Clinical and economic evaluation of exenatide for formulary decisions

Summary

This study was designed to provide an assessment for healthcare organisations to make formulary decisions on Byetta®* (exenatide).

The drug was shown to be efficacious in treating type 2 diabetes mellitus when used as adjunctive therapy with other oral antidiabetic drugs. The Centre for Outcomes Research Diabetes Model analyses suggested that the drug treatment fails to achieve a threshold level of the incremental cost-effectiveness ratio until 50 years have passed. Sensitivity analyses suggested that maintaining an appropriate haemoglobin A_1c level is essential for the treatment to be cost effective. A budget impact analysis with a hypothetical plan produced a 2.1 cent base case compliance-adjusted, per member per month cost in Year 1. One-way sensitivity analyses indicated that the two major determinants are compliance and the percentage of individuals expected to be on exenatide.

The author's conclude that further research is needed to include exenatide for the formulary decision.

• managed care
• diabetes
• cost effectiveness
• exenatide
• formulary
• budget impact analysis

Introduction

Diabetes mellitus (DM) is a metabolic disease that is characterised by defects in insulin secretion, insulin action, or both. Both the prevalence rate for DM and the number of diabetic patients have increased steadily in the US since a national system for ascertaining diagnosed diabetes was established in 19581. In the year 2005 alone, 1.5 million new cases were diagnosed among people aged ≥20 years2. Over 20.8 million people are currently affected by the disease, representing approximately 7% of the US population3. The direct costs of treating diabetes and its complications account for approximately 12% of the total US healthcare expenditure4,5.

DM has been classified into four different types by the World Health Organization and the National Diabetes Data Group: type 1 (insulin-dependent DM); type 2 (non-insulin-dependent DM); gestational DM; and diabetes secondary to other conditions1. This study focuses only on type 2 DM, which is the most predominant form in the US, affecting 90–95% of all diagnosed diabetic patients2. Also referred to as adult-onset diabetes, type 2 DM results from a progressive β-cell dysfunction on the background of insulin resistance6. Many DM patients can control their blood glucose level by changing their lifestyle, losing weight and taking oral medications2.

Haemoglobin A_1c (HbA_1c) is commonly used to assess the blood glucose level of diabetic patients over the preceding 3 months7. The American Association of Clinical Endocrinologists recommends 6.5% (or lower) of the target HbA_1c level for a diabetic patient8.

If untreated or improperly maintained, DM causes numerous complications. The most common acute metabolic complication in type 2 DM is hyperosmolar hyperglycaemic coma. Chronic complications include retinopathy, nephropathy, neuropathy and peripheral vascular disease6.

Recently, Byetta®* (exenatide), a new non-insulin medication, was introduced to the market to help people with type 2 DM by allowing a delay in the initiation of insulin therapy and insulin dependence9. Since its introduction, many pharmacy and therapeutics committees at various healthcare plans have had to decide whether or not Byetta should be put on formulary, given the cost of the medication.

Byetta is an injectable form of exenatide available in two dosage strengths, 5 and 10 μg. The US Food and Drug Administration indication for Byetta is adjunctive therapy in type 2 DM patients who have taken metformin, a sulfonylurea, a thiazolidinedione (TZD), or a combination of metformin with either of the latter two medications, but still have not achieved adequate glycaemic control9. Exenatide enhances the natural ability of pancreatic β-cells to secrete insulin under a hyperglycaemic state, suppresses inappropriate glucagon secretion and slows gastric emptying9. In clinical trials, the most common side effects of the drug were nausea, vomiting and diarrhoea. When exenatide is used in combination with sulfonylurea, hypoglycaemia was the second most common side effects reported during the initial weeks of treatment10–12.

The purpose of this study was to analyse the clinical and economic data of Byetta and to determine the potential impact on adding it to the formulary of a hypothetical health plan. A review of the Centre for Outcomes Research (CORE) Diabetes Model analyses performed by the manufacturer is presented and an assessment of the cost effectiveness of Byetta is provided using an interactive budget impact model.

Methodology

The CORE Diabetes Model, which is widely reviewed and non-product specific, was used to evaluate the cost effectiveness of exenatide. The CORE Diabetes Model is based on a series of secondary (Markov) models and Monte Carlo simulations, with tracker variables, that capture the long-term and progressive nature of diabetes and its complications4. To minimise sample bias, non-parametric bootstrap procedures were used. These first- and second-order Monte Carlo simulation procedures provide an unbiased estimation of sample mean13. In addition, a step therapy protocol requiring metformin and/or sulfonylurea prior to exenatide was imposed on the simulation. The model compares long-term health outcomes of exenatide with those of placebo by changing hbA_1c value and weight loss, and compares their impact on morbidity and mortality. These variables are chosen based on several clinical studies that showed the effects of exenatide on patients with type 2 DM10–12. Details of the CORE Diabetes Model simulation input assumptions can be found in the managed care dossier of Byetta¹⁴.

