A Health Economic Model to Determine the Long-Term Costs and Clinical Outcomes of Raising Low HDL-Cholesterol in the Prevention of Coronary Heart Disease

By Roze, S; Liens, D; Palmer, A; Berger, W; Et al

ABSTRACT

Objectives: The aim of this study was to describe a health economic model developed to project lifetime clinical and cost outcomes of lipid-modifying interventions in patients not reaching target lipid levels and to assess the validity of the model.

Methods: The internet-based, computer simulation model is made up of two decision analytic sub-models, the first utilizing Monte Carlo simulation, and the second applying Markov modeling techniques. Monte Carlo simulation generates a baseline cohort for long-term simulation by assigning an individual lipid profile to each patient, and applying the treatment effects of interventions under investigation. The Markov model then estimates the long-term clinical (coronary heart disease events, life expectancy, and quality-adjusted life expectancy) and cost outcomes up to a lifetime horizon, based on risk equations from the Framingham study. Internal and external validation analyses were performed.

Results: The results of the model validation analyses, plotted against corresponding real-life values from Framingham, 48, AFCAPS/ TexCAPS, and a meta-analysis by Gordon et al., showed that the majority of values were dose to the y = x line, which indicates a perfect fit. The R^sup 2^ value was 0.9575 and the gradient of the regression line was 0.9329, both very dose to the perfect fit (=1).

Conclusions: Validation analyses of the computer simulation model suggest the model is able to recreate the outcomes from published clinical studies and would be a valuable tool for the evaluation of new and existing therapy options for patients with persistent dyslipidemia.

Introduction

Cardiovascular (CV) disease, the leading cause of premature death and disability in Westernized societies, is a major contributor to the escalating costs of health care. In the US, coronary heart disease (CHD) accounts for approximately one in every five deaths per year and is associated with an annual morbidity cost of more than $142 billion1. A similar pattern of disease has been observed in Europe, with CHD accounting for 1.95 million deaths each year, with over one in five men and women dying from the disease2. Moreover, CV disease cost the healthcare systems of the European Union around euro105 billion in 2003, with CHD accounting for just under euro23 billion2. In the UK, CHD is responsible for around 1.75 billion per year in health care system expenditure3.

Among associated CV risk factors, lipid abnormalities (dyslipidemia) play a major role. In addition to dietary and lifestyle changes, the primary focus of lipid management in patients with or at high risk of developing CHD is the lowering of low density lipoprotein cholesterol (LDL-c), with an HMG-CoA reductase inhibitor (statin)4,5. A recent prospective meta-analysis of 14 randomized trials (data from 90056 patients) confirmed that statin treatment was effective in reducing CV events in most patient subgroups, irrespective of level of risk (primary or secondary prevention), gender, age or presence of Type 2 diabetes6. Furthermore, each mmol/L decrease in LDL-c was associated with a significant 21% reduction in major vascular event (equivalent to 0.5% per mg/ dL). However, the 5-year absolute remaining risk of a major CV event was 60% higher in patients presenting with a high density lipoprotein cholesterol (HDL-c) 0.9 mmol/L as compared to patients with HDL-c > 1.1 mmol/L. Roughly one patient in about five with low HDL-c or prior myocardial infarction suffered a CV event, a markedly higher rate than the roughly one patient in seven from the overall population with a CV event6. These data confirm that the adverse influence of low HDL-c and prior MI on CV prognosis is strong and similar in magnitude.

Addressing LDL-c lowering alone is not sufficient. Atherosclerosis is a complex process that is a balance between atherogenic factors, mainly ApoB-containing particles [LDL, VLDL, Lp(a)] and anti-atherogenic factors, mainly Apo A1-containing particles (HDL). The independent relationships of LDL-c and HDL-c for CV risk were identified 30 years ago in the Framingham Heart study7. This landmark study was the first to identify low HDL-c as an independent cardiovascular risk factor. In this analysis patients were divided into tertiles for both LDL-c and HDL-c. High LDL-c exerted little effect on coronary risk in patients with high HDL-c. In contrast, low HDL-c markedly increased the risk of coronary heart disease at any level of LDL-c7. In 1998, the CHD risk in relation to HDL-c and LDL-c was subsequently confirmed in the UK Prospective Diabetes Study population by stepwise multivariate Cox analysis. The two first factors associated with CHD in the final model were LDL-c followed by HDL-c (p < 0.001) and then HbA^sub 1c^, systolic blood pressure and smoking8. More recently, the INTERHEART study confirmed a strong linear relationship between the ApoB/Apo A-I ratio and CV risk that accounted for up to 60% of the population attributable risk9.

