroke, which includes a score of 0–6 indicating state, from no signs of disability to death. The first group of patients with an mRS of 0–2 (good outcome) consisted of those patients who were in good health or were independent in performing daily activities. The second group included patients with an mRS of 3–5, who had to rely on others for daily activities. The third group included patients with an mRS of 6, who did not survive (death).

According to expert opinions and related previous studies^22,24, these states can change in such a way that patients with mRS 0–2 can maintain the same state or change into another two states, also, this is the case for those patients with an mRS of 3–5. These transitions were equal in each model arm. Patients in the independent and dependent states at 12 months and beyond who were unable to move to dependent and independent states, stayed in the same health state or died. We also incorporated the deaths from other causes in the model, using WHO’s life table for Iran²⁵. Previous studies were reviewed for transition probabilities. We could use studies with similar states, so we chose Davis et al.’s²² transitions, which took its transitions probabilities from LSR^26,27 (Figure 1).

Figure 1. (A) Long-run model used to simulate lifetime patient outcomes. (B) Short-run decision analytic model used to model cost-effectiveness analysis of tissue-type plasminogen activator.

Since the tPA treatment effect completes after 3 months and plateaus after 6 months²⁸, we used that time for the length of the Markov model cycle.

Initial distribution of stroke patients, to be used for entering into the Markov model, was based on our samples and Ganesalingam et al.²⁴ for IV tPA. In order to ensure the validity of our initial distribution into the three mRS states (independent, dependent, and death), we consulted experts who had extensive experience working with stroke patients to check the initial distribution. However, the initial distribution did not change significantly after the expert panel (Table 1).

Table 1. Markov model parameters and range of values for sensitivity analysis, per one patient.

	Base case	Sensitivity range	Distribution	Reference
Transition probabilities
Movement from independent to:
Independent	0.875	0.77–0.89		[22,26,27]
Dependent	0.0938	0.07–0.11		[22,26,27]
Movement from dependent to:
Independent	0.1111	0.03–0.18		[22,26,27]
Dependent	0.7407	0.65–0.81		[22,26,27]
Initial distribution
Independent	0.26	0.07–0.45		[24]
Dependent	0.55	0.36–0.74		[24]
Utility
Independent	0.74			[22,29]
Dependent	0.38			[22,29]
QALY*
Independent tPA	0.30	0.1–0.50	Uniform	[21,22]
Dependent tPA	0.12	0–0.32	Uniform	[21,22]
Independent no tPA	0.24	0.04–0.44	Uniform	[21,22]
Dependent no tPA	0.14	0–0.34	Uniform	[21,22]
Costs ($)
Independent tPA	636.10	224.9–1078	Lognormal
Dependent tPA	1,857.90	0–9,565	Lognormal
Death tPA	2,506.30	391.30–5,820	Lognormal
Independent no tPA	1,108.20	267–35,26.14	Lognormal
Dependent no tPA	1,846.50	239–11,038.90	Lognormal
Death no tPA	1,759.40	232.60–6,913.20	Lognormal

* Quality adjusted life year.

To categorize patients into the mentioned three states, proportional sampling was used. To determine the mRS of patients who had received tPA at the time of discharge, surviving patients were contacted and their mRS was determined in an interview. To maintain homogeneity in the two groups and to reduce the effects of confounding factors such as gender, age, costs, and mRS value on the treatment process, the no tPA group was matched with the tPA group.

Cost and outcome

The total costs were entered into the model, including the direct patient charges and the insurance company share registered in their medical records, and rehabilitation costs. Rehabilitation costs were estimated by interviewing patients and their families and in some cases through their medical records. The sum of the insurer share of hospitalization and rehabilitation costs per patient were entered into the model.

To measure the outcomes, since no Iranian study has been conducted to calculate QALY for ischemic stroke, we had to use the life years gain (LYG) from Fagan et al.²¹ and utility from the study of Davis et al.²², which took its utility from Dorman et al.²⁹, to calculate the QALY of patients, because both of them had the same setting as our study.

Fagan et al.²¹ studied LYG separately for each mRS for 1,000 patients in a 30-year time horizon. In the present study, we first calculated life years gained for a patient in the lifetime horizon, then we used average LYG mRS 0, 1, 2 and average LYG mRS 3, 4, 5 to calculate the LYG of each state in the two arms (Table 1).

One of the advantages of Markov models is that costs and outcomes, which may occur at different time periods, can be easily discounted. We also calculated the long-term cost, which consists of hospitalization and rehabilitation costs, with a discount rate of 0.06 and 0.03 for the long-term QALY based on local discount rates³⁰. All costs were calculated based on the value of US dollars in 2016³¹.

The ratio of incremental cost-effectiveness or ICER was calculated by Treeage pro 2011 software and a decision was made as to whether to reject or accept the new technology.

Sensitivity analysis

Considering uncertain parameters of the model including QALY, initial distribution, and the transition probabilities, basic and probabilistic sensitivity analyses were carried out using a variation range of 20%, and calculating the costs using the lowest and highest amounts. According to CEA guidelines, the range of sensitivity analysis could be taken from previous studies and or researchers own views; accordingly we selected the 20% range.^14,17 (Table 1).

