B. Decision tree for budget scenario with drug E
11.2 Reporting the Budget-Impact Analysis
11.2.5 Reporting the Scenario and Sensitivity Analyses
Finally, the results of the scenario analyses and one-way sensitivity analyses should be presented in tabular and/or graphical format. The following are examples of sce- nario analyses that should be presented:
• Alternative current patterns of drug use by the health plan
• Alternative projected rates of uptake for the new drug
Budget-impact results in 24 patients with locally advanced, nonresectable, or metastatic pancreatic cancer (Danese et al. 2008, Table 4)
Budget component Erlotinib + gemcitabinea Gemcitabine monotherapy
Treatment 358,900 250,800
Administration 44,400 40,300
Side effects 63,400 55,600
Total 466,700 346,700
Difference 120,000
Per member per month 0.020
Reprinted from Danese et al. 2008, Copyright 2008, with permission from Elsevier Note: Values are 2006 USA dollars
aResults assumed erlotinib was used in 40% of gemcitabine-treated patients in the erlo- tinib + gemcitabine treatment strategy
$0
$100,000
$300,000
$200,000
$400,000
$500,000
$50,000
$250,000
$150,000
$350,000
$450,000
Treatment Administration Adverse events Total Contributions to total cost of treatment (US $)
Erlotinib + gemcitabine Gemcitabine monotherapy
Budget impact of adding erlotinib to gemcitabine therapy for locally advanced, nonresect- able, or metastatic pancreatic cancer in the base-case analysis for a health plan with 500,000 members. Results assumed erlotinib was used in 40% of gemcitabine-treated patients in the erlotinib + gemcitabine treatment strategy (Reprinted from Danese et al. 2008, Copyright 2008, with permission from Elsevier)
Results for a new combination therapy for advanced pancreatic cancer (Danese et al.
2008, Fig. 3).
181
• Alternative redistribution patterns from the current drug to the new drug
• Alternative health plan population characteristics, including age, sex, and condi- tion incidence or prevalence rates
• Alternative restrictions on use planned by the health plan
• Alternative prices for the new drug
In Box 11.8, we present a hypothetical example of how to present the results of a scenario analysis when varying the estimated uptake rates for a new drug over the budget-impact analysis time horizon.
Box 11.8 Results of a Hypothetical Scenario Analysis Estimating the Budget Impact with Varying the Drug Uptake Rates
The budget-impact results are shown below for three different uptake scenar- ios for a new drug added to the treatment mix for a specific condition. In the health plan, we have assumed that 100 people are eligible for treatment, the total annual drug-related costs are $2000 (new drug), $1500 (drug A), $1000 (drug B), and $500 (drug C). We also assumed that the annual condition- related costs when taking drugs are $350 (new drug), $600 (drug A), $600 (drug B), and $1000 (drug C).
Results of scenario analyses for treatment mix with new drug Treatment shares
before new drug (current scenario)
Scenario 1 with new druga
Scenario 2 with new drugb
Scenario 3 with new drugb Treatment share
New drug 0% 10% 20% 10%
Drug A 20% 18% 20% 20%
Drug B 40% 36% 40% 40%
Drug C 40% 36% 20% 30%
Total drug-related costs $90,000 $101,000 $120,000 $105,000 Total condition-related
costs
$76,000 $71,900 $63,000 $69,500
Total drug- and condition-related costs
$166,000 $172,900 $183,000 $174,500
New drug scenarios vs. current scenario:
drug costs only
N/A $11,000 $30,000 $15,000
New drug scenarios vs. current scenario:
condition costs only
N/A −$4100 −$13,000 −$6500
New drug scenarios vs.
current scenario: drug and condition costs
N/A $6900 $17,000 $8500
N/A not applicable
aAssuming treatment shares for the new drug are taken equi-proportionately from all cur- rent drugs
bAssuming treatment shares for the new drugs are only taken from drug C 11 Reporting Budget-Impact Analyses
One-way sensitivity analyses can be presented in the form of tables or tornado diagrams both in interactive computer programs and in reports and publications.
Tornado diagrams present the results in an easy-to-interpret format. Tornado dia- grams should include the ranges tested in the analysis and indicate which bar repre- sents which end of the range. The following are examples of input parameter values for which one-way sensitivity analyses should be presented:
• Inputs with observed variability (e.g., standard deviation or standard error of the mean) in efficacy rates for current and new drugs based on formal meta-analyses
• Inputs with observed variability in rates of side effects for current and new drugs based on formal meta-analyses
• Inputs with observed variability in rates of discontinuation for current and new drugs based on formal meta-analyses
• Inputs with observed variability in drug-related or condition-related costs In Box 11.9, we present an example of a one-way sensitivity analysis presented in a tornado diagram.
Box 11.9 One-Way Sensitivity Analysis Results Presented as a Tornado Diagram for a New Combination Therapy for Advanced Pancreatic Cancer
Sensitivity of results expressed as either per-member per-month or total annual costs to assumed ranges of input parameter values (Danese et al. 2008, Fig. 4).
Administration cost Gemcitabine cost Duration of gemcitabine
Erlotinib cost
% Gemcitabine as chemotherapy Duration of erlotinib
Gemcitabine AE cost Erlotinib utilization
Erlotinib AE cost
% to receive chemotherapy Medicare population
0.005 PMPM
30,000 0.013 PMPM
75,000 0.020 PMPM
120,000 0.021 PMPM
165,000 Cost PMPM and total cost difference (US $)
20 Wk 12%
0%
88%
53%
100%
60%
32% 48%
−25% +25%
−25% +25%
−25% +25%
−25% +25%
−25% +25%
9 Wk 15 Wk
12 Wk
Sensitivity analyses: Total cost difference and cost per member per month (PMPM) between erlo- tinib + gemcitabine and gemcitabine monotherapy. AE adverse event. Values at the end of each bar indicate either the low and high values used or the change from the base-case. (Reprinted from Danese et al. 2008, Copyright 2008, with permission from Elsevier)
183