The importance of HDL-c as an independent CV risk factor and the importance of treating it are recognized in several national and international clinical guidelines. In 1993, the Adult Treatment Panel of the National Cholesterol Education Program recommended the use of drug therapy that targeted HDL-c levels, and the Adult Treatment Panel III (ATP III) guidelines developed in 2001 supported this recommendation4,10. Moreover, a target HDL-c of < 1 mmol/L was identified by the International Atherosclerosis Society in 2003 as a major risk factor for atherosclerotic disease development, and the Third Joint Task Force of European and Other Societies on Cardiovascular Disease Prevention in Clinical Practice reported that HDL-c levels < 1 mmol/L in males and < 1.2 mmol/L in females were markers of increased risk of CHD5'11. In addition, the 2005 French guidelines, from the Agence Franaise de Scurit Sanitaire des Produits de Sant (AFSSAPS) on the treatment of dyslipidemia, recognized that an HDL-c level below 1.03 mmol/L (40 mg/dL) is associated with an increased risk for CV events and recommend the initiation of a therapeutic treatment in such cases12.

Altogether, these data indicate that a substantial number of patients with dyslipidemia who are receiving monotherapy with a statin may well require additional treatment to reach recommended target HDL-c levels, and reduce the risk of CV disease events and the enormous healthcare expenditure attributable to CHD. A recent study has shown that a new add-on strategy, when co-administrated with statin, has been shown to effectively raise HDL-c in patients with CHD and arrest the progression of atherosclerosis13. In order to support healthcare decision makers when considering the reimbursement of add-on lipid-modifying strategies such as this, a model was developed to project lifetime clinical and cost outcomes of such therapy in patients not at cholesterol goal. The aim of this paper was to describe the model in some detail (in line with the ISPOR Task Force recommendations on computer modeling) and to compare the clinical outcomes of preliminary analyses with the model to observed outcomes from Framingham, and from the Scandinavian Simvastatin Survival Study (4S), the Air Force/Texas Coronary Atherosclerosis Prevention Study (AFCAPS/TexCAPS) and a further study (metaanalysis) also not used to construct the model14-18. The model was developed by CORE – Center for Outcomes Research and the accompanying analyses were funded by an unrestricted grant from Merck KGaA, Germany.

Methods

Model description

To project long-term clinical and economic outcomes data for patients receiving first line cholesterol treatment (usually a statin), with or without an add-on intervention, an internet-based computer simulation model programmed in TreeAge Pro (TreeAge Software Inc., Williamstown, Massachusetts, USA) was developed. The model has a user-friendly HTML interface to allow widespread user access over the internet and allow cohort, treatment and economic settings to be readily adjusted, and simulations to be set up.

The model architecture consists of two analytic decision sub- models designed to work in combination. The first sub-model generates a baseline cohort for the analysis and assigns individual lipid profiles to each patient and then applies the treatment effects of the interventions under investigation. The second submodel uses standard Markov techniques and projects the long-term clinical and economic results of the different treatment strategies, based on risk equations from the Framingham study15.

Lipid profile sub-model

For each patient, an individual lipid profile is randomly drawn from these distributions using Monte Carlo simulation techniques19 (Figure 1).

With the baseline lipid profiles assigned to the patients, all patients are assumed to receive a one-off treatment effect with a first-line cholesterol-lowering treatment (assumed to be a statin, by default). Patients with lipid parameter levels outside user- defined limits (such as HDL-c below 1 mmol/L), despite the statin treatment effect, are randomly assigned to either continue with statin monotherapy or receive addon therapy with a hypothetical intervention aimed at correcting the remaining lipid abnormality. ‘Cutoff’ levels suchas these can be modified through the user interface. The treatment effects associated with the hypothetical add-on intervention are applied as a percentage change in HDL-c and/ or LDL-c and/or triglyceride levels and persist over patient lifetimes. The effects of the hypothetical add-on intervention are sampled from a normal distribution based on mean and standard deviation of effect change.

