Budget Impact Analysis

This page explains what budget impact analysis measures, which information must be included, and how eligible population, treatment uptake, treatment mix, resource use and prices produce annual expenditure estimates. It also explains how reference and adoption scenarios are compared, which assumptions drive the forecast, and why affordability is different from cost-effectiveness.

Budget impact analysis is commonly shortened to BIA. Healthcare systems, insurers, hospitals, government programmes and regional budget holders use it for financial planning. The analysis estimates how expenditure may change after adoption; it does not determine whether an intervention provides good value for money.

What budget impact analysis measures

BIA estimates the financial consequences of moving from an expected future without an intervention to an expected future with it. The result is normally reported for each budget period and cumulatively over a short- to medium-term planning horizon.

BIA is a forecast, not a record of expenditure that has already occurred. Its results depend on assumptions about population, adoption, treatment mix, timing, prices, resource use and capacity.

A BIA may report:

• Reference-scenario expenditure for each period.
• Adoption-scenario expenditure for each period.
• Annual or periodic budget impact.
• Cumulative budget impact.
• Budget impact per covered member.
• Budget impact per treated patient.
• Budget impact per member per month when relevant.
• Results under alternative adoption or pricing scenarios.

How BIA differs from cost-effectiveness analysis

Budget impact analysis and cost-effectiveness analysis answer related but different questions.

Cost-effectiveness analysis asks whether differences in health outcomes justify differences in costs under a stated decision rule. Budget impact analysis asks whether the expected change in expenditure can be accommodated by a particular budget holder over a specified planning period.

An intervention can be cost effective but difficult to afford when:

• The eligible population is large.
• Uptake is rapid.
• Treatment costs occur early.
• Savings occur much later.
• Implementation requires substantial initial spending.
• The budget is fixed or narrowly defined.

An intervention can also reduce expenditure without being clinically effective or cost effective. Projected savings alone do not establish health benefit or value for money.

Start with the budget holder and decision

A BIA must identify whose budget is being analysed. Without a named budget holder, it is unclear which population, services, prices and expenditure categories belong in the calculation.

The analysis should identify:

• The budget holder.
• The covered population.
• The technology, policy or service being considered.
• The intended place in the treatment pathway.
• The reference scenario.
• The adoption scenario.
• The eligible and treated populations.
• The current and future treatment mixes.
• Uptake, displacement, switching and discontinuation.
• Relevant resources and prices.
• The planning horizon and reporting periods.
• Implementation and capacity constraints.
• Decision-relevant uncertainty scenarios.

Local population, treatment-pathway, price and utilisation data should be used when they are relevant and reliable. Evidence from another jurisdiction should not be treated as automatically transferable.

How a budget impact model is built

A BIA connects population estimates, treatment patterns, resources, prices and timing in a transparent calculation. Each stage should be traceable so a reviewer can see how changing an assumption affects annual and cumulative expenditure.

1. Define the budget holder and decision. Identify the organisation, covered population, intervention and financial question.
2. Define the reference scenario. Describe the expected future without adoption.
3. Define the adoption scenario. Describe expected practice after adoption.
4. Estimate the eligible population. Apply the relevant clinical, demographic, coverage and policy criteria.
5. Estimate treatment uptake. Determine how many eligible people are expected to receive treatment.
6. Allocate treatment mix. Distribute treated patients across the available options and remove displaced treatments.
7. Measure budget-relevant resources. Include acquisition, administration, monitoring, implementation and other applicable resources.
8. Apply prices and unit costs. Use values that reflect the budget holder’s actual decision context.
9. Model treatment timing. Account for initiation, duration, discontinuation, switching and partial-year treatment.
10. Calculate scenario expenditure. Estimate reference- and adoption-scenario expenditure for every period.
11. Calculate budget impact. Report periodic and cumulative differences.
12. Test uncertainty. Examine alternative population, uptake, price, timing, resource and capacity assumptions.
13. Validate the model. Check patient flows, treatment shares, formulas, units and scenario logic.
14. Interpret and report the forecast. Explain affordability implications without claiming that BIA establishes cost-effectiveness.

Estimating the eligible population

The eligible population is the group that meets the clinical, demographic, coverage and policy conditions for the intervention. It should be estimated separately for each budget period when population size or eligibility can change over time.

