B. Decision tree for budget scenario with drug E
7.4 Computing Framework for Combined Cost-Effectiveness and Budget-Impact Model
The choice of computing framework might be influenced by the decision to use the same computer model to estimate the cost-effectiveness and budget impact of a new drug. Of course, the purposes of these two analyses are very different. Cost- effectiveness analysis is designed to estimate the value of a new drug compared with a standard-of-care comparator. The time horizon of a cost-effectiveness analysis is generally long enough to capture the full costs and benefits of the new drug, and costs and outcomes are typically discounted if they occur over multiple years. A budget-impact analysis is designed to provide information for budget planning within the planning time horizon comparing the treatment mix with and without the new drug included on the formulary. For budget-impact analyses, costs and out- comes are not discounted. Cost-effectiveness analyses may be reviewed by those with modeling expertise, while budget-impact analyses are intended for adaptation and use for budget planning by those who are not expert modelers. Therefore, the complexity appropriate for a cost-effectiveness analysis may not be appropriate for a budget-impact analysis.
If a static approach will suffice for the budget-impact analysis, the preferred approach may be to keep the budget-impact analysis separate from the cost- effectiveness analysis while ensuring that consistent input parameter values are used in both analyses. However, if a dynamic approach is necessary for the budget- impact analysis, it may be convenient to leverage the calculations in the cost- effectiveness analysis. These calculations may include disease progression, treatment switching, titration, discontinuation, mortality, or other health outcomes calculations.
If a combined model is desired, the model will need to be programmed with two different modes:
• Single cohort mode: This mode is appropriate for the cost-effectiveness analy- sis. In this mode, the model tracks the costs and health outcomes of a single initial cohort over a fixed time horizon, often over the remaining lifetime of the cohort. Results are typically presented as discounted totals over the full time horizon.
• Open population mode: This mode is appropriate for the budget-impact analysis.
In this mode, an initial prevalent cohort enters the model in the first year and is tracked for the duration of the budget-impact analysis time horizon. In each year, a newly incident cohort enters the model and is also tracked until the end of the budget-impact analysis time horizon. Each cohort is typically tracked separately in the model. Results are calculated as annual undiscounted totals across all the cohorts included in each year of the budget-impact analysis time horizon.
7 The Computing Framework and Calculations
Programming a combined model with both of these modes may be efficient, but the model calculations will be more complex than if the cost-effectiveness and budget- impact analyses are kept separate. It is important to consider whether the efficiency gained is worth the additional complexity.
In general, when developing the computing framework for a budget-impact analysis, it is critical to maintain focus on the target user. Since the primary purpose of a budget-impact analysis is to help health care budget holders plan their budgets over their typical planning time horizons, it is critical that the computing frame- work be kept as simple as possible while including enough detail about the condition and treatment process to ensure credibility. Simplicity helps ensure that the analy- sis is transparent and can be readily adapted to the environment specific to each budget holder.
Exercises
Exercise 7.1 Explain when it is better to use a static computing framework rather than a dynamic computing framework. What are the advantages of using a dynamic computing framework?
Exercise 7.2 Identify a condition for which a static computing framework would be appropriate. Outline the components of the computing approach that need to be considered when developing the underlying calculations for this model. Why is the static approach appropriate for this condition?
Exercise 7.3 Identify a condition for which a dynamic computing framework would be appropriate. Outline the components of the computing approach that need to be considered when developing the underlying calculations for this model. Why is the dynamic approach appropriate for this condition?
Exercise 7.4 Describe the differences in the computing frameworks that result from using a static versus a dynamic approach for the conditions identified in Exercises 7.2 and 7.3.
Exercise 7.5 For the condition that requires a dynamic approach in Exercise 7.3, how could the computing framework be revised to transform it into a static approach? How might the results be affected if a static approach is used instead of a dynamic approach?
Exercise 7.6 A typical assumption within a dynamic computing framework is that the treatment mix for a given budget year is the same for each of the cohorts that has entered the model. How could you account for different cohorts having different treatment mixes? Explain or show an example.
Exercise 7.7 Drug C is getting ready to go to market to treat condition X. One hundred people are treated for condition X every year. Patients are cured after 1 year of treatment. Condition X is currently treated with drugs A and B. These exist- ing drugs and drug C must be administered by a physician, and one administration is required for each prescription. It is common for patients with condition X to seek additional physician care while being treated. Drug C has been shown to reduce the
127
number of additional physician visits patients sought while being treated. Using the table below, calculate the budget and health outcomes impact of introducing drug C to the market for three budget years. What computing framework did you use and why? What key values messages should the manufacturer of drug C project?
