Cost-Minimisation Analysis and the Equivalence Assumption It Rests On

Cost-minimisation analysis (CMA) is the simplest form of economic evaluation used in health technology assessment (HTA): it compares only the costs of two or more interventions, on the premise that their health outcomes have already been shown to be equivalent. Where cost-effectiveness analysis (CEA) and cost-utility analysis (CUA) exist specifically to weigh a cost difference against an outcome difference, CMA skips that step entirely — because, by assumption, there is no outcome difference to weigh. That shortcut is also the method’s single point of failure: CMA is only valid for as long as the equivalence assumption underneath it actually holds, and a large methodological literature has spent the past two decades arguing that it usually does not hold as cleanly as analysts have claimed.

What Cost-Minimisation Analysis Is

CMA is used when prior clinical evidence — typically a randomized equivalence trial or non-inferiority trial — has already established that two interventions produce statistically indistinguishable health outcomes. Once that premise is accepted, the analytic question collapses from “is the extra benefit worth the extra cost?” to simply “which option costs less?” The output of a CMA is a single figure: the cost difference between Option A and Option B, calculated the same way costs are costed in any other economic evaluation (drug acquisition, administration, monitoring, adverse-event management, and any downstream healthcare resource use attributable to each option).

Because there is no incremental-cost-effectiveness-ratio to compute — the numerator (cost) is all that is being compared, and the denominator (outcome difference) is defined as zero — CMA produces a much simpler result than a full cost-effectiveness analysis. This simplicity is exactly why it is an attractive method when it genuinely applies: generic-versus-branded drug substitutions, two delivery routes of the same active compound, or two care pathways a trial has already shown converge on the same clinical endpoints are the classic textbook cases.

The Equivalence Assumption CMA Rests On

Every conclusion a cost-minimisation analysis draws is downstream of one premise: that the interventions being compared do not differ in outcomes in any way that matters. This is not a background caveat — it is the entire justification for comparing costs alone. If that premise is wrong, a CMA does not produce an approximately-right answer with some uncertainty around it; it produces an answer to a question (“which is cheaper, given equal benefit?”) that no longer describes the real decision, which is “which choice offers better value once both cost and benefit are considered?”

The methodological literature on this point is unambiguous, and dates back at least two decades. Andrew Briggs and Bernie O’Brien’s influential 2001 Health Economics paper “The Death of Cost-Minimisation Analysis?” argued that the standard way analysts established equivalence — running a hypothesis test on the outcome difference and treating a non-significant result (“failure to reject the null of no difference”) as proof of equivalence — is a misuse of frequentist statistics. Failing to detect a statistically significant difference is not the same claim as demonstrating there is no clinically meaningful difference; the two are routinely conflated. A later re-examination by Dakin and Wordsworth (“Cost-Minimisation Analysis Versus Cost-Effectiveness Analysis, Revisited,” Health Economics, 2013) revisited the same question and found the underlying critique still held: genuinely justified CMAs are rare, because true equivalence is a much stronger and harder-to-establish claim than “no significant difference was observed.”

The practical failure mode is specific and worth naming directly: a non-inferiority trial that is underpowered, or that uses an overly generous non-inferiority margin, can return a “non-inferior” result even when a real, clinically relevant difference in outcomes exists. If an analyst then treats that non-inferiority result as license to run a CMA, the resulting cost comparison inherits the trial’s blind spot — the modeled “equivalent” outcomes were never demonstrated to the standard the CMA implicitly claims. This is why HTA bodies and peer reviewers now scrutinize the equivalence claim itself before accepting a CMA submission, rather than taking it as given once a non-inferiority trial is cited.

How CMA Differs From Cost-Effectiveness and Cost-Utility Analysis

The distinction is entirely about what happens when outcomes are not identical. Cost-effectiveness analysis measures outcomes in a natural clinical unit (life-years gained, cases averted, symptom-free days) and reports the ratio of cost difference to outcome difference — the incremental cost-effectiveness ratio (ICER). Cost-utility analysis is the specific, standardized form of CEA that always measures outcomes in quality-adjusted life years (QALYs), which is what lets HTA bodies such as NICE compare value for money across completely unrelated disease areas using one common reference case. Both methods exist specifically because most real interventions differ from their comparator by at least a small margin on both cost and outcome — the whole apparatus of an ICER, a cost-effectiveness threshold, or a cost-effectiveness plane is built to handle that trade-off.