A simplified calculation is:

Eligible Populationₜ = Covered Populationₜ × Disease Frequencyₜ × Eligibility Proportionₜ

The treated population may then be estimated as:

Treated Populationₜ = Eligible Populationₜ × Treatment Uptakeₜ

The calculation may require additional stages for diagnosis, treatment eligibility, contraindications, coverage restrictions, referral, capacity or patient acceptance.

The model should represent incident and prevalent cases, mortality, recovery, discontinuation, switching, loss of eligibility, retreatment and capacity constraints where relevant. Several percentages should be multiplied together only when each represents a distinct stage in the patient pathway.

Distinguishing prevalence and incidence

A prevalence-based model estimates the number of people living with the condition during a period. An incidence-based model follows newly occurring cases and may track their consequences over time.

The choice depends on the condition, intervention and planning question. Combining prevalent and incident populations without accounting for overlap can double-count patients.

The model should explain:

• Whether the population is prevalence-based, incidence-based or mixed.
• Whether patients remain eligible in later periods.
• How mortality, recovery or loss of eligibility is handled.
• Whether new patients enter the model each year.
• Whether previously treated patients remain on treatment.
• Whether retreatment is possible.
• How capacity limits affect the treated population.

Distinguishing uptake from treatment share

Uptake describes how many eligible patients receive active treatment. Treatment or market share describes how treated patients are distributed across the available options.

These percentages represent different stages and should not be treated as interchangeable.

For example, if 60% of eligible patients receive active treatment and the new intervention has a 30% share, the new intervention does not treat 30% of the entire eligible population unless the model explicitly defines it that way.

A simplified calculation is:

Patients Receiving New Interventionₜ = Eligible Populationₜ × Uptakeₜ × New Intervention Shareₜ

Every percentage should have a clearly labelled denominator.

Comparing the reference and adoption scenarios

The reference scenario represents the expected future without adoption. It is not necessarily unchanged current practice. Treatment patterns, prices, population size and clinical practice may change even if the new intervention is not adopted.

The adoption scenario represents the expected future with the intervention. It should include:

• Patients receiving the new intervention.
• Existing treatments displaced by the new intervention.
• Patients who become newly treated.
• Changes in administration or monitoring.
• Changes in adverse events.
• Changes in downstream services.
• Implementation and capacity effects.
• Changes in treatment duration or discontinuation.

The scenarios should differ only where adoption is expected to cause a change. Treatment shares should reconcile to 100% within the relevant treated population, and mutually exclusive groups must not double-count patients or costs.

Measuring treatment and pathway costs

BIA should include expenditure relevant to the stated budget holder. Resource quantities should remain separate from prices or unit costs so the model is auditable and each input can be changed independently.

For each resource:

Resource Costᵢ,ₜ = Resource Quantityᵢ,ₜ × Unit Costᵢ,ₜ

Relevant resource categories may include:

• Treatment acquisition.
• Administration.
• Diagnostic testing.
• Routine monitoring.
• Adverse-event management.
• Hospital, outpatient or primary-care services.
• Condition-related healthcare.
• Implementation and training.
• Infrastructure or equipment.
• Patient switching.
• Discontinuation.
• Treatment wastage.
• Credible healthcare offsets.

A saving should be included only when the model identifies which resource is avoided, when it is avoided, whose budget changes and what evidence supports the estimate.

Choosing the correct price

The price used in a BIA should reflect the decision context. Depending on the organisation and available information, this may be:

• List price.
• Reimbursed price.
• Negotiated net price.
• Tender price.
• Contract price.
• Acquisition cost.
• Tariff.
• Locally approved unit cost.

List and confidential net prices are not interchangeable. If the actual price is confidential, the model may need a protected input or scenario range without publishing the confidential value.

For every important price or unit cost, the model should state:

• The source.
• The applicable organisation or jurisdiction.
• The currency.
• The price year.
• Whether taxes, rebates, discounts or dispensing fees are included.
• Whether the amount is expected to change over time.
• Any inflation or price-growth assumption.
• Any uncertainty applied to the value.

Calculating expenditure and budget impact

Expenditure is calculated for each treatment or pathway in every reporting period. The values are summed to obtain total reference- and adoption-scenario expenditure.

