Budget Impact Model Analysis

Author

Richmond Silvanus Baye

Introduction

Budget Impact Analyses (BIAs) estimate how a healthcare budget holder’s spending might change if they decide to cover a new medical intervention or modify an existing policy. These analyses assess financial effects at a population level, helping decision-makers understand the overall cost implications (Von Hein, 2025). In the pharmaceutical industry, a Budget Impact Analysis (BIA) helps evaluate the financial consequences of introducing a new drug. For example, if a manufacturer (e.g., Johnson and Johnson or Merck ) launches a higher-cost cancer therapy, the BIA may seek to answer the following questions.

Cost of the New Treatment: What’s the price of the new drug or technology?
How Many People Will Use It?: How many patients are expected to need it?
What Does It Replace?: Will it substitute an existing treatment (saving money) or be an add-on (increasing costs)?
Total Budget Impact: Will the overall spending go up, down, or stay the same?

With these guiding questions in mind, we will implement a BIA in R following the parameters and key assumptions in Mirko Von Hein’s Video. Link to the youtube

Let’s begin by loading the required packages. I like to use pacman because it does not require me to call the libraries all over again after installing the packages.

Code

pacman::p_load(tidyverse, dplyr, ggplot2, rmarkdown, purrr)

In this walk through tutorial, our aim it to assess the budget impact of introducing Drug A into the market against an existing Standard of Care (SoC). To model this scenario, one can assume a situation where Drug A is available against a situation without Drug A. The goal here is to estimate the budget impact at the national level over a period of 10 years.

Lets begin by closely following the cost inputs from Mirko’s model. We know there are two main costs: 1. Treatment Related Cost 2 Conditions Related Costs.

We assume the drug acquisition cost for A and SoC is 500 and 300 respectively. We also assume that treatment(SoC) is oral so there is not admin cost but for drug A there is an admin cost of 200 for administering injection. We also assume there is a monitoring cost for 100 and 50 for both drug A and SoC. The share of patient with adverse event due to drug A is 5% while the share of patients with rare side effects for SoC is 2%

Code

cost_components <- data.frame(
  Component = c("Drug", "Admin", "Monitor", "AE", "Hospitalization", "Outpatient"),
  Cost_A = c(500, 200, 100, 5, 0, 175),
  Cost_SoC = c(300, 0, 50, 60, 75, 175)
)

total_cost <- data.frame(
  Treatment = c("Drug A", "SoC"),
  Total_Cost = c(colSums(cost_components[, c("Cost_A", "Cost_SoC")]))
)

print(total_cost)

         Treatment Total_Cost
Cost_A      Drug A        980
Cost_SoC       SoC        660

From this, we can compute the total cost per year per patient for Drug A and SoC respectively as $960 and $660.

We also define the epidemiological landscape (Incidence and Prevalence): The U.S. population of 335 million serves as our baseline. Current cancer prevalence is approximately 6 million patients (1.8% of the population), with about 1.96 million new cases diagnosed annually. Focusing on treatment-eligible populations, we estimate 1.51 million patients are in active treatment, 904,500 receive systemic therapy, and 361,800 require second- or third-line treatments. These figures, drawn from SEER and American Cancer Society data

Epidemiological Imput Data

(Based on SEER, CDC, and ACS data)

Cascade Step	Assumption	Population Size	Notes
Total US Population	-	335,000,000	(2025 estimate)
Cancer Prevalence	1.8%	6,030,000	~18 million survivors (5.4% lifetime)
Active Treatment Cases	25%	1,507,500	Patients currently in therapy
On Systemic Therapy	60%	904,500	Chemo/targeted/immunotherapy
On 2nd/3rd Line Therapy	40%	361,800	Progressive/refractory disease

Code

# Cancer Cascade
cancer_cascade <- data.frame(
  Step = c("Total US", "Prevalence", "Active Treatment", "Systemic Therapy", "2nd/3rd Line"),
  Assumption = c(NA, "1.8%", "25%", "60%", "40%"),
  Population = c(335000000, 6030000, 1507500, 904500, 361800)
)