Results

Costs

Direct costs of patients with ischemic stroke during 2015 and a quarter of 2016 in the first Hospital were reviewed. The number of stroke patients in this Hospital during that period was 837 patients; the direct total costs and unit costs were $1,038,968.7 and $1,241.3, respectively. Direct total cost in this study meant the total bill of the patient who is hospitalized in the hospital. Of this cost, a certain part (∼90%) is paid by a third party payer and the rest is paid by the patient as copayment.

The mean age in the group treated with tPA was 69.7 years, and it was 70.7 years in the no tPA group, and 36.17% of patients in each group were women. The average direct cost spent per patient who received tPA was $3,039.317, and the cost in the no tPA group was $2,055.297, and additional costs of tPA (purchasing tPA) were $18,938. Other costs for all patients in two group stroke patients in this study are presented in Table 2. To evaluate the cost-effectiveness of tPA, the average direct costs paid by the payers were calculated for the two groups of patients on discharge using the Markov Model. Furthermore, total costs of rehabilitation and the share of third party payers from the costs of rehabilitation for each of the three mRS states were also calculated in each group after the discharge (see Table 2).

Table 2. Demographic characteristics.

	tPA	No tPA
Number of patients, n	47	47
Women, n	17	17
Age (mean)	69.7	70.7
Cost of tPA ($)	18,938.5	–
Third party payer costs ($)
Inpatient costs	84,605.64	80,835.20
Rehabilitation costs	2,768.68	6,095.61
Drugs	11,635	7,909.90
Inpatient visits	8,523.80	9,605.90
Surgery	740.60	581.20
Diagnostic	13,514.30	5,723.30
Laboratory	4,836.60	1,813.60

Base case analysis

The results of the cost-effectiveness of tPA in the first 4.5 h after the onset of symptoms compared to no tPA from the perspective of the third party payers per 1,000 patients in two different time horizons are shown in Table 3, which indicates in lifetime horizon 0.20 QALY gained and $1,921.48 cost decrease in the tPA group per 1,000 patients treated. The incremental cost-effectiveness ratio (ICER) was $9,201.14 per QALY gained. On average, the cost in each QALY unit, i.e. average cost-effectiveness ratio (ACER) of treatment with tPA, was $5,335.68 and treatment with no tPA was 7,444.03 per QALY. This is located in the southeast quadrant of the Cost-Effectiveness plane and the ICER value suggests that tPA is cost-effective compared to no tPA.

Table 3. Base-case analysis per 1,000 patients: Costs, QALYs, and ICER of tPA within 4.5 h of symptom onset compared with no tPA in patients with acute ischemic stroke.

Time horizon	Strategy	QALY	Cost ($)	ICER	ACER
1 year	tPA*	0.42	3,730.53	3,087.60	8,843.69
	No tPA	0.39	3,817.54		9,697.77
Life time	tPA	1.64	8,796.96	9,201.14	5,335.68
	No tPA	1.43	10,718.45		7,444.03

* Tissue type plasminogen activator.

Sensitivity analysis

For all uncertain parameters of the model including QALY, the cost, transition probability, and the initial distribution based on variation range presented in Table 1, the basic sensitivity analysis was performed. According to the tornado diagram, four first parameters had the greatest impact on costs; these parameters are shown in Figure 2. One-way and two-way sensitivity analysis was done on these variables. According to the results of the analysis, the result was not sensitive to changes of these variables. To perform probabilistic sensitivity analysis (PSA) with 10,000 stimulations, the distribution of variables was determined using the EasyFit 5.5 Professional software. The country’s threshold is 3-times the country’s GDP per capita ($5,757.80) equal to $17,273.40. Results of the cost-effectiveness acceptability curves show that tPA in various thresholds is always a cost effective strategy (Figure 3).

Figure 2. Tornado diagram for uncertain parameters. Initial QALYs of tPA, transition probability and initial distribution had the greatest impact. Abbreviations. QALY, Quality Adjusted Life-Year; tPA, tissue type Plasminogen Activator; mRS, modified Rankin Scale; Pr, probability; Incr, Incremental.

Figure 3. Cost-effectiveness acceptability curve, favors treatment with intravenous tPA in the 4.5-h time window over no tPA from the perspective of a third party payer in Iran.

Discussion

According to the results of the study, the use of tPA in the first 4.5 h after the symptoms onset and stroke attack is cost-effective from a third party payer perspective. According to the sensitivity analysis and the effective factors on ICER, the health policy-makers should plan and focus more on stroke patients quality-of-life, in order to achieve better results from tPA treatment. This study was conducted at the time when the tPA treatment was not covered by third party payers. Fortunately, in Iran, tPA has being covered by third party payers since June 2016. The effects of this treatment on the rehabilitation costs, which usually starts after the end of the first treatment cycle, were evident, and these costs were significantly less in the intervention group compared to the no tPA group. Reducing the cost of rehabilitation in the tPA group had a significant impact on the total costs in long-term and ICER value. Therefore, tPA can be introduced as a cost-effective strategy against the no tPA strategy. In all related evaluations carried out, similar results have been obtained and tPA has been accepted as a cost-effective treatment method compared to no tPA. It is worth mentioning that, after the health transformation plan implemented in 2015 in Iran, there have been some initiatives to reduce the burden of stroke in Iran, including developing Stroke units in some selected hospitals. We hope that this will help to improve the effectiveness and make tPA treatment worthwhile. Yet, due to problems such as the high cost of tPA, the need for required equipment and structural facilities in health centers, and the delay in taking stroke patients to health centers and missing golden time, the use of tPA is very limited in our country. Providing or improving facilities such as stroke units, trained staff, or stroke systems may appropriately increase the tPA treatment rate and improve safe delivery of tPA.