Table 1. Baseline cohort characteristics that are defined at the user interface

Long-term projection sub-model

As described above, the first sub-model generates lipid profiles for patients in two sub-groups: (1) those with persistently abnormal lipid level(s) after initial statin treatment who continue with statin monotherapy; and (2) those with persistently abnormal lipid level(s) who receive add-on treatment with a hypothetical intervention. After 1000 simulated patients have had a statin hypothetical intervention treatment effect applied to their lipid profiles, mean post-treatment levels of HDL-c, LDL-c, and triglycerides are calculated. The progress of these two groups of patients is then projected using the second sub-model.

The second sub-model comprises standard Markov model architecture and is designed to evaluate the long-term development of CHD complications, including angina, myocardial infarction (MI) and CHD death. It is based on five health states: no CHD, history of MI, history of angina, history of MI and angina, and dead (Figure 2). The cycle length is 1 year, and the time horizon can be set from 1 year to patient lifetimes, by which time all simulated patients have left the model following either CHD-related death or death from other causes. A proportional hazard regression model, constituting a risk function based on data published from The Framingham Heart Study, is used to calculate health state transition probabilities and risk adjustments. The regression model, published by D’Agostino et al., is used to calculate the probability of initial and subsequent coronary events, including angina, MI, and CHD death, allowing the model to use specific transition probabilities for primary and secondary prevention, according to the population studied15. Non CHD-related mortality is calculated using country-, age- and gender-specific life tables. Separate probabilities are calculated for male and female patients and the proportion of patients belonging to each state at baseline can be modified through the user interface. Risk factors in the Framingham formula include age, gender, menopausal status, total cholesterol, HDL-c, triglycerides, systolic blood pressure, antihypertensive medication, diabetes, smoking status, and alcohol consumption. Total cholesterol is calculated from mean values of HDL-c, and triglycerides generated by the first sub-model using the Friedwald equation, where total cholesterol (mmol/L) = HDL-c + LDL-c + (triglycerides/2.2). Initial and subsequent risk of CHD events in both males and females were observed by D’Agostino et al. to be influenced significantly by HDL- c level15.

Figure 1. Calculation algorithm of the first (lipid profile) sub- model. HDL-c = high density lipoprotein cholesterol; LDL-c = low density lipoprotein cholesterol; TG = triglyceride

Figure 2. Markov structure of the second (long-term projection) sub-model. *HI = Hypothetical intervention; CHD = coronary heart disease; MI = myocardial infarction; Hx = history

In the current model, CHD mortality rates are influenced primarily by changes in lipid levels. To account for the total annual risk of mortality, the annual probability of non-CHD death is taken from age- and gender-dependent life tables. In this way, by simulating lipid treatment over a lifetime horizon and accounting for CHD-related and non-CHD death, the model is able to project mean life expectancy for simulated patients. With every yearly cycle, the risks of CHD events for simulated patients are recalculated using the new patient ages in the Framingham formula. The 4-year risks of CHD events provided by the Framingham formula are converted to annual probabilities.

Costs

Pharmacy costs for anti-lipid therapy, and direct costs for the treatment of CHD-related health states represented in the model are accounted. Baseline add-on pharmacy costs are applied over patient lifetimes. Analyses are performed from a third-party healthcare payer perspective, with incremental cost-effectiveness ratios expressed in terms of cost per life year gained and cost per quality adjusted life year gained. Health state utilities are derived from published sources20,21. Costs for angina and MI are split into first year event costs and follow-up costs for all subsequent years. Pharmacy and CHD complication costs, along with annual discount rates for clinical and cost outcomes, and economic settings, can be modified at the user interface to adapt the model to various country- specific settings.

Uncertainty in the model

The long-term projection sub-model does not calculate confidence intervals for outcomes as it is deterministic in nature and does not utilize probabilistic simulation. However, in keeping with ISPOR guidelines that recommend the use of either deterministic or probabilistic simulation in health economic analyses, the model handles uncertainty by utilizing second order Monte Carlo simulation to sample baseline patient lipid profiles and lipid treatment effects from probability distributions14. In addition, sensitivity analyses are performed on model input parameters to test their impact on the model’s projections and assess the robustness of the model’s results. Sensitivity analyses may, for example, be performed on lipid ‘cut off levels, drug efficacy, the gender distribution of the population, the cost of CHD complications and discount rates on clinical and cost outcomes.

In summary, the model can project long-term clinical and cost outcomes for lipid modifying interventions used in user-defined populations for either primary or secondary prevention, based on risk factors included in the Framingham formulae, in a variety of settings.