# Eligible Patients Over 5 Years
annual_incident <- 150000
years <- 0:5
eligible_patients <- 361800 + (annual_incident * years)

# Print tables
print(cancer_cascade)

              Step Assumption Population
1         Total US       <NA>  335000000
2       Prevalence       1.8%    6030000
3 Active Treatment        25%    1507500
4 Systemic Therapy        60%     904500
5     2nd/3rd Line        40%     361800

Code

print(data.frame(Year = years, Eligible_Patients = eligible_patients))

  Year Eligible_Patients
1    0            361800
2    1            511800
3    2            661800
4    3            811800
5    4            961800
6    5           1111800

Now that we have determined the eligible patient population, we can now build the market share for a situation with Drug A as against a situation without Drug A

Code

No_New_Drug <- data.frame(
    Year = c("Year 0", "Year 1", "Year 2", "Year 3", "Year 4", "Year 5"),
    Drug_A = c(0, 0, 0, 0, 0, 0),
    SoC = c(1.0, 1.0, 1.0, 1.0, 1.0, 1.0)  # 100% as 1.0
)

With_New_Drug <- data.frame(
    Year = c("Year 0", "Year 1", "Year 2", "Year 3", "Year 4", "Year 5"),
    Drug_A = c(0, 10, 25, 40, 60, 80),  # Market penetration %
    SoC = c(1.0, 0.9, 0.8, 0.75, 0.6, 0.2)  # 90% as 0.9, etc.
)
print(No_New_Drug)

    Year Drug_A SoC
1 Year 0      0   1
2 Year 1      0   1
3 Year 2      0   1
4 Year 3      0   1
5 Year 4      0   1
6 Year 5      0   1

Code

print(With_New_Drug)

    Year Drug_A  SoC
1 Year 0      0 1.00
2 Year 1     10 0.90
3 Year 2     25 0.80
4 Year 3     40 0.75
5 Year 4     60 0.60
6 Year 5     80 0.20

The assumption is with the introduction of the new treatment into the market, drug A will begin to take some of the market from SoC over time.

Cost per patient

Code

unit_cost_A <- sum(cost_components$Cost_A)
unit_cost_SoC <- sum(cost_components$Cost_SoC)

# Step 4: Loop over years to calculate total cost
total_costs <- data.frame(
  Year = paste0("Year ", 0:5),
  Eligible = eligible_patients,
  DrugA_Share = With_New_Drug$Drug_A,
  SoC_Share = With_New_Drug$SoC
)

total_costs$Cost_DrugA <- total_costs$Eligible * total_costs$DrugA_Share * unit_cost_A
total_costs$Cost_SoC   <- total_costs$Eligible * total_costs$SoC_Share * unit_cost_SoC
total_costs$Total_Cost <- total_costs$Cost_DrugA + total_costs$Cost_SoC

# View result
print(total_costs[, c("Year", "Eligible", "Cost_DrugA", "Cost_SoC", "Total_Cost")])

    Year Eligible  Cost_DrugA  Cost_SoC  Total_Cost
1 Year 0   361800           0 238788000   238788000
2 Year 1   511800  5015640000 304009200  5319649200
3 Year 2   661800 16214100000 349430400 16563530400
4 Year 3   811800 31822560000 401841000 32224401000
5 Year 4   961800 56553840000 380872800 56934712800
6 Year 5  1111800 87165120000 146757600 87311877600

Similarly, we can create one for a world with the new drug

Code

# Step 0: Define components and cost structure
cost_components <- data.frame(
  Component = c("Drug", "Admin", "Monitor", "AE", "Hospitalization", "Outpatient"),
  Cost_A = c(500, 200, 100, 5, 0, 175),
  Cost_SoC = c(300, 0, 50, 60, 75, 175)
)

unit_cost_A <- sum(cost_components$Cost_A)
unit_cost_SoC <- sum(cost_components$Cost_SoC)

# Step 1: Define population and market shares
years <- 0:5
eligible_patients <- 361800 + 150000 * years

With_New_Drug <- data.frame(
  Year = paste0("Year ", years),
  Drug_A = c(0, 0.10, 0.25, 0.40, 0.60, 0.80),
  SoC = c(1.0, 0.90, 0.75, 0.60, 0.40, 0.20)
)

