Cost–effectiveness of adding ezetimibe to atorvastatin vs switching to rosuvastatin therapy in Portugal

Abstract

Background:

Statin monotherapy is the mainstay of low-density lipoprotein cholesterol (LDL-C) management for high cardiovascular risk patients in Portugal; however, several therapeutic options are available and predicted to have different clinical and economic impacts. The aim of this study was to evaluate the cost-effectiveness of adding ezetimibe 10 mg (EZ10) to atorvastatin 10 or 20 mg (A10/20) vs switching to rosuvastatin 10 or 20 mg (R10/20) in Portuguese patients with coronary heart disease (CHD) and/or diabetes who are currently above the LDL-C goal.

Methods:

A Markov model was used to describe CHD disease progression and the lifetime costs and utilities associated with each disease state were used to estimate the gains in life-years and quality-adjusted life-years (QALYs), as well as the incremental cost-effectiveness ratio (ICER), of the two treatment regimens. Model inputs, such as age, gender, and prevalence of cardiovascular risk factors of the dyslipidemic Portuguese patients were obtained from the Portuguese cohort of the Dyslipidemia International Study (DYSIS). The efficacy of each treatment regimen, the cost of drugs and of treating CHD events, and the utilities for each disease state were derived from published sources.

Results:

The estimated lifetime discounted number of QALYs gained by patients treated with A10/20 was 8.70, while in those switching to R10/20 it was 8.81 and in those adding EZ10 it was 8.93. Discounted total health costs were estimated to be €11,131 for A10/20, but €14,511 and €16,571 for R10/20 and A10/20 + EZ10, respectively. The ICER of adding ezetimibe vs switching to rosuvastatin was €16,465/QALY. Based on the Portuguese cost-effectiveness willingness-to-pay threshold of €30,000/QALY, adding ezetimibe vs switching to rosuvastatin would be a cost-effective use of resources in Portugal. Sensitivity analyses in patients with differing clinical histories (CHD or diabetes or both) yielded similar values, with no ICER over €30,000/QALY.

Conclusions:

From the perspective of the National Health Service, prescribing ezetimibe to high cardiovascular risk patients being treated with atorvastatin vs switching them to rosuvastatin is projected to be a cost-effective use of resources in Portugal.

Keywords::

• Cost-effectiveness
• Ezetimibe
• Statin
• Rosuvastatin
• Atorvastatin
• Coronary heart disease
• Cardiovascular disease

Introduction

In the European Union, cardiovascular diseases (CVDs), which include coronary heart disease (CHD) and cerebrovascular diseases, are the major causes of morbidity and mortality1. Diseases of the circulatory system are the major causes of death in Portugal, accounting for ∼1/3 of all deaths2. The main risk factors contributing to CVDs, and the ones which are modifiable, are elevated cholesterol, hypertension, smoking, and obesity3–5. The results of a recently published meta-analysis of randomized controlled trials suggest the incidence of major vascular events can be reduced by ∼20% for each 1 mmol/L decrease in low-density lipoprotein cholesterol (LDL-C)4.

The mainstay of LDL-C management in Portugal is statin monotherapy. However, in the Dyslipidemia International Study (DYSIS), a multi-center, cross-sectional study designed to assess the lipid profiles of patients treated with statins in 11 European countries and Canada, the results from Portugal showed that ∼50% of patients (including high cardiovascular risk patients) failed to attain the therapeutic goal for LDL-C; over half of the study population had high cardiovascular risk, and 50% of them had pre-existing CVD6. Other studies7–9 have reported that, despite the efficacy of statins in lowering LDL-C levels, many patients who are at high risk for heart disease with hypercholesterolemia require additional LDL-C level reduction, and these patients are the least likely to reach established goals. Thus, there is a clear need for safe agents that improve the attainment of lipid goals7.

One strategy used to increase achievement of target LDL-C levels is to change from a less effective to a more effective statin. Rosuvastatin has been shown to reduce LDL-C more effectively than many other statins10–12, and also is a safe therapeutic option for high cardiovascular risk patients to achieve their lipid and apolipoprotein targets9. Alternative treatment strategies, such as ezetimibe co-administration, can help patients to achieve greater lipid control through dual inhibition, and potentially lower their CHD risk7. There is evidence that adding ezetimibe to the 10 mg/day starting dose of atorvastatin can facilitate LDL-C goal attainment in high cardiovascular risk patients with hypercholesterolemia, even with lower doses of the statin7. Recently, a large trial demonstrated incremental clinical benefit when adding ezetimibe to statin therapy by reducing cardiovascular events13^,14.