CMA is not a cheaper substitute for CEA/CUA that an analyst can choose when a trade-off calculation feels unnecessary — it is only valid in the narrow case where the outcome side of that trade-off has been shown to be zero. Even a small, genuine outcome difference is enough to invalidate a CMA and require a full CEA or CUA instead, because once outcomes differ at all, the comparison needs some way to express how much that difference is worth relative to its cost — which is precisely the calculation CMA is built to avoid making.

Why Cost-Minimisation Analysis Has Become Less Common

Two things have pushed CMA out of routine use since Briggs and O’Brien’s critique. First, the equivalence assumption is now something reviewers actively test rather than accept on citation: HTA bodies increasingly ask analysts to justify why a CMA — rather than a CEA reporting a formal ICER — is the right method for a given submission, and a CMA built on a single underpowered or narrowly-scoped non-inferiority trial draws exactly the scrutiny Briggs and O’Brien’s critique predicted it should. Second, running a full cost-effectiveness analysis has become the lower-risk default even when two interventions are expected to perform similarly: a CEA that happens to show a near-zero outcome difference still produces a defensible ICER and a transparent sensitivity analysis, whereas a CMA that turns out to rest on a shaky equivalence claim has no such fallback — the entire analysis has to be redone as a CEA regardless. Reporting guidelines reflect the same shift: the CHEERS 2022 checklist for reporting economic evaluations still recognizes cost-minimisation studies as one of several legitimate economic-evaluation types, but treats the justification for using cost-minimisation as a distinct, checkable reporting item precisely because that justification is where these studies most often fall short.

None of this means CMA is obsolete. Where equivalence genuinely has been demonstrated with adequate statistical power — a generic bioequivalence finding, for instance, or a delivery-route switch of an identical active compound with a tight, prespecified equivalence margin cleared on both bounds — comparing costs alone remains the correct, most transparent method. The caution is about the assumption being asserted casually rather than demonstrated rigorously, not about the method itself being wrong when its precondition is actually met.

Where CMA Fits Among the Economic-Evaluation Types

CMA is one of several named economic-evaluation types a health-economics researcher needs to be able to distinguish and justify. Budget impact analysis asks a different question entirely — not “is this good value?” but “can the payer’s budget absorb this?” — and is typically run alongside, not instead of, a CEA/CUA or CMA. Markov cohort models are a modeling technique for projecting costs and outcomes over time within a CEA/CUA, not a competing evaluation type. For background on how these methods relate to health economics as a field, see what health economics covers; for the trial-design evidence a defensible equivalence claim depends on, see non-inferiority vs. superiority trial design.

Frequently Asked Questions

When is cost-minimisation analysis the right method to use?

Only when a prior study — usually a randomized equivalence or non-inferiority trial with adequate statistical power and a prespecified, clinically justified margin — has already established that the interventions being compared do not differ in outcomes in any way that matters to patients or payers. If that evidence does not exist, or is weaker than that (a single underpowered non-inferiority trial, an indirect comparison, expert opinion), a cost-effectiveness or cost-utility analysis is the appropriate method instead.

What happens if the equivalence assumption in a CMA turns out to be wrong?

The analysis is invalidated, not just weakened. A CMA reports only a cost difference; if the interventions actually differ in outcomes, that cost difference is being presented without the outcome context needed to judge whether it represents good or poor value. The correct response is to redo the evaluation as a full cost-effectiveness or cost-utility analysis incorporating the real outcome difference, not to append a caveat to the cost-only result.

Is failing to find a statistically significant difference in a trial the same as proving equivalence?

No. This is the specific error Briggs and O’Brien’s 2001 critique targeted: failing to reject the null hypothesis of no difference (often because a trial was underpowered, or used a wide non-inferiority margin) is a different, weaker claim than affirmatively demonstrating equivalence within a prespecified, clinically meaningful margin. Treating the two as interchangeable is the most common way a CMA’s foundational assumption goes wrong.

How does CMA relate to the CHEERS 2022 reporting checklist?

CHEERS 2022 treats cost-minimisation as one of several recognized types of economic evaluation and requires that a study using it explicitly justify why comparing costs alone is appropriate — i.e., document the evidence for equivalence, not just assert it. See the CHEERS 2022 checklist guide for the full 28-item reporting structure this applies within.

This guide covers the general methodology of cost-minimisation analysis, its dependence on the equivalence assumption, and its relationship to other forms of economic evaluation used in health technology assessment. It does not constitute guidance on any specific submission or reimbursement decision.