Adapting a global cost-effectiveness model to local country requirements: posaconazole case study

Abstract

Background:

Many countries have various requirements for local economic analyses to assess the value of a new health technology and/or to secure reimbursement. This study presents a case study of an economic model developed to assess the cost-effectiveness of posaconazole vs standard azole therapy (fluconazole/itraconazole) to prevent invasive fungal infections (IFIs), which was adapted by at least 11 countries.

Methods:

Modeling techniques were used to assess the cost-effectiveness of posaconazole vs fluconazole/itraconazole as IFI prophylaxis in patients with acute myelogenous leukemia or myelodysplastic syndromes and chemotherapy-induced neutropenia. For the core model, the probabilities of experiencing an IFI, IFI-related death, and death from other causes were estimated from clinical trial data. Long-term mortality, drug costs, and IFI treatment costs were obtained from secondary sources. Locally changed parameters were probabilities of long-term death and survival, currency, drug costs, health utility, IFI treatment costs, and discount rate.

Results:

Locally adapted cost-effective modeling studies indicate that prophylaxis with posaconazole, compared with fluconazole/itraconazole, prolongs survival, and, in most countries, is cost-saving. In all countries, the model predicted that prophylaxis with posaconazole would be associated with an increase in life-years, with increases ranging from 0.016–0.1 life-year saved. In all countries, use of the model led to posaconazole being approved by the appropriate reimbursement authority.

Limitations:

The study did not have power to detect differences between posaconazole and fluconazole or itraconazole separately. The risk of death after 100 days was assumed to be equal for those who did and did not develop an IFI, and equal probabilities of IFI-related and other death during the trial period were used for both groups.

Conclusions:

A core economic model was successfully adapted locally by several countries. The model showed that posaconazole was cost-saving or cost-effective vs fluconazole/itraconazole and led to positive reimbursement listings.

Key words::

• Pharmacoeconomic model
• Posaconazole
• Cost-effectiveness
• Invasive fungal infection
• Neutropenia
• International adaptations

Introduction

Rising global healthcare costs mean that evidence of the quality, safety, and efficacy of a drug is no longer sufficient to ensure reimbursement for use in public markets. In a growing number of areas, evidence of cost-effectiveness is also required1,2. Individual countries have their own pharmacoeconomic guidelines or requirements dictating how cost-effectiveness studies need to be designed and reported, and many country-specific guidelines have been summarized by the International Society for Pharmacoeconomics and Outcomes Research (ISPOR)3.

National guidelines may insist on the presentation of local data, especially if available clinical trials were undertaken wholly outside the jurisdiction of interest and one or more components of evidence cannot be generalized2. Mathematical modeling can provide evidence on clinical and economic outcomes in a form that can promote informed decisions about clinical practices and healthcare resource allocations4. Decision-analytic models also provide a framework within which evidence from a variety of sources can be compiled2. Often a core model is developed early in a product’s life cycle (i.e., when local needs have not been identified). Such core models may then be adapted for local use. For local adaptations, it is important to carefully consider which parameters need to be tailored to the specific healthcare system (e.g., prices and treatment patterns) and which may be generalized (e.g., treatment effect or relative risk reduction)2,5.

Local adaptation of an existing model may allow cost-effectiveness data to be produced in a timely fashion because a lag between the availability of efficacy and cost-effectiveness data could mean that decision-makers are faced with demands to adopt a new drug before adequate health economic data are available, potentially resulting in an inappropriate decision or an unacceptably long delay6.

Global cost-effectiveness models need to be able to deliver a number of components in order for them to be adapted and used locally. Models need to be flexible for input of local data; transparent for clarity of design and results; designed for minimal required adaptation locally; robust to assumptions; and able to be used in health technology assessment, reimbursement, or other decision-making situations. Model developers should conduct studies according to the highest possible standards of quality and communicate results with adequate disclosure of assumptions4.

The incidence of invasive fungal infection (IFI) has risen dramatically over the past 20 years, and costs associated with treatment are high7,8. Preventing IFI in high-risk patients, such as those with acute myelogenous leukemia (AML) or myelodysplastic syndrome (MDS) who have chemotherapy-induced neutropenia, may therefore reduce overall healthcare costs8,9. Posaconazole is an extended-spectrum triazole with demonstrated efficacy as prophylaxis for and treatment of IFI9–12. Posaconazole is approved for use in the US and Europe for the prophylaxis of IFI in immunocompromised patients, and, additionally, in Europe as treatment for refractory IFI13,14.

This report presents a case study of how a cost-effectiveness model was developed for posaconazole in a global headquarters and then distributed and adapted by at least 11 countries.