Decision-makers need to have information about the impact of these therapeutic options on long-term incidence of CHD as well as the projected lifetime costs and benefits relative to existing lipid-lowering therapies. A Markov model, developed and validated by Cook et al.15, projecting the long-term benefits and costs of alternative lipid-lowering strategies in patients with hypercholesterolemia has been tested in countries like Finland3, Germany, Spain, and Norway16; the UK5, and the Netherlands17, with results suggesting that ezetimibe-statin co-administration is a more cost-effective treatment than statins alone.

In Portugal, atorvastatin is currently the preferred statin for second line treatment of dyslipidemic patients following simvastatin, while rosuvastatin continues to be the leading brand for the treatment of LDL-C. Usually, for those uncontrolled atorvastatin treated patients who require more effective LDL-C reductions, prescribers decide to switch them to rosuvastatin treatment. Therefore, it is relevant to test this common treatment pattern against an alternative scenario of adding ezetimibe to atorvastatin. The purpose of this analysis was to evaluate the cost-effectiveness of adding ezetimibe to atorvastatin 10 or 20 mg in high cardiovascular risk CHD and/or diabetes patients currently not at LDL-C goal vs switching from atorvastatin to rosuvastatin in Portugal. The analysis was based on the aforementioned published Markov model by Cook et al.15 developed to project and evaluate the lifetime costs and health outcomes, including life-years (LYs).

Methods

Patient population

The population included in this analysis was high cardiovascular risk patients with a history of CHD and/or diabetes who did not reach LDL-C target levels (≥2.5 mmol/L or ≥100 mg/dL) with atorvastatin 10 mg or 20 mg (A10/20) treatment. Patient profiles were obtained from the Portuguese cohort of the DYSIS and are described in Table 1. A total of 37 patients met the inclusion criteria. The DYSIS protocol was approved by the Portuguese National Data Protection Commission and written consent was obtained from all patients, in accordance with the Declaration of Helsinki6.

Table 1. Patient profiles.

	Atorvastatin (n = 37)*
Demographics
Age (mean, SD)	65.73 (10.22)
Male gender	18 (49%)
Hypertension
DBP (mean, SD)	77.35 (10.43)
SBP (mean, SD)	134.46 (18.43)
Medications for hypertension	21 (57%)
Cholesterol (mean, SD)
LDL-C	132.33 (27.30)
HDL-C	49.19 (8.69)
Triglycerides	168.65 (117.32)
Total cholesterol	219.32 (40.99)
CHD risk factors
History of CHD	18 (49%)
Diabetes	30 (81%)
Smoking (yes)	13 (35%)
Treatment regimen
Atorvastatin 10 mg	16 (43%)
Atorvastatin 20 mg	21 (57%)

*Values are presented as n (%) unless indicated otherwise.

DBP, diastolic blood pressure; SBP, systolic blood pressure. LDL-C, low-density lipoprotein cholesterol; HDL-C, high-density lipoprotein cholesterol; CHD, coronary heart disease.

Model

This was a cost-effectiveness analysis performed with a Markov model for CHD disease progression as previously described15. As depicted in Figure 1, this model allows for patients to enter either in the primary or secondary CHD prevention states. Patients without a history of CHD (primary prevention diabetic or CHD risk equivalent based on National Cholesterol Education Program Adult Treatment Panel III18 or other guidelines19) enter the model in the primary prevention state, and those with a history of CHD enter the model in the secondary prevention state. The model was developed to project lifetime (in annual cycles up to the age of 100 years) costs and benefits of lipid therapy in high cardiovascular risk patients with CHD and/or diabetes who did not reach LDL-C target levels on A10/20 and were, therefore, switched to receive either atorvastatin plus ezetimibe 10 mg (EZ10) or rosuvastatin 10 or 20 mg (R10/20).

Figure 1. Decision tree model structure.