# Step 2: Calculate patient splits
drugA_patients <- eligible_patients * With_New_Drug$Drug_A
soc_patients <- eligible_patients * With_New_Drug$SoC

# Step 3: Compute cost per component per year
calculate_component_costs <- function(patients, cost_vector) {
  sapply(1:nrow(cost_components), function(i) {
    patients * cost_vector[i]
  })
}

component_names <- cost_components$Component

cost_A_matrix <- t(mapply(calculate_component_costs, patients = drugA_patients, 
                          MoreArgs = list(cost_vector = cost_components$Cost_A)))
cost_SoC_matrix <- t(mapply(calculate_component_costs, patients = soc_patients, 
                            MoreArgs = list(cost_vector = cost_components$Cost_SoC)))

# Step 4: Sum across components
total_cost_A <- rowSums(cost_A_matrix)
total_cost_SoC <- rowSums(cost_SoC_matrix)
total_cost_combined <- total_cost_A + total_cost_SoC

# Step 5: Assemble final table
final_costs <- data.frame(
  Year = paste0("Year ", years),
  Eligible_Patients = eligible_patients,
  DrugA_Patients = drugA_patients,
  SoC_Patients = soc_patients,
  Cost_DrugA = round(total_cost_A),
  Cost_SoC = round(total_cost_SoC),
  Total_Cost = round(total_cost_combined)
)

print(final_costs)

    Year Eligible_Patients DrugA_Patients SoC_Patients Cost_DrugA  Cost_SoC
1 Year 0            361800              0       361800          0 238788000
2 Year 1            511800          51180       460620   50156400 304009200
3 Year 2            661800         165450       496350  162141000 327591000
4 Year 3            811800         324720       487080  318225600 321472800
5 Year 4            961800         577080       384720  565538400 253915200
6 Year 5           1111800         889440       222360  871651200 146757600
  Total_Cost
1  238788000
2  354165600
3  489732000
4  639698400
5  819453600
6 1018408800

Code

# Reuse earlier values
total_eligible <- eligible_patients  # 6 values: Year 0 to Year 5
total_cost_WITH <- final_costs$Total_Cost  # From With_New_Drug block

# World WITHOUT: all patients on SoC
unit_cost_SoC <- sum(cost_components$Cost_SoC)
total_cost_WITHOUT <- eligible_patients * unit_cost_SoC

# Budget impact
budget_impact_total <- total_cost_WITH - total_cost_WITHOUT

# Per-patient costs
per_patient_WITH <- total_cost_WITH / eligible_patients
per_patient_WITHOUT <- total_cost_WITHOUT / eligible_patients
budget_impact_per_patient <- per_patient_WITH - per_patient_WITHOUT

# Assemble results
budget_summary <- data.frame(
  Year = paste0("Year ", years),
  Eligible_Patients = eligible_patients,
  Cost_WITHOUT = round(total_cost_WITHOUT),
  Cost_WITH = round(total_cost_WITH),
  Budget_Impact = round(budget_impact_total),
  CostPerPatient_WITHOUT = round(per_patient_WITHOUT, 2),
  CostPerPatient_WITH = round(per_patient_WITH, 2),
  BudgetImpact_PerPatient = round(budget_impact_per_patient, 2)
)

print(budget_summary)

    Year Eligible_Patients Cost_WITHOUT  Cost_WITH Budget_Impact
1 Year 0            361800    238788000  238788000             0
2 Year 1            511800    337788000  354165600      16377600
3 Year 2            661800    436788000  489732000      52944000
4 Year 3            811800    535788000  639698400     103910400
5 Year 4            961800    634788000  819453600     184665600
6 Year 5           1111800    733788000 1018408800     284620800
  CostPerPatient_WITHOUT CostPerPatient_WITH BudgetImpact_PerPatient
1                    660                 660                       0
2                    660                 692                      32
3                    660                 740                      80
4                    660                 788                     128
5                    660                 852                     192
6                    660                 916                     256