Preliminary comparative clinical outcome modeling analyses

As the long-term projection sub-model is based on the Framingham equations and incorporates non-CHD causes of death to evaluate life expectancy, we performed an analysis against published data by simulating the progress of four cohorts with the same baseline characteristics as reported in the Framingham study15. For males with no history of cardiovascular disease, the model projected a risk of 1.12% for any CHD events after 2 years, compared to a value of 1% published by D’Agostino et al.15 In the female cohort with no history of cardiovascular disease, the model projected a CHD-event risk of 2.02% compared to the published value of 2%. In patients with a history of cardiovascular disease, the model predicted 2- year CHD event risks of 9.15% and 3.20% for males and females, respectively. The published values were 10% and 3.5% in the male and female groups, respectively15.

Further comparative clinical analyses were made against published epidemiological or clinical studies whose outcomes data were not used to calculate transition probabilities in the model. Trials were chosen based on the appropriateness of the outcomes data reported, and they were required to be of high quality and present suitable baseline cohort data.

In a meta-analysis of four landmark studies investigating the risk of CV disease associated with low HDL-c levels, Gordon et al. reported that each 1 mg/dL increment in HDL-c was associated with a 2-3% decrease in the risk of CHD events18. By recreating the population from the meta-analysis and simulating its progress over 10 years, with a range of HDL-c changes, the model projected a risk reduction of 2.32% for CHD events associated with each 1 mg/dL increase in HDL-c, falling within the range reported in the metaanalysis.

A second analysis was performed in the secondary prevention setting against data from the 4 S on the cumulative incidence of non- fatal and fatal CHD events16,22. The 4S was a randomized, double- blind, placebo-controlled trial that demonstrated statin treatment (simvastatin 20-40 mg daily) reduced the risk of death by 30% (p = 0.0003) over a median followup of 5.4 years in patients with previous myocardial infarction or stable angina pectoris22. By recreating the same cohort characteristics and simvastatin treatment effects in terms of lipid profile reported in the 4S study, the model projected a cumulative incidence of 14.2% for non-fatal and 5.5% for fatal CHD events over the duration of the trial; the 4S study reported 15.9% and 5.0%, respectively16.

A further comparative analysis was performed against data from AFCAPS/TexCAPS17. This was a primary prevention clinical trial with a primary composite endpoint encompassing first acute major coronary events (fatal or non-fatal myocardial infarction, unstable angina, or sudden cardiac death) and secondary endpoints including unstable angina, combined fatal and non-fatal myocardial infarction, and revascularization. The model’s projection of the 5-year cumulative incidence of angina for the AFCAPS/ TexCAPS population was compared to the cumulative incidence of this endpoint reported in AFCAPS/ TexCAPS (the primary endpoint and remaining secondary endpoints of AFCAPS/TexCAPS were not considered representative of outcomes projected by the model). A linear regression model (y = x) was used to plot the model’s projections and the AFCAPS/TexCAPS values and showed a close correlation with an R^sup 2^ value of 0.9851 (perfect fit = 1) (Figure 3).

Figure 3. Correlation plot comparing cumulative incidence of angina calculated by the CHD model with results from AFCAPS/ TexCAPS17

The results of the modeling analyses above were plotted against corresponding real-life values from the published studies (including primary and secondary endpoint data from AFCAPS/TexCAPS) to as\sess goodness of fit (Figure 4). The majority of values on the chart are shown to be on or close to the y = x line, which indicates a perfect fit. The R^sup 2^ value of 0.9575 indicated a very close correlation. Furthermore, the gradient of the regression line in Figure 4 was 0.9329, close to the perfect fit line (gradient =1).

Discussion

The main aim of this paper was to describe a new computer simulation model designed to evaluate the long-term clinical and cost outcomes associated with first-line and add-on lipid interventions (to correct persistent dyslipidemia), including HDL-c raising therapy. Preliminary comparative analyses with the model against data used to create it and against three published studies that were not used in development, indicated that the model can closely reproduce real life clinical trial data.

Figure 4. Correlation plot comparing results calculated by the CHD model with published results from studies (Framingham15, Gordon et al.18, 4S16 andAFCAPS/TexCAPS17). CHD = coronary heart disease

Advantages of the model include a user-friendly HTML interface that allows widespread access over the internet and gives the user the ability to readily adjust cohort, treatment effect and economic setting data. Potential, limiting assumptions in the model are that treatment for dyslipidemia is considered to be lifelong, and that the treatment dose and treatment effects of dyslipidemia medication are assumed to remain constant over patient lifetimes. A further limitation of the model is the use of the Friedwald equation to estimate total serum cholesterol. Whilst this is frequently used in Cardiology, it was based on a relatively small patient population and should not be used in cases where serum triglylcerides concentration exceeds 4.52 mmol/L.