In the model, different health states were defined according to whether the patient had previously experienced a CHD event and whether the patient experienced a fatal or non-fatal event during the annual cycle. Each health state was assigned an expected cost and utility in order to estimate the cost-effectiveness of the alternative treatment options. The probability of annual movement between health states depended on the patient’s risk factor profile and their current health state. The risk of CHD events was based on the Framingham Heart Study risk equations20, and the risk of non-CHD deaths was based on national statistics on death rates2. Cardiovascular disease mortality rates and overall death rates (Table 2) were obtained from the National Institute of Statistics (year 2009)2. Non-CHD death rates by age and gender were calculated by subtracting the number of CHD deaths from the overall mortality and dividing these totals by the population total in each group.

Table 2. Non-CHD death rates per 100,000 inhabitants by age and gender.

Age range	Non-CHD death rates
	Male	Female
20–24	58.3	24.4
25–29	58.3	24.4
30–34	58.3	24.4
35–39	207.6	95.8
40–44	207.6	95.8
45–49	483.8	210.5
50–54	483.8	210.5
55–59	929.4	409.0
60–64	929.4	409.0
65–69	2130.6	1075.3
70–74	2130.6	1075.3
75–79	8229.8	6588.0
80–84	8229.8	6588.0
85+	8229.8	6588.0

Data source: Instituto Nacional de Estatistica, Statistics Portugal. Portuguese Life Table 2007–2009. http://www.ine.pt/ [Accessed October 1, 2013].

Efficacy

Efficacy data (i.e. the percentage change) for total cholesterol and high-density lipoprotein cholesterol (HDL-C) was obtained from the study described by Stein et al.7 for the addition of ezetimibe to atorvastatin. For the switch from atorvastatin to rosuvastatin, total cholesterol efficacy data was approximated from the results of the Mercury I8 and II9 studies (Table 3).

Table 3. Summary of efficacy data (% change).

Treatment change	Lipid parameter	% change (SE)	Notes
Addition of ezetimibe	TC	−17.3 (0.6)	Derived from Stein et al.7
	HDL-C	2.1	Derived from Stein et al.7
Switching from atorvastatin to rosuvastatin	TC	−8.0 (0.6)	Approximated from Schuster et al.8 and Ballantyne et al.9
	HDL-C	1.0	Rosuvastatin assumed to have the same variability as adding ezetimibe

HDL-C, high-density lipoprotein cholesterol; TC, total cholesterol.

Utility weights

Data inputs for health state utilities were retrieved from previous studies21–24, as defined in Table 4. For fatal events (fatal myocardial infarction or non-CHD death), the utility value was assumed to be zero, while the primary prevention state, with no previous history of CHD, was equated with perfect health (i.e. utility = 1).

Table 4. Summary of health state utilities.

Health state	Utility value	Notes
Healthy primary prevention	1	No previous history of CHD.
Non-fatal MI	0.76	Derived from Cook et al.16, van Nooten et al.17, Reiner et al.19
Angina pectoris	0.77	Derived from Cook et al.16, van Nooten et al.17, Reiner et al.19
Subsequent year CHD	0.808	Derived from NCEP18 and assumed to be the same as someone with stable angina
Healthy secondary prevention	0.808	Derived from NCEP18 and assumed to be the same as someone with stable angina.
Fatal MI	0
Non-CHD death	0

CHD, coronary heart disease; MI, myocardial infarction.

Costs

The analysis was performed from a third party payer perspective (that of the National Health Service). Future costs and outcomes were discounted at 5% per annum, as recommended by the Portuguese guidelines for health economics evaluations. The time horizon considered was lifetime (up to the age of 100 years).

Drug costs were obtained from the National Authority of Medicines and Health Products (INFARMED)25 in February 2015. For atorvastatin, the generic price was considered (reference price). For primary prevention patients (including diabetics with no previous CHD event), the annual costs considered were the costs of two physician visits and two sets of laboratory tests26^,27. The annual cost of treating an event included all the resource consumption (e.g. hospitalizations, consultations, medication, laboratory tests, and examinations) needed to treat the event, as described in Table 513.

Table 5. Health states and medication costs.

Resource	Cost (€)
Events (annual cost)
Healthy primary prevention	141.40
Non-fatal MI	8010.01
Angina pectoris	4180.94
Fatal MI	6632.87
Subsequent year CHD	1790.49
Non-CHD death in primary state*	70.70
Non-CHD death in secondary state†	1087.98
Medication (daily cost)
Rosuvastatin 10 mg	0.80
Rosuvastatin 20 mg	1.23
Ezetimibe 10 mg	1.53

*Half the annual cost of primary prevention.