For treatment or pathway i in period t:

Expenditureᵢ,ₜ = Patientsᵢ,ₜ × Cost per Patientᵢ,ₜ

For each scenario:

Total Scenario Expenditureₜ = Σᵢ Expenditureᵢ,ₜ

Periodic budget impact is:

Budget Impactₜ = Adoption Scenario Expenditureₜ − Reference Scenario Expenditureₜ

Cumulative budget impact is:

Cumulative Budget Impact = Σₜ₌₁ᵀ Budget Impactₜ

A positive budget impact indicates increased expenditure. A negative budget impact indicates projected savings. Neither result establishes cost-effectiveness, clinical value or the final adoption decision.

Additional budget measures

Some budget holders require results in units that match their planning processes.

Possible additional measures include:

Budget Impact per Treated Patient = Budget Impact ÷ Number Treated

The denominator should be stated, for example patients receiving the new intervention or all treated patients.

Budget Impact per Covered Member = Budget Impact ÷ Covered Population

When monthly reporting is appropriate:

Per-Member-Per-Month Impact = Annual Budget Impact ÷ Covered Members ÷ 12

These measures should supplement rather than replace total annual expenditure. A small per-member amount can still represent a substantial total financial commitment.

Accounting for treatment timing and duration

Annual budget impact depends on when patients start treatment and how long they remain treated. Assuming a full year of treatment for everyone can overstate expenditure when patients enter gradually, discontinue, die, recover or switch.

The model should distinguish:

• One-time and recurring costs.
• Initiation and maintenance periods.
• Partial- and full-year treatment.
• Incident and prevalent patients.
• Treatment discontinuation.
• Mortality or recovery.
• Switching between treatments.
• Delayed implementation.
• Treatment waning.
• Retreatment.
• Future price changes.

If treatment starts are assumed to occur evenly throughout the year, an average partial-year adjustment may be appropriate. The assumption should be explicit and tested when it materially affects the result.

Capacity and implementation constraints

Adoption may be limited by workforce, infrastructure, diagnostic capacity, training, procurement, referral pathways or service availability. A model that assumes immediate unrestricted uptake may overstate both expenditure and the number treated.

Implementation constraints can affect:

• The maximum number treated.
• The speed of uptake.
• The timing of costs.
• Waiting lists.
• Displacement of existing services.
• Implementation expenditure.
• Regional access.
• The feasibility of projected savings.

Capacity should be modelled as a constraint when it affects realistic adoption, not treated as an afterthought.

Worked example: a year-one expenditure increase

In an illustrative example, a payer has 10,000 eligible patients. Existing treatment costs £2,100 per patient and a new treatment costs £2,400 per patient. If 30% of the eligible population receives the new treatment in year one, 3,000 patients receive the new treatment and 7,000 remain on existing treatment.

Reference-scenario expenditure is:

10,000 × £2,100 = £21,000,000

Adoption-scenario expenditure is:

(7,000 × £2,100) + (3,000 × £2,400) = £21,900,000

Year-one budget impact is:

£21,900,000 − £21,000,000 = £900,000

The result indicates a projected year-one expenditure increase of £900,000. It is not a cost-effectiveness conclusion and does not show whether the additional expenditure produces sufficient health benefit.

The example assumes full-year treatment, no implementation costs, no discontinuation and no downstream offsets. A real BIA should test whether those assumptions are appropriate.

Reporting annual cash flows

BIA normally presents undiscounted expenditure for each financial period because the budget holder needs to understand expected cash flows. Discounting should not obscure the amount expected to be paid in each period.

If present-value results are required, they should be reported separately from the undiscounted periodic results. The currency, price year, inflation assumptions, reporting frequency and whether values are nominal or real should be stated.

Testing the assumptions that drive the forecast

BIA depends on assumptions about population size, uptake, treatment mix, duration, prices, implementation and resource offsets. Scenario analysis should show which assumptions materially change annual or cumulative expenditure.

Relevant alternatives should be tested for:

• Covered population.
• Disease frequency.
• Eligibility.
• Diagnosis or referral.
• Uptake.
• Treatment share.
• Displacement.
• Treatment duration.
• Discontinuation.
• Prices and discounts.
• Administration and monitoring.
• Implementation timing.
• Capacity.
• Adverse events.
• Healthcare offsets.
• Price changes.
• Alternative planning horizons.

