Cost-Effectiveness Analysis vs. Budget Impact Model: Which Do Payers Want?

A hospital value analysis committee asks for your economic case. Your consultant sends a cost-effectiveness analysis with an incremental cost-effectiveness ratio of 18,000 dollars per quality-adjusted life year. The committee reads it, says nothing useful, and asks again what it will cost them next year. Everyone is frustrated, and nobody is wrong. You answered a question they did not ask.

Cost-effectiveness analysis and budget impact modelling are both legitimate, both standard, and not interchangeable. Knowing which one a given buyer wants is one of the cheapest advantages available to a medtech commercial team.

The two questions

A cost-effectiveness analysis asks whether the health gain is worth the extra cost, from a defined perspective, over a long horizon. It divides the incremental cost of your technology by the incremental health benefit and produces a ratio, usually cost per quality-adjusted life year. The horizon is often lifetime. The perspective is usually the health system or society. The audience is an assessor deciding whether the technology represents value at all.

A budget impact model asks a narrower question: if this specific organisation adopts this technology for its eligible population, what happens to its spending in year one, year two and year three. No discounting to lifetime. No abstract willingness-to-pay threshold. Just a population count, a rate of uptake, unit costs, offsets, and a net number per year.

Both are covered by the ISPOR good practices reports, which remain the reference set for how each model should be built and reported. If your model departs from them, expect to justify it.

Who wants which

National HTA bodies want cost-effectiveness. NICE's health technology evaluations manual sets out the reference case: the perspective, the discount rates, the outcome measure, the requirement for a probabilistic sensitivity analysis. Germany, France and most European assessors ask a version of the same question, even where the thresholds are less explicit.

In the United States, formal cost-effectiveness plays a different role. ICER's value assessment framework produces public cost-effectiveness reviews that influence negotiation, but no US payer is required to use them. What US buyers ask for directly, almost every time, is budget impact.

That is a structural fact rather than a preference. A commercial plan budgets annually, members switch plans, and a saving that lands in year seven is a saving for somebody else. Hospitals are worse still: their fiscal year is the horizon that matters, and a device that shortens length of stay only helps if the beds are actually a constraint they are paying for.

So the working rule: Europe and national assessment want cost-effectiveness; US plans, integrated delivery networks and hospital committees want budget impact. Digital health products with an international plan need both, which is why the international reimbursement pathways differ so sharply in what they ask for.

What a credible cost-effectiveness analysis contains

A defined comparator that reflects current practice, not a straw man. A model structure that matches the disease: a decision tree for a one-off intervention, a Markov or state-transition model for a chronic condition. Transition probabilities and utilities from published, citable sources, with the source named for each parameter. Costs from a defensible national source. A discount rate consistent with the jurisdiction. One-way sensitivity analysis to show which parameters drive the result, and a probabilistic analysis to show the spread.

The most common weakness is not maths. It is the utility values. If your quality-of-life gain rests on a mapping exercise from an unvalidated instrument, an assessor will discount the entire result. This is where knowing the minimum clinically important difference of your outcome measure protects you: an effect below it cannot credibly carry a utility gain.

What a credible budget impact model contains

The eligible population, derived from the plan's own membership or the hospital's own volumes, not from national prevalence multiplied by an optimistic share. Uptake over three years, with an explicit and conservative curve. The full acquisition cost, including implementation, training, integration and support, because buyers will add those back if you omit them.

Then offsets, which is where the argument is won or lost. An avoided admission is only an offset if the payer pays for admissions. A shortened procedure is only an offset if the theatre time is redeployed. Staff time saved is rarely cash saved. Credible models separate hard offsets from soft ones and price the soft ones at zero in the base case.

Finally, the reimbursement mechanism. If the device is paid inside a bundle or a DRG, the budget impact of adoption is different from a separately payable code. This is the same code-level reality that drives payment classification and it belongs in the model, not in a footnote.

Building both without doubling the work

The two models share most of their inputs: population, event rates, unit costs, effect size. Build one parameter table with sourced values, version it, and drive both models from it. When a new trial or real-world evidence analysis updates an input, both outputs update together and stay consistent, which matters because assessors and hospital committees do occasionally compare your documents.

Keep the documentation to model-audit standards. If your model informs a regulated claim, the record-keeping expectations in the 21 CFR Part 11 guide are a reasonable benchmark for version control and traceability.

A short diagnostic

Ask the buyer three questions. Over what period do you make this decision? Whose budget absorbs the cost? Do you have a threshold you compare technologies against? An answer of one to three years, our own budget, and no threshold means budget impact. An answer of lifetime, the health system, and yes means cost-effectiveness. Mixed answers usually mean a national assessment feeding local adoption, and you need both.

The market side matters too. Which of the 25+ markets you are entering determines which model gets read first, and MedTech Compass scores those differences rather than leaving them to guesswork. Sequence the evidence spend accordingly and see the broader market access picture before commissioning either model.

Sources

- ISPOR — Good Practices reports for health economics and outcomes research - NICE — Health technology evaluations: the manual (PMG36) - ICER — Value Assessment Framework - CMS — Coverage with Evidence Development - FDA — Real-World Evidence programme

This article is general information about health economic modelling, not financial, regulatory or legal advice. Confirm methods and requirements with the relevant assessor or payer before submitting.