A budget impact model analysis allows a health plan to obtain a predictable picture of the economic impact by adding exenatide to their formulary over a 5-year period. The model compares pharmacy costs, quantifying the cost of adding exenatide to a health plan against alternative antidiabetic medications. The alternative medications are a TZD, insulin glargine and insulin glargine plus a bolus dose of a rapid-acting insulin analogue. Total economic cost is calculated by multiplying incremental unit drug cost by the projected number of patients.

The characteristics of the health plan in this budget impact model analysis were: a total of 2.6 million individuals with an age distribution of 28.5% at ≤18 years, 61% at 18–65 years and 10.5% at ≥65 years. The gender distribution was 51.4% female and 48.6% male, with a prevalence of type 2 DM of 5.6% and drug treatment rates of members between the ages of 18 and 65 years of 80%. A price per day of therapy for exenatide was assumed to be $6.77 (for 10 μg). This figure is higher than the figure of $5.75 for 10 μg from the exenatide trial assumptions (average wholesale price of Byetta for 10 μg/day obtained from Drug Topics Red Book including prices of alternative drugs, November 2006). Furthermore, patient segments for exenatide were projected to be 1.6, 2.8, 4.0, 4.9 and 5.6% for Years 1 through to 5, respectively, and an 18% predicted uptake of exenatide14.

Results

The result of the CORE Diabetes Model analysis and that of the budget impact model analysis each provide a different aspect of the cost effectiveness of the drug treatments.

Incremental cost effectiveness

The results of the CORE Diabetes Model analyses showed that exenatide failed to reach a threshold level of an incremental cost-effectiveness ratio (ICER) of <$50,000 until 50 years have passed since the beginning of the treatment given that the body weight has been maintained or lost throughout the trial15. The ICER reaches a value of $43,814 with respect to life expectancy and a value of $48,921 with respect to quality-adjusted life years at 50 years of treatment14. Table 1 shows the result of the CORE Diabetes Model analyses. One notable point of the analyses is that there is very little change in ICER values from the 20- to the 50-year horizon. These figures offer an indication that the drug will not be cost effective because it takes at least 20 years to be marginally cost effective, and the incremental cost effectiveness of the treatment after 20 years becomes almost negligible.

Table 1. CORE Diabetes Model analysis (base case): Byetta vs. placebo*.

Time horizon	LE	LE QALE	Costs	ICER (LE)	ICER (QALY)
5 years	0.02	0.07	$6,986	$359,757	$104,697
20 years	0.24	0.26	$13,522	$56,244	$52,180
50 years	0.35	0.32	$15,477	$43,814	$48,921

LE, life expectancy; QALE, quality-adjusted life expectancy; ICER, incremental cost-effectiveness ratio; QALY, quality-adjusted life year.

* Values are mean per patient and discounted at 3% per year.

The high ICER in the earlier period of treatment may be explained by the fact that most of the type 2 DM complications have not yet occurred. According to a report by the Centres for Disease Control and Prevention, 86% of type 2 DM patients are over 40 years of age3. Therefore, it is expected that for the majority of individuals with type 2 DM, exenatide will not be incrementally cost effective.

A one-way sensitivity analysis was performed by varying the discount rate as well as the HbA_1c level (Figure 1). By adjusting the HbA_1c obtained from the pivotal exenatide trials by ± 20%, the ICER ranged from $41,547 to $58,215 at the 50-year time horizon. The mean change in HbA_1c level of the trials was 0.9 with standard errors of 0.1 after 30 weeks of treatment14. By varying the discount rate from 0 to 5%, ICER values changed from $44,297 to $52,512. The base discount rate was 3% per annum (US Department of Labour, 2003). These results underscore the need to achieve appropriate HbA_1c levels to maintain the cost effectiveness of exenatide.

Figure 1. One-way sensitivity analysis on ICER.