Methods

A decision-analytic model was developed by Schering-Plough Corp. (subsequently acquired by Merck Sharp & Dohme Corp., Whitehouse, NJ) and i3 Innovus (now OptumInsight, Waltham, MA) to estimate the cost-effectiveness of antifungal prophylaxis therapy with posaconazole among patients with acute myelogenous leukemia (AML) or myelodysplastic syndrome (MDS) at high risk for IFI because of chemotherapy-induced neutropenia (Figure 1). The model was first described in detail in O’Sullivan et al.15 and has subsequently been adapted and used by various third parties. The decision analytic model consisted of an initial decision tree followed by a Markov model with 1-month cycle lengths. In the model, patients were assumed to receive prophylaxis with posaconazole or standard azole therapy (fluconazole or itraconazole), and distributions of patients receiving fluconazole and itraconazole in the comparator group were assumed to be the same as those in a clinical trial of Cornely et al.9 (i.e., 81% fluconazole and 19% itraconazole) conducted in patients with AML or MDS who were considered at high risk of IFI. The duration of antifungal prophylaxis in the model was assumed to equal the mean duration of prophylaxis in the Cornely et al.9 trial. The base case risk for experiencing IFI was assumed to be treatment-specific; probabilities of experiencing an IFI, IFI-related death, and death from other causes during the study period were also estimated with data from the clinical trial9. One-month Markov cycles were used to extrapolate survival to a lifetime horizon. Patients alive at the end of the trial period (100 days) were assumed to have a risk for death specific to underlying disease. Table 1 contains the original model parameters and data sources9,15–18. The probability of IFI with either standard azole therapy (fluconazole/itraconazole) or posaconazole are core parameters that were not changed at the country level when the model was adapted for local use.

Figure 1.  Model structure.

Table 1.  Global model parameters and data sources14,15.

Model Parameter	Estimate	Data source
Efficacy of prophylaxis
Probability of IFI with FLU/ITRA	0.11	Cornely et al.9
Probability of IFI with POS	0.05	Cornely et al.9
Survival after prophylaxis
Probability of IFI-related death,  given IFI^a	0.43	Cornely et al.9
Probability of other death  within 100 days^a	0.16	Cornely et al.9
Relative survival for AML  after 100 days	0.21	NCI (SEER)16 ^b
Relative survival for MDS  after 100 days	0.08	Kantarjian et al.17

^aEstimate was pooled as fatality rates were not markedly different between treatment arms15.

^bCanada also included Cancer Survival Statistics (Statistics Canada)18.

AML, acute myelogenous leukemia; FLU, fluconazole; IFI, invasive fungal infection; ITRA, itraconazole; MDS, myelodysplastic syndrome; POS, posaconazole; SEER, Surveillance, Epidemiology, and End Results.

The core model was used to estimate total costs, IFIs avoided, life-years gained, utilities gained, and incremental cost (if applicable) of posaconazole compared with standard azole therapy from a third-party perspective in various currencies. One-way sensitivity analyses and second-order probabilistic Monte Carlo sensitivity analyses were conducted to assess the robustness of model results and the effects of parameter uncertainty on the study findings, particularly as related to treatment efficacy and IFI costs. The sensitivity analyses varied parameters of the model such as risk of IFI, IFI-related death, other death within 100 days, mean duration of prophylaxis, and the cost of IFI treatment. Table 2 shows the parameters that were adapted locally for each country, including probabilities of death and survival, currency, cost of treating an IFI, and the discount rate (for cost and outcomes)5,15,19–33.

Table 2.  Model parameters that were changed locally11,15–24,29.

	Long-term survival
Country	AML	MDS	Probability of IFI-related death, given IFI	Probability of other death within 100 days	Year^a	Cost of IFI treatment	Discount, %
USA (global model)15	0.21	0.08	0.43	0.16	2006	USD 40,583^b	3
Germany5	German university clinic data		0.48 (FLU/ITRA) 0.36 (POS)	0.16	2008	€21,000^c	–
Spain19	0.21	0.08	0.48 (FLU/ITRA) 0.38 (POS)	0.16 (FLU/ITRA) 0.13 (POS)	2007	€63,35928	3
France20	0.21	0.08	0.43	0.15	2009	€51,033 (calculated)	3
Switzerland21	0.21	0.08	0.43	0.15	Not stated	CHF 83,725	3
Canada22	0.21	0.08	0.43	0.15	2007	CAD 61,98429,30	5
The Netherlands23	0.79^§	0.91^d	0.45	0.16	Not stated	€37,20423,31	1.5
South Korea24	Included additional data from KNSO life table 2006		0.39 (FLU/ITRA) 0.07 (POS)	0.16	Not stated	KRW 34,444,263^e	–
Australia25	0.21	0.08	0.39 (FLU/ITRA) 0.07 (POS)	0.17	Not stated	AUD 186,055	5
Belgium26	0.21	0.08	0.43	0.15	Not stated	€39,203^f	3
Russia27	0.21	0.08	0.48 (FLU/ITRA) 0.36 (POS)	0.15	Not stated	Not stated	5

All countries used local data for the cost of posaconazole, fluconazole, and itraconazole.