A potential criticism of the model is that it makes long-term projections of CHD events based on the Framingham equations and does not account for other macrovascular disease, such as cerebrovascular and peripheral vascular disease. Whilst lipid lowering treatment can influence the onset and sequelae of all these conditions, the model focuses on CHD outcomes because of the overwhelming clinical and economic impact this disease has. The Framingham equations were generated from a study of predominantly white, middle-class patients in North America and, therefore, may not be applicable to other countries. Framingham, however, represents a large source of data on CHD risk and is widely used and well accepted. Moreover, validation of the Framingham CHD prediction scores against multiple ethnic groups has shown that the functions performed well for White and Black patients, and could be adapted to accurately predict CHD events in Hispanic, Japanese-American and Native-American populations23.

Given the important potential benefits of raising HDL-c, it is noteworthy that a Pan-European study of 8545 patients highlighted a high prevalence of low HDL-c in patients diagnosed with dyslipidemia, irrespective of statin treatment24. The prevalence of low HDL-c was 33% in men (HDL-c < 1.03 mmol/L) and 40% in women (HDL- c < 1.29mmol/L). Very low HDL-c (HDL-C < 0.90 mmol/L) occurred in 14% of treated patients, with a similar prevalence in untreated patients. This indicates that insufficient attention is paid to treating this major CHD risk factor.

Conclusions

We developed a computer simulation model to evaluate the long- term clinical and cost outcomes for patients with persistent lipid abnormalities despite first-line treatment. Validation analyses suggest that the model is able to recreate the outcomes from published clinical studies. It is hoped that future analyses of real- life interventions with this model will prove useful in terms of supporting clinical and reimbursement decision makers evaluating new and existing therapy options, including add-on HDL-c raising therapies, for patients with persistent dyslipidemia on statin treatment.

Acknowledgments

Declaration of interest: The health economic model described in this paper was developed by CORE – Center for Outcomes Research, a unit of IMS, and the preparation of this manuscript was supported by an unrestricted grant from Merck KGaA, Darmstadt, Germany. Merck KGaA is the manufacturer of sustained-release nicotinic acid as an add-on therapy in the treatment of dyslipidemia with persistent low HDL-c levels.

References

1. Thorn T, Haase N, Rosamond W, et al. Heart disease and stroke statistics – 2006 update: a report from the American Heart Association Statistics Committee and Stroke Statistics Subcommittee. Circulation 2006;113:e85-151

2. European cardiovascular disease statistics 2005. European Heart Network . Available at http://www.ehnheart.org/files/ statistics%202005-092711A.pdf [last accessed 20 July 2006]

3. British Heart Foundation. BHF 2003 CHD statistics summary. Available at http://www.bhf.org.uk/professionals/uploaded/ bhf%20heartstats%202003%20-%204-page.pdf [last accessed 20 July 2006]

4. Adult Treatment Panel III. Executive summary of the third report of The National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults. J Am Med Assoc 2001;285: 2486-97

5. de Backer G, Ambrosioni E, Borch-Johnsen K, et al. European guidelines on cardiovascular disease prevention in clinical practice [Third Joint Task Force of European and Other Societies on Cardiovascular Disease Prevention in Clinical Practice]. Eur Heart J 2003;24:1601-10

6. Baigent C, Keech A, Kearney PM, et al. Efficacy and safety of cholesterol-lowering treatment: prospective meta-analysis of data from 90056 participants in 14 randomised trials of statins. Lancet 2005;366:1267-78

7. Castelli WP, Garrison RJ, Wilson PW, et al. Incidence of coronary heart disease and lipoprotein cholesterol levels. The Framingham Study. J Am Med Assoc 1986;256:2835-8

8. Turner RC, Millns H, Neil HA, et al. Risk factors for coronary artery disease in non-insulin dependent diabetes mellitus: United Kingdom Prospective Diabetes Study (UKPDS: 23). Br Med J 1998;316:823-8

9. Yusuf S, Hawken S, Ounpuu S, et al. Effect of potentially modifiable risk factors associated with myocardial infarction in 52 countries (the INTERHEART study): case-control study. Lancet 2004;364:937-52