Summary of previous work

Pan et al.¹⁵ conducted a study in China to evaluate the cost-effectiveness of tPA up to 4.5 h after its administration from the perspective of the health system. A decision tree and Markov model were used to measure the cost-effectiveness. The costs’ information was derived from the study of TIMS-China, and information about the outcomes was derived from the analysis of four clinical trials. The mean age of the patients was 63 years. In a 1-year time period, based on a 105,000 yen threshold, the ICER value was 181.607 and tPA was not cost-effective, but, in the 2nd and 30th years, the values of ICER were 94,257 and 15,474, respectively, and the tPA was cost-effective. Boudreau et al.^13,14 conducted two studies in the US to evaluate the cost-effectiveness of tPA from the perspective of the health system within 0–3 h and 3–4.5 h after the onset of stroke symptoms using the Markov model. The mean age of patients in their studies was 66 and 70 years. In the first study, the results indicated the QALY of 0.39 and a $25,000 decrease in cost when tPA was used, so the use of tPA was cost-effective¹⁴. In the second study, the QALY was 0.49 and a $1,495 increase was seen in the cost when the tPA was used, so the tPA with ICER of 6,255 was cost-effective compared to the $100,000 threshold¹³. Tanny et al.¹² conducted a similar study in Australia from the perspective of society. In their study, the costs’ information was derived from patients who were hospitalized between 2003 and 2011 in a Hospital in Melbourne, and the outcomes’ information was collected and derived from meta-analysis clinical trials that were previously conducted on the subject. The mean patient age was 68 years and a 1-year time period was considered. Their results showed a 0.04 increase in QALY and a $1,495 decrease in the cost when the tPA was used, so the tPA with an ICER of 1,478 was cost-effective.

A summary of the ACER and ICER of previous studies is shown in Table 4. In all of this study, tPA is the dominant strategy.

Table 4. Summarize the ICER and ACER for various countries (cost per QALY).

Study	Country	Unit cost	Unit effect	ACER		ICER
				tPA	No tPA
Sinclair et al. 2001	Canada	CAN $	QALY	785	1,105	109
Stahl et al. 2003	US	US $	QALY	18,984	19,156	43,400
Sandercock 2002/4	UK	£	QALY	11,545	13,301	96,565
Moodie et al. 2004	Australia	AUS$	DALYs	4,109	4,107	3,841
Mar et al. 2004	Spain	€	QALY	2,437	2,276	4,528
Ehlers et al. 2006	Denmark	US $	QALY	31,896	42,551	33,523
Araujo et al.	Brazil	R $	QALY	–	–	36,836
Tung et al. 2011	US	US $	QALY	16,008	15,724	21,978
Tanny et al. 2013	Australia	US $	QALY	62,104	65,619	2,002
Boudreau et al. 2013	US	US $	QALY	16,720	17,111	6,255
Boudreau et al. 2014	US	US $	QALY	66,993	80,102	64,102
Pan et al. 2014	China	CNY	QALY	24,209	25,016	15,473
Present study	Iran	US $	QALY	5,335	7,444	9,201

Limitations

Our study has some limitations. First, we used the calculated QALY value of other countries because this value has not been calculated in Iran yet, so we used sensitivity analysis for determining the robustness of the results. Second, difficulties in access to patients treated with tPA or small number of patients who met the administration criteria, might lower the generalizability of the results. Third, when we started this study and selected the sample, tPA was not included in insurance benefit packages in Iran, which is why there were only a small number of patients treated with tPA.

Conclusion

In conclusion, the balance of hospitalization and rehabilitation costs and QALYs favors treatment with intravenous tPA in the 4.5-h time window over no tPA from the perspective of a third party payer in Iran. This strategy saves the limited health system resources in the long run. This strategy may be costly for the third party payer and health system at the time of hospitalization and in the short-term, but the treatment of patients who are eligible for thrombolytic therapy in the long-term reduces the financial burden of the stroke by lowering rehabilitation costs. We might benefit more if we start the tPA treatment at a shorter time from the onset of symptoms. Additional research is needed to study implications of tPA use in all hospitals in the country and to improve the administration of tPA when patients are eligible and protocols are in place.

Transparency

Declaration of funding

There is no funding to report for this study.

Declaration of financial/other relationships

The authors have no financial or other relationships to disclose. JME peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Acknowledgments

The authors thank Dr Moradi, Mr Daroudi, Miss Lalianpoor, Miss Soltani, and Miss Riahi, who helped collect data.