^aYear is given if available.

^bUnpublished data from the HCUP-NIS analysis, Schering-Plough Corp., 2007.

^cGerman cost of treating IFI estimated as the difference between the cost for patients with proven/probable IFI vs the cost for patients with possible/no IFI32.

^dDutch long-term survival (probability of 1-year survival): Integraal Kankercentrum Amsterdam33 data 1988–2003 (AML) and 2001–2003 (MDS)23.

^eKorean Hospital Survey HIRA data.

^fIMS Belgian Hospital Disease Database.

€, Euro; £, Great British pound; AML, acute myelogenous leukemia; AUD, Australian dollar; CAD, Canadian dollar; CE, cost-effective; CHF, Swiss franc; CS, cost saving; FLU, fluconazole; HCUP-NIS, Healthcare Cost and Utilization Project Nationwide Inpatient Sample; HIRA, Health Insurance Review Agency; IFI, invasive fungal infection; IMS, International Medical Services; ITRA, itraconazole; KNSO, Korean National Statistics Office; KRW, Korean won; LYS, life-years saved; MDS, myelodysplastic syndrome; POS, posaconazole; QALY, quality-adjusted life-year; RUR, Russian ruble; USD, United States dollar.

The core model and technical report were made freely available to local countries, and in most cases the adaptation and analysis was completed locally. The specific study outcome for this analysis was incremental cost per life-year saved.

Results

Table 3 summarizes the results of the various locally adapted cost-effective modeling studies, which indicate that prophylaxis with posaconazole, compared with fluconazole or itraconazole, prolongs survival, and, in most countries, is cost-saving5,15,19–27. In Germany, South Korea, and Belgium, additional cost per life-year saved was associated with the use of posaconazole compared with fluconazole or itraconazole; however, the incremental cost was well below the pre-determined cost-effectiveness thresholds for these countries.

Table 3.  Cost-effectiveness of posaconazole prophylaxis among high-risk patients with neutropenia.

Country	Change in cost*	LYS	Incremental cost/LYS	Probabilistic sensitivity analysis
USA15	USD −600	0.07	Saving	73% (CS); 96% (CE at USD 50,000/LYS)
Germany5	€690	0.10	€6,820	85% (CE at €20,000/LYS)
Spain19	−€1564	0.09	Saving	87% (CS); 98% (CE at €30,000/LYS)
France20	−€859	0.02	Saving	76% (CS); ∼83% (CE at €30,000/LYS)
Switzerland21	CHF −1118	0.016	Saving	73% (CS); 88% (CE at CHF 100,000/LYS)
Canada22	CAD −185	0.016	Saving	53% (CS); 70% (CE at CAD 50,000/LYS)
Netherlands23	−€183	0.10	Saving	90% (CE at €20,000/QALY)
Scotland	−£188	Not stated	Saving	Not stated
South Korea24	KRW 1,457,512	0.092	KRW 15,798,087	50% (CE at KRW 22,000,000)
Australia25	AUD −10,076	0.09	Saving	Not done
Belgium26	€79	0.067	€1173	83% (CE at €40,000/LYS)
Russia27	RUR −24,779	0.02	Saving	70% (CE at RUR 1,775,000/LYS)

*Change in cost is the total cost of treatment with posaconazole compared with the total cost of treatment in the comparator arm; a negative figure therefore indicates cost-saving for posaconazole.

AUD, Australian dollar; CAD, Canadian dollar; CE, cost-effective; CHF, Swiss franc; CS, cost saving; €, Euro; £, Great British pound; KRW, Korean won; LYS, life-years saved; QALY, quality-adjusted life-years; RUR, Russian ruble; USD, United States dollar.

All reported probabilistic sensitivity analyses (PSA) showed that prophylaxis with posaconazole was likely to be cost-effective at the pre-determined cost-effectiveness threshold for that country (Table 3)5,15,19–27. It should be noted that, while the model allowed for PSA to be conducted, countries that were not required to conduct a PSA as part of their local reimbursement process may not have conducted the analysis.

In all countries, the model predicted that prophylaxis with posaconazole would be associated with an increase in life-years, with increases ranging from 0.016–0.1 life-year saved.

Table 4 shows the outcome of the modeling study in each country from a payer’s perspective; in all countries, posaconazole was approved by the respective reimbursement authority5,15,19–27. In the first instance, the analyses were presented in health technology assessment reports to the reimbursement authorities; in most countries, data from the locally adjusted model were also published in journals or presented at congresses.

Table 4.  Outcome of use of the locally adapted global model.