Budget impact of adding exenatide

Figure 2 summarises the findings of the budget impact analysis. In Years 1–5, a compliance-adjusted total cost was $655,032, $1,146,306, $1,637,580, $2,006,035 and $2,292,611, respectively. A study by Cramer16 reviewed adherence to DM medications and found overall type 2 DM adherence to be 62–64% of the prescribed dosage. These compliance-adjusted figures provide a more realistic estimate of the medication costs (corresponding 63% adherence rate used). The resulting per member per month (PMPM) cost for the base case in the first year was $0.021.

Figure 2. Compliance-adjusted total cost.

An acceptable PMPM cost varies between health plans as factors such as prevalence within the specific health plan and change of PMPM cost in the coming years influence the acceptance level. Given the variety of DM medications available on the market and the cost of exenatide compared with alternative drug treatments, the drug failed to provide an attractive option for a hypothetical health plan.

A one-way sensitivity analysis was performed on the budget impact model by changing the following variables: compliance; percentage of patients expected to be on exenatide; the price of exenatide; and the prevalence of type 2 DM. By varying compliance from 0 to 365 days, PMPM cost ranged from $0.00 to $0.029. The range of PMPM cost changed from $0.013 to $0.029 when the percentage of patients expected to be on exenatide varied from 11 to 25%. A variation in the price of exenatide from $6.09 to $6.45 changed the PMPM cost from $0.019 to $0.023. Finally, changing the prevalence of type 2 DM from 4.2 to 7%, PMPM cost in the model varied from $0.016 to $0.026. The results are reported in Figure 3. The sensitivity analysis results showed that the two biggest drivers of the model are compliance and the percentage of individuals expected to be on exenatide.

Figure 3. One-way sensitivity analysis on budget impact model.

Discussions

Whilst the CORE Diabetes Model has been tested and reviewed in many studies, the results in the real world may vary when applied with different assumptions regarding population characteristics and demographics and/or scenarios regarding patients’ compliance along the time horizons.

The baseline demographics and characteristics of the patient population used in the CORE Diabetes Model were identical to those in the Byetta pivotal trials. As clinical studies have specific inclusion and exclusion criteria that limit the application of the results, the results of the CORE Diabetes Model also have their own limitations for applications.

The base CORE Diabetes Model assumes a uniform compliance for all medications by the patient throughout the simulations, which is not a realistic scenario in practice. Compliance of Byetta may differ in practice from the scenario because the drug requires twice-daily injections and often causes side effects such as severe nausea and vomiting. Fear of injections and side effects are common factors that impede patient compliance17,18.

In addition, the CORE Diabetes Model assumes that the separation between Byetta and placebo is maintained throughout the 50-year treatment period. It is unlikely that this separation will be observed in the real world. Again, if any of the conditions assumed for this study do not match real-world conditions, the economic assessment on cost effectiveness of Byetta will have very little implication in practice.

Although the onset of type 2 DM has traditionally been observed in individuals over the age of 40 years, onset in those under 40 years is now observed, particularly in ethnic minority groups3,6. It should be noted that this change and its economic impacts are expected to be investigated further in future studies.

Moreover, most hyperglycaemic medications for type 2 diabetes eventually fail to control blood glucose levels after a certain time period of treatment. Data is not available either to prove or disprove this aspect of cost effectiveness of the antidiabetic drug.

Conclusions

The review of the CORE Diabetes Model and the budget impact model analysis indicated that exenatide treatment will not be incrementally cost effective for the majority of individuals with type 2 DM. The results of the sensitivity analyses of the CORE Diabetes Model suggested that maintaining appropriate HbA_1c levels is essential for exenatide treatment to be cost effective. One-way sensitivity analyses of the budget impact model demonstrated that the two major determinants of the model are compliance and the percentage of individuals expected to be on exenatide.

The authors suggest the following recommendations to assist healthcare organisations in determining whether to include Byetta on their formulary. These recommendations are based on a step therapy requirement of metformin and/or a sulfonylurea before adding Byetta to the drug formulary.

The requirements of prior authorisation after the fourth pharmacy claim should be to ensure that: (i) the HbA_1c level has been decreased from the pre-Byetta treatment level; (ii) body weight has also been maintained or lost from the weight before the treatment; and (iii) patients are able to tolerate the side effects of the drug.

These conclusions underscore the need for healthcare organisations to assess their own populations in order to make decisions well measured with respect to their disease management programmes.

Acknowledgements

Declaration of interest: The authors have declared no conflict of interest and have received no payment in the preparation of this manuscript.

The authors would like to acknowledge the support of Richard Fry, Director of Programs at the Foundation of Managed Care Pharmacy, for this study.

Notes