Scenarios should be coherent combinations of assumptions rather than arbitrary changes that create impossible patient flows or treatment shares.

Checking that the model works correctly

A BIA can produce plausible-looking totals even when its patient flows or formulas are wrong. Model-integrity checks should test both the calculations and the relationships among population, treatment and expenditure.

Checks should confirm that:

• Treatment shares reconcile to 100% where required.
• Mutually exclusive patient groups do not overlap.
• Reference and adoption populations are comparable.
• Eligible, treated and untreated populations reconcile.
• Incident and prevalent populations are not double-counted.
• Partial-year treatment is handled correctly.
• Displaced treatments and costs are removed.
• Discontinuation, mortality and switching are applied consistently.
• Currency, price year and units are consistent.
• Annual results reconcile with cumulative results.
• Scenario controls cannot create negative patients or impossible shares.
• Zero and missing values are distinguishable.
• Headline results trace back to population, resource and price inputs.

Independent review and testing are particularly important when the model will inform a major financial commitment.

How affordability rules should be interpreted

Some organisations use affordability thresholds or budget-impact tests to trigger additional negotiation, implementation planning or commercial arrangements. These rules are jurisdiction-specific policy mechanisms rather than universal economic principles.

An affordability threshold may change over time and may apply only to a particular organisation, expenditure category or period. The applicable source, jurisdiction, date and consequences of crossing the threshold should be reported.

A local threshold should not be presented as a general rule for all healthcare systems.

What BIA can and cannot decide

BIA provides information about expected expenditure and financial feasibility. It does not incorporate every factor relevant to healthcare coverage, reimbursement or adoption.

BIA can:

• Estimate periodic and cumulative expenditure changes.
• Show which population, uptake, price and resource assumptions drive the forecast.
• Compare reference and adoption scenarios.
• Examine implementation and capacity effects.
• Support budgeting, negotiation and financial planning.

BIA cannot:

• Determine whether an intervention is cost effective.
• Establish that projected savings imply clinical value.
• Determine whether health outcomes justify additional expenditure.
• Resolve equity or ethical questions.
• Guarantee actual future expenditure.
• Replace implementation planning.
• Make the final adoption or reimbursement decision.

Common mistakes and how to avoid them

Budget-impact errors often arise from unclear populations, inconsistent scenarios, double-counted patients or unrealistic uptake assumptions.

• BIA assesses affordability, not cost-effectiveness.
• The reference scenario is a forecast rather than automatically unchanged current practice.
• Uptake and treatment share represent different stages.
• Every percentage requires a clearly defined denominator.
• Displaced treatments and costs must be removed.
• Incident and prevalent populations must not be double-counted.
• Timing and treatment duration can materially change annual expenditure.
• List and confidential net prices are not interchangeable.
• Discounting should not replace budget-period cash flows.
• Savings do not prove clinical value.
• Local thresholds are not universal affordability rules.
• Capacity constraints should limit adoption when they apply.
• Missing values should not be silently treated as zero.

BIA conclusions should remain tied to the budget holder, covered population, intervention, reference and adoption scenarios, uptake, treatment mix, prices, time horizon, evidence, assumptions and local policy context.

What should be reported

A BIA report should state the budget holder, covered population, intervention, scenarios, eligible-population method, prevalence or incidence approach, treatment mix, uptake, displacement, treatment duration, quantities, unit costs, prices and discounts, currency and price year, implementation and capacity assumptions, planning horizon, periodic and cumulative results, per-member measures where relevant, uncertainty scenarios, validation checks, sources, limitations and the distinction between affordability and value for money.

Sources

• Sullivan SD, Mauskopf JA, Augustovski F, Jaime Caro J, 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 in Health. 2014;17(1):5-14.
• Mauskopf JA, Sullivan SD, Annemans L, Caro J, Mullins CD, Nuijten M, et al. Principles of good practice for budget impact analysis: report of the ISPOR Task Force on good research practices, budget impact analysis. Value in Health. 2007;10(5):336-347.
• National Institute for Health and Care Excellence. NICE technology appraisal and highly specialised technologies guidance: the manual (PMG36). Published 2022; updated 2026.