Country	Publication	Outcome
USA (global model)	Published in Value in Health 200915	Reimbursement Local adaptation for USA
Germany	Presented at ECCMID 20085	Reimbursement^a
Spain	Presented at ECCMID 200819	Reimbursement
France	Published in J Med Econ 201120	Reimbursement
Switzerland	Published in Oncology 201021	Reimbursement
Canada	Presented at ECCMID 200822	Reimbursement
Netherlands	Published in Eur J Haematol 200823	Reimbursement
Scotland	Not published	Reimbursement
South Korea	Presented at ISPOR 200924	Reimbursement
Australia	Presented at HTAi 200925	Reimbursement
Belgium	Presented at ISFN 200826	Reimbursement^b
Russia	Published in CMAC27	Reimbursement

^aAlso distributed for local use.

^bAlso used by hospital formulary.

CMAC, Clinical Microbiology and Antimicrobial Chemotherapy (Russia); ECCMID, European Congress of Clinical Microbiology and Infectious Diseases; Eur J Haematol, European Journal of Haematology; HTAi, Health Technology Assessment International; ISFN, International Symposium on Febrile Neutropenia; ISPOR, International Society for Pharmacoeconomics and Outcomes Research; J Med Econ, Journal of Medical Economics.

Discussion

This case study reports how a global cost-effectiveness model was successfully adapted to local requirements by importation of local parameters. The study-specific outcome was incremental cost/life-year saved. This analysis shows that, in most countries, use of posaconazole is cost-saving5,15,19–27. There was an additional cost per life-year saved in Germany, South Korea, and Belgium; however, the incremental cost was well below the pre-determined cost-effectiveness thresholds for these countries. Local use of the core model led to reimbursement of posaconazole by local authorities in all countries.

As noted by O’Sullivan et al.15 in the original publication of the posaconazole cost-effectiveness model, a number of limitations should be noted: the lack of power to detect differences between posaconazole and fluconazole or itraconazole separately; the assumption that risk of death after 100 days was equal for those who did and did not develop IFI; and, in the original model, the use of equal probabilities of IFI-related and other death during the trial period for both prophylaxis groups, even though overall and IFI-related survival benefits for the posaconazole group were demonstrated in Cornely et al.9 When treatment-specific case fatality rates were used as an alternative analysis in O’Sullivan et al.15, however, posaconazole was also observed to be cost-saving. Furthermore, when treatment-related death probabilities were used in various local adaptations of the model (Table 2)5,15,19–33, posaconazole cost-saving or cost-effectiveness was consistently demonstrated (Table 3)5,15,19–27.

Interestingly, an independent decision-analytic model developed by Collins et al.34 at the University of Michigan Health System showed results similar to those of the global model, in which posaconazole prophylaxis was associated with overall treatment costs lower than treatment costs associated with fluconazole or itraconazole prophylaxis in patients with prolonged neutropenia. A recent independent review by Lyseng-Williamson35 has also reviewed the cost-effectiveness analyses of posaconazole in this population, presenting results from this original core model. The review reports findings similar to those of our own research; in spite of some inherent limitations, data from several countries support the use of posaconazole as a dominant or cost-effective antifungal prophylaxis relative to standard oral azole prophylaxis in patients with neutropenia at high risk of developing IFI35.

The model has the capacity to run both univariate sensitivity analyses, as well as the more robust probabilistic sensitivity analyses. Results from both types of sensitivity analysis suggest that the model is sensitive to variations in the risk of IFI and the mean duration of therapy with posaconazole. In the base case analysis, the probabilistic sensitivity analysis suggests that posaconazole has a 90% probability of having an incremental cost-effectiveness ratio at or below a $50,000 per life-year saved threshold and a 95% probability of being at or below a $100,000 per life-year saved threshold.

Conclusion

A core economic model was developed and successfully adapted locally by several countries. The model showed that posaconazole was cost-saving or cost-effective vs standard azole therapy (fluconazole and itraconazole) and led to positive reimbursement listings.

Transparency

Declaration of funding

This work was funded by Merck Sharp & Dohme Corp., Whitehouse Station, NJ, USA.

Declaration of financial/other relationships

Manishi Prasad is an employee of Merck Sharp & Dohme Corp., a subsidiary of Merck & Co., Inc., Whitehouse Station, NJ. At the time of the study, George Papadopoulos was an employee of Merck Sharp & Dohme Corp. Sheena Hunt is an employee of ApotheCom, a company that received funding from Merck Sharp & Dohme Corp., for help in the publication of this study. JME Peer Reviewers on this manuscript have no relevant financial relationships to disclose.

Acknowledgments

The authors gratefully acknowledge the support of i3 Innovus, Medford, MA, USA, in the development of the model. The authors would also like to thank the researchers in all the countries who used and adapted this model locally.