†Half the annual cost of the subsequent CHD year.

Cost for the year in which a non-fatal event occurred included both the event costs and the subsequent CHD maintenance costs. Once a non-fatal event occurred or the patient entered the secondary prevention state, the patient received only the subsequent CHD maintenance costs26^,27. If a non-CV fatal event occurred in the primary prevention state, the costs considered were half those of a year of the primary prevention state. If a non-CV fatal event occurred in the secondary prevention state, the costs considered were half those of a year of the subsequent CHD annual costs.

Portuguese-specific annual direct costs of CVD health states were estimated based on a national expert panel discussion conducted by the Delbecq method. The expert panel consisted of Portuguese clinical experts (general practitioners and specialists) in CVD treatment representative of all geographic main regions of Portugal. Experts answered questions concerning the mean resource consumption associated with each CVD health state26^,27. Direct annual costs for each health state were determined by multiplying resource consumption by the unit cost of each individual resource (hospitalizations, consultations, medication, laboratory tests, and examinations) (Table 5).

Cost-effectiveness analyses

An analysis was performed to compare the cost-effectiveness of adding ezetimibe to atorvastatin vs switching to rosuvastatin. The incremental cost-effectiveness ratio (ICER) was estimated as the ratio between the difference in the total cost of the interventions and the difference in LYs or quality-adjusted life-years (QALYs) gained.

Sensitivity analyses

One-way sensitivity analyses were performed to test the robustness of the findings and evaluate the impact of uncertainty in the following parameters: discount rate (0% and 7%), annual CHD risk estimates (±10%), event costs (±20%), QALY estimates (±10%), total cholesterol efficacy of switching to rosuvastatin (95% CI), and total cholesterol efficacy values (95% CI). The analyses were also performed on the following pre-specified sub-groups of patients: those with CHD only, those with diabetes only, and those with both diabetes and CHD.

Results

As shown in Table 1, Portuguese patients undergoing treatment with A10/20 who met all the inclusion criteria for this analysis had a mean age of 65.7 years. Among the patients, 35% were current smokers, 57% were receiving medications for hypertension, 81% were diabetic, and 49% had previous history of CHD. In addition, the mean HDL-C was 49 mg/dL, triglyceride was 169 mg/dL, and LDL-C was 132 mg/dL.

In Table 6, the base-case results are presented, with discounted and undiscounted averages for costs, LYs and QALYs. Patients treated with A10/20 had an average life expectancy of 14.84 undiscounted LYs, while patients switching to R10/20 or adding EZ10 to atorvastatin gained an additional 0.27 and 0.58 undiscounted LYs, respectively. Moreover, the average discounted number of QALYs gained by patients treated with A10/20 was 8.70, while in those switching to R10/20 it was 8.81 and in those adding EZ10 it was 8.93. Discounted total health costs were estimated to be €11,131 for A10/20, but €14,511 and €16,571 for R10/20 and A10/20 + EZ10, respectively. Based on these results, the base-case ICER was calculated as €16,465 (Table 6). Considering the Portuguese cost-effectiveness willingness-to-pay threshold of €30,000/QALY, adding ezetimibe vs switching to rosuvastatin would be a cost-effective use of resources in Portugal.

Table 6. Base-case results for the addition of ezetimibe to atorvastatin vs switching to rosuvastatin.

	A10/20	Switch to R10/20	A10/20 +  EZ10
Costs (€)
Discounted	11,131	14,511	16,572
Undiscounted	17,146	22,518	25,894
Life-years (LYs)
Discounted	9.55	9.67	9.80
Undiscounted	14.84	15.11	15.42
Quality-adjusted life-years (QALYs)
Discounted	8.70	8.81	8.93
Undiscounted	13.56	13.81	14.09
Incremental cost-effectiveness ratio (€/QALYs)	16,465

A10/20, atorvastatin 10 mg or 20 mg; R10/20, rosuvastatin 10 mg or 20 mg; EZ10, ezetimibe 10 mg.

Sensitivity analysis

Allowing the variables in the model to fluctuate around the base-case scenario, as described in the Methods, yielded ICERs below the €30,000/QALY threshold (Table 7). The variables with the largest effect on the ICER were the discount rate and the total cholesterol efficacy. When the base-case data were analyzed separately by disease group, patients with diabetes only and with both CHD and diabetes produced the highest and lowest ICERs: €19,977/QALY and €11,820/QALY, respectively (Table 7).