10. Grundy SM, Cleeman JI, Merz CN, et al. Implications of recent clinical trials for the National Cholesterol Education Program Adult Treatment Panel III guidelines. Circulation 2004;110:227-39

11. International Atherosclerosis Society. Harmonised clinical guidelines on prevention of atherosclerosic vascular disease. 2003. Available at http://www.athero.org [last accessed 20 July 1996]

12. Agence Franaise de Scurit Sanitaire des Produits de Sant (AFSSAPS). Prise en charge therapeutique du patient dyslipidemique. 2005. Available at http://afssaps.sante.fr/pdf/5/ rbp/dysreco.pdf [last accessed 20 July 1996]

13. Taylor AJ, Sullenberger LE, Lee HJ, et al. Arterial Biology for the Investigation of the Treatment Effects of Reducing Cholesterol (ARBITER) 2: a double-blind, placebo-controlled study of extended-release niacin on atherosclerosis progression in secondary prevention patients treated with statins. Circulation 2004;110:3512- 7

14. Weinstein MC, O’Brien B, Hornberger J, et al. Principles of good practice for decision analytic modeling in health-care evaluation: report of the ISPOR Task Force on Good Research Practices – Modeling Studies. Value Health 2003;6:9-17

15. D’Agostino RB, Russell MW, Huse DM, et al. Primary and subsequent coronary risk appraisal: new results from the Framingham study. Am Heart J 2000;139:272-81

16. Pedersen TR, Olsson AG, Faergeman O, et al. Lipoprotein changes and reduction in the incidence of major coronary heart disease events in the Scandinavian Simvastatin Survival Study (4S). 1998. Atheroscler Suppl 2004;5:99-106

17. Downs JR, Clearfield M, Weis S, et al. Primary prevention of acute coronary events with lovastatin in men and women with average cholesterol levels: results of AFCAPS/TexCAPS. Air Force/Texas Coronary Atherosclerosis Prevention Study. J Am Med Assoc 1998;279:1615-22

18. Gordon DJ, Probstfield JL, Garrison RJ, et al. High-density lipoprotein cholesterol and cardiovascular disease. Four prospective American studies. Circulation 1989;79:8-15

19. Briggs AH. Handling uncertainty in cost-effectiveness models. Pharmacoeconomics 2000; 17:479-500

20. Palmer S. A cost-effectiveness model comparing alternative management strategies for the use of glycoprotein IIb/IIIa antagonists in non-st-elevation acute coronary syndrome. Report to the National Institute for Clinical Excellence 2002

21. Main C. A rapid and systematic review of the clinical effectiveness and cost-effectiveness of clopidogrel used in combination with aspirin compared to aspirin alone in the treatment of non-st-segment elevation acute coronary syndrome (ACS). National Institute for Clinical Excellence 2004

22. Anon. Randomised trial of cholesterol lowering in 4444 patients with coronary heart disease: the Scandinavian Simvastatin Survival Study (4S). Lancet 1994;344:1383-9

23. D’Agostino Sr RB, Grundy S, Sullivan LM, Wilson P. Validation of the Framingham coronary heart disease prediction scores: results of a multiple ethnic groups investigation. J Am Med Assoc 2001;286:180-7

24. Bruckert E, Baccara-Dinet M, McCoy F, Chapman J. High prevalence of low HDL-cholesterol in a pan-European survey of 8545 dyslipidaemic patients. Curr Med Res Opin 2005;21: 1927-34

CrossRef links are available in the online published version of this paper: http://www.cmrojournal.com

Paper CMRO-3603_3, Accepted for publication: 13 October 2006

Published Online: 21 November 2006

doi: 10.1185/030079906X148490

S. Roze(a), D. Liens(b), A. Palmer(a), W. Berger(c), D. Tucker(a) and C. Renaudin(b)

a CORE – Center for Outcomes Research, A Unit of IMS, Basel, Switzerland

b Merck Sant, Lyon, France

c Merck KGaA, Darmstadt,Germany

Address for correspondence: Stphane Roze, CORE – Center for Outcomes Research, A Unit of IMS, Gewerbestrasse 25, 4123 Basel, Switzerland. Tel.: +41 61 383 0757; Fax: +41 61 383 0759

Key words: Coronary heart disease – Cost-effectiveness – Dyslipidemia – HDL-c – Modeling – Statin

Copyright Librapharm Dec 2006

(c) 2006 Current Medical Research and Opinion. Provided by ProQuest Information and Learning. All rights Reserved.