Drug
Treatment mix without drug C Treatment mix with drug C Year 1 (%) Year 2 (%) Year 3 (%) Year 1 (%) Year 2 (%) Year 3 (%)
Drug A 80 75 70 75 70 65
Drug B 20 25 30 20 23 25
Drug C 0 0 0 5 7 10
Drug
Drug cost per prescription
Number of administrations per year
Cost per administration
Number of physician visits per year
Cost per physician visit
Drug A $100 15 $50 20 $300
Drug B $150 12 $50 18 $300
Drug C $200 12 $50 10 $300
Exercise 7.8 Condition Y is similar to condition X in Exercise 7.7 except that condition Y is treated for the remainder of the patient’s lifetime. In the first year, 100 people are treated for condition Y. Each subsequent year, five new patients begin treatment for condition Y. Annual mortality if treated with drug A or B is 5%. Drug C reduces annual mortality to 1%. As with condition X, these drugs must be adminis- tered by a physician, and one administration is required for each prescription. It is common for patients with condition Y to seek additional physician care while being treated. Drug C has been shown to reduce the number of additional physician visits sought while being treated. Using the tables below, estimate the budget and health outcomes’ impact of introducing drug C to the market for three budget years. What computing framework did you use and why? What key values messages should the manufacturer of drug C project?
Drug
Treatment mix without drug C Treatment mix with drug C Year 1 (%) Year 2 (%) Year 3 (%) Year 1 (%) Year 2 (%) Year 3 (%)
Drug A 80 75 70 75 70 65
Drug B 20 25 30 20 23 25
Drug C 0 0 0 5 7 10
Drug
Annual mortality (%)
Drug cost per prescription
Number of administrations per year
Cost per administration
Number of physician visits per year
Cost per physician visit
Drug A 5 $100 15 $50 20 $300
Drug B 5 $150 12 $50 18 $300
Drug C 1 $200 12 $50 10 $300
7 The Computing Framework and Calculations
Exercise 7.9 Exercises 7.7 and 7.8 are very similar. However, there are differ- ences in how the population changes over time. For each budget year, compare the costs, health outcomes, and budget impact between the exercises. What differences do you observe? Are the differences significant? Why or why not? How did the dynamic approach affect the results?
Exercise 7.10 What are the advantages and disadvantages of combining a cost- effectiveness analysis and a budget-impact analysis into a single Excel-based model? Discuss the implications of the choice of computing framework when com- bining these analyses into a single model.
References
Brogan A, Talbird S, Thompson J, Kim Y, Olson J, Peterson J, et al. Budget impact of the introduc- tion of elvitegravir/cobicistat/emtricitabine/tenofovir, the first integrase inhibitor-based single- tablet antiretroviral regimen for HIV treatment, to US third-party payers. Poster presented at the 16th Annual European Congress of the International Society for Pharmacoeconomics and Outcomes Research. Dublin, Ireland. 2013
National Institute for Health and Care Excellence (NICE). Guide to the methods of technology appraisal 2013. 2013. http://nice.org.uk/process/pmg9. Accessed 5 Aug 2016.
Sax PE, Meyers JL, Mugavero M, Davis KL. Adherence to antiretroviral treatment and correlation with risk of hospitalization among commercially insured HIV patients in the United States.
PLoS One. 2012;7(2):e31591.
Sullivan SD, Mauskopf JA, Augustovski F, Caro JJ, Lee KM, Minchin M, et al. Budget impact analysis—principles of good practice: report of the ISPOR 2012 Budget Impact Analysis Good Practice II Task Force. Value Health. 2014;17(1):5–14. doi:10.1016/j.jval.2013.08.2291.
129
© Springer International Publishing AG 2017
J. Mauskopf et al., Budget-Impact Analysis of Health Care Interventions, DOI 10.1007/978-3-319-50482-7_8
Chapter 8
Uncertainty Analysis
Josephine Mauskopf and Stephanie Earnshaw
Abstract The purpose of budget-impact analysis is to project the potential future impact of the introduction of a new drug or other intervention on payer or provider budgets. Because estimates of current input values as well as assumptions about the structural model elements and changes in many input values over the analysis time horizon are needed, the results are estimated with uncertainty. Therefore, it is impor- tant for the model to include a method for performing uncertainty analyses.
Uncertainty analyses allow the user to test the impact of different structural ele- ments, assumptions, and input parameter values on the outcomes of the budget- impact analysis. In this chapter, methods for testing the impact on the results are presented (1) for alternative scenarios created using data and assumptions known to the budget holder and (2) for estimated ranges of input parameter values using uncertain data estimates and assumptions.
Keywords Budget-impact analysis • Uncertainty • Sensitivity analysis • Scenario analysis • One-way sensitivity analysis • Probabilistic sensitivity analysis
Chapter Goal
To describe the sources of uncertainty within budget-impact analyses for new drugs and to provide recommendations for demonstrating the impact of this uncertainty on the analysis results.
J. Mauskopf (*) • S. Earnshaw
RTI Health Solutions, RTI International, Research Triangle Park, Durham, NC, USA
e-mail: jmauskopf@rti.org
The results of any budget-impact analysis contain inherent uncertainty. The pur- pose of budget-impact analysis is to project the potential impact on payer or pro- vider budgets of the introduction of a new drug.1 Because estimates of current input values, assumptions, and predictions about the future form the basis for the analysis, the results are estimated with uncertainty. Therefore, it is important for the analysis to include a method for performing both scenario and sensitivity analyses. These analyses allow the user to test the impact of different structural elements, assump- tions, and input parameter values on the outcomes of the budget-impact analysis.