Table 7. Sensitivity analysis results.

Variable	ICER (€/QALY)	% Change
Base case	16,465
Discount rate
0%	11,886	−28%
7%	18,566	13%
Annual CHD risk
−10%	17,826	8%
10%	15,349	−7%
Event costs
−20%	16,505	0.24%
20%	16,424	−0.25%
QALY estimates
−10%	17,102	4%
10%	15,873	−4%
Efficacy for TC
Switching to rosuvastatin
Lower 95% CI	14,829	−10%
Upper 95% CI	18,530	13%
Ezetimibe Addition
Lower 95% CI	18,588	13%
Upper 95% CI	14,778	−10%
Patient sub-group
Diabetes only	19,977	21%
CHD only	19,312	17%
CHD and diabetes	11,820	−28%

CHD, coronary heart disease; CI, confidence interval; ICER, incremental cost-effectiveness ratio; QALY, quality-adjusted life-year; TC, total cholesterol.

In addition, we performed an analysis using data from DYSIS, where we assumed that the characteristics of patients taking simvastatin in that study were similar to the atorvastatin patients in the current study. This assumption was made because the data in DYSIS were collected before atorvastatin became available as a generic formulation. Our reasoning was that generic atorvastatin will be prescribed more frequently by physicians in the future and, consequently, it is reasonable to expect that future dyslipidemia populations treated with atorvastatin will have characteristics similar to the DYSIS population treated with simvastatin. When we compared the results of this analysis to the base case, there was little difference between the costs and QALYs observed and the ICER was €21,494/QALY.

Discussion

The purpose of this analysis was to evaluate the cost-effectiveness of the addition of ezetimibe to high cardiovascular risk patients not at goal (i.e. LDL-C ≥ 100 mg/dL) on atorvastatin in comparison to switching to rosuvastatin. The results suggest that adding ezetimibe to atorvastatin provides more gains in LYs and QALYs than switching from atorvastatin to rosuvastatin and that it is more cost-effective. In this study, the incremental cost-effectiveness ratio of switching patients from A10/20 to A10/20 + EZ10 vs to R10/20 was €16,465/QALY, which means that atorvastatin + ezetimibe provides greater value for patients at high cardiovascular risk for CHD and who are not at the LDL-C goal.

Due to differences in patients’ baseline characteristics and the treatment regimens evaluated, and because base-case information is country-specific, comparison of these results with other countries is difficult. However, previous studies3^,5^,14^,15^,28 have suggested that adding ezetimibe to a statin was cost-effective when compared to statin monotherapy. In Germany, Spain, and Norway, co-administration of ezetimibe with a statin in patients who failed to reach their lipid goal was projected to yield cost-effectiveness ratios under €18,000/LY gained in CHD patients and under €30,000/LY gained in non-CHD diabetic patients when compared with statin-only treatment14. The corresponding ratios for a statin titration strategy ranged up to €25,949/LY gained and €47,561/LY gained, respectively14. The results of a Dutch study showed that the cost-effectiveness of adding ezetimibe to simvastatin vs continuing atorvastatin or simvastatin monotherapy was under the acceptable threshold of €30,000/QALY15. Likewise, in a Finnish study, switching from generic simvastatin to co-administration with branded ezetimibe was more cost-effective than switching to generic atorvastatin or branded rosuvastatin in the secondary prevention population, not at the LDL-C goal3. In the UK, switching to ezetimibe plus simvastatin 40 mg had an estimated ICER of £11,571/QALY vs doubling the statin dose in patients with acute coronary syndrome (the INFORCE cohort)5. An earlier cost-effectiveness analysis of a hypothetical cohort of 1000 UK men aged 55 years predicted that the combination of ezetimibe plus a statin would prevent an additional 43 non-fatal myocardial infarctions, seven non-fatal strokes, and 26 cardiovascular deaths over a lifetime, compared with doubling the statin dose26.