Code

# Total cost by component
total_by_component <- colSums(cost_A_matrix + cost_SoC_matrix)

# Cost per patient (total over 6 years / total patients)
total_patients <- sum(eligible_patients)
per_patient_by_component <- total_by_component / total_patients

# Display
detailed_summary <- data.frame(
  Component = cost_components$Component,
  Total_Cost = round(total_by_component),
  Per_Patient_Cost = round(per_patient_by_component, 2)
)

print(detailed_summary)

        Component Total_Cost Per_Patient_Cost
1            Drug 1727814000           390.84
2           Admin  401574000            90.84
3         Monitor  321433500            72.71
4              AE  154815150            35.02
5 Hospitalization  180969750            40.94
6      Outpatient  773640000           175.00

Clustered Column Chart for Budget Impact Results (Per Year)

Code

# Assume budget_summary exists from earlier step

bim_plot_data <- budget_summary %>%
  select(Year, Cost_WITH, Cost_WITHOUT, Budget_Impact) %>%
  pivot_longer(cols = -Year, names_to = "Scenario", values_to = "Cost")

ggplot(bim_plot_data, aes(x = Year, y = Cost / 1e6, fill = Scenario)) +
  geom_col(position = "dodge") +
  labs(title = "Budget Impact Results Per Year",
       y = "Cost (£ millions)", x = "Year") +
  scale_fill_manual(values = c("Cost_WITH" = "steelblue", 
                               "Cost_WITHOUT" = "grey60",
                               "Budget_Impact" = "firebrick")) +
  theme_minimal()

Clustered Bar Chart for Detailed Cost by Cost Type

Code

# Assume detailed_summary from earlier exists

ggplot(detailed_summary, aes(x = Component, y = Total_Cost / 1e6, fill = Component)) +
  geom_col(position = "dodge") +
  labs(title = "Detailed Cost by Cost Type",
       y = "Total Cost (£ millions)", x = "Cost Component") +
  theme_minimal() +
  theme(legend.position = "none")

Stacked Area Chart for Market Share Over Time

Code

# Assume With_New_Drug exists
market_share_long <- With_New_Drug %>%
  pivot_longer(cols = c(Drug_A, SoC), names_to = "Treatment", values_to = "Share")

ggplot(market_share_long, aes(x = Year, y = Share, fill = Treatment)) +
  geom_area(alpha = 0.7, position = "stack") +
  labs(title = "Market Share Over Time", y = "Market Share", x = "Year") +
  theme_minimal()

Line Chart for Predicted Patient Population (Eligible)

Code

patient_data <- data.frame(
  Year = paste0("Year ", 0:5),
  Eligible_Patients = eligible_patients
)

ggplot(patient_data, aes(x = Year, y = Eligible_Patients)) +
  geom_line(group = 1, color = "darkgreen", size = 1.2) +
  geom_point(color = "darkgreen") +
  labs(title = "Predicted Eligible Patient Population Over Time",
       y = "Patients", x = "Year") +
  theme_minimal()

Warning: Using `size` aesthetic for lines was deprecated in ggplot2 3.4.0.
ℹ Please use `linewidth` instead.

Line Chart for Market Uptake (Drug A and SoC)

Code

market_uptake_long <- With_New_Drug %>%
  pivot_longer(cols = c(Drug_A, SoC), names_to = "Treatment", values_to = "Share")

ggplot(market_uptake_long, aes(x = Year, y = Share, color = Treatment)) +
  geom_line(size = 1.2) +
  geom_point() +
  labs(title = "Market Uptake Over Time",
       y = "Market Share (%)", x = "Year") +
  theme_minimal()

`geom_line()`: Each group consists of only one observation.
ℹ Do you need to adjust the group aesthetic?

Bar Chart for Total Eligible Patients by Year

Code

ggplot(patient_data, 
       aes(x = Year, y = Eligible_Patients)) +
  geom_col(fill = "skyblue") +
  labs(title = "Eligible Patients Over Time", y = "Eligible Patients", x = "Year") +
  theme_minimal()

Things to consider

Validate the inputs you are considering
Check standardized templates
Follow proper guidelines as issued by ISPOR
Make the model as simple and interactive as possible.