Sensitivity analyses comparing ezetimibe co-administration vs switching to rosuvastatin showed that the results were robust, in line with results from other European countries14. The results were most sensitive to changes in the variables of patient history, CHD risk, total cholesterol efficacy, and discount rates. Nevertheless, the maximum values were always below the €30,000/QALY threshold, even when considering diabetic patients only. Furthermore, according to the results, treating patients with ezetimibe + atorvastatin would be increasingly more cost-effective than treating patients with rosuvastatin with increasing CHD risk, since ICER values decreased with increasing risk.

This analysis had several limitations. First, like all model-based studies, there are limitations due to the assumptions that had to be made. Cardiovascular events were estimated and not directly observed. However, modeling is still the best available technique to estimate long-term events where observational studies are unfeasible. The model used in this study has been previously validated and has been demonstrated to project CV events accurately14.

Second, the disease model relied on the relationship between lipid levels and risk of CHD, and consequently the prediction of outcomes is not based on data from large outcomes studies. However, the linear relationship between LDL-C and cardiovascular events is well established (e.g. CTT meta-analysis)4 and, although it stems mainly from data derived from clinical trials of statin therapy, some recently presented data suggests additional reduction of CVD risk following use of ezetimibe (i.e. IMPROVE-IT13^,14 and SHARP29 studies). Third, it was not possible to derive the mean change in lipids when switching from atorvastatin to rosuvastatin directly, because the publications for the Mercury I/II studies only presented data as changes from randomization. Lastly, other factors which might influence the utilization and due effectiveness of lipid-lowering agents, such as adherence and tolerability profiles, weren’t considered. Nevertheless, based on the available evidence it is not expected that these factors would significantly favor any of the evaluated strategies and, therefore, change the overall results.

Achievement of LDL-C goals is an important objective of lipid-lowering therapy in clinical practice. However, it has been widely shown that many high risk CHD patients fail to achieve their goals and even patients who have reached LDL-C goals may still be at higher CV risk due to other risk factors. So it is clear, at all levels, that there is a global need to provide patients with more effective statins or, alternatively, agents that can be combined with low- and high-dose statins to effectively increase the attainment of lipid goals. Even though atorvastatin was, for a time, recognized as one of the most effective statins in enabling patients to reach their LDL-C goals, rosuvastatin has been found to reduce LDL-C more effectively than atorvastatin, as well as other statins8–12. In addition, recent trials7 have shown that the added efficacy of ezetimibe co-administered with atorvastatin can facilitate LDL-C goal attainment in high cardiovascular-risk patients with hypercholesterolemia, even with lower doses of the statin. The current study provides evidence of the cost-effectiveness of atorvastatin plus ezetimibe in high risk CHD patients. This treatment option was predicted to provide more LYs and QALYs gained than switching to rosuvastatin.

With the current economic concerns and cost constraints, economic evaluations are critical to provide decision-makers with the necessary information to make well informed choices. It is believed that these results will inform such decisions regarding the treatment of dyslipidemia in Portugal. They particularly highlight ezetimibe as a relevant LDL-C treatment option expected to generate efficiency gains to the National Healthcare Service under the tested circumstances.

Conclusions

Adding ezetimibe on statin regimen has been shown to be effective in lowering LDL-C. Moreover, based on the analysis conducted, it is projected to reduce CHD events and provide gains in longevity and quality-of-life. Our analysis suggests that the addition of ezetimibe 10 mg to atorvastatin 10 or 20 mg in high cardiovascular risk patients not at the LDL-C goal will be cost-effective compared to switching to rosuvastatin 10 or 20 mg and that the ICER is well below the preferred value of €30,000/QALY in Portugal.

Abbreviations:
A10/20	=	Atorvastatin 10/20 mg
A20/40	=	Atorvastatin 20/40 mg
CHD	=	Coronary heart disease
CVD	=	Cardiovascular disease
dL	=	Deciliter
DYSIS	=	Dyslipidemia International Study
EZ10	=	Ezetimibe 10 mg
HDL-C	=	High-density lipoprotein cholesterol
ICER	=	Incremental cost-effectiveness ratio
L	=	Liter
LDL-C	=	Low-density lipoprotein cholesterol
LY	=	Life-years
mg	=	Milligram
mmol	=	Millimole
QALY	=	Quality-adjusted life-years
R10/20	=	Rosuvastatin 10/20 mg
TC	=	Total cholesterol

Transparency

Declaration of funding

This study was funded by Merck Sharp & Dohme.

Declaration of financial/other relationships

All authors are current employees of Merck and all hold stocks and/or stock options.