Same two drugs. Same data. Four different verdicts.

Every economic evaluation is the same fraction. Only one part of it ever moves.

Strip any economic evaluation down and you find an identical skeleton: two options, the cost of each and the effect of each, comparing the difference in cost against the difference in effect.

numerator: always £, never changes denominator: the only moving part

You've already met the top of that fraction. In Module 1 you met opportunity cost: in a fixed budget, extra cost means health displaced elsewhere, and the ratio itself, cost-per-unit-of-health, is the exchange rate you'll formalise later as the ICER. Hold that; it's the spine of everything in this module.

Now look at what's fixed and what floats. The cost side is always in the same currency: pounds. Cost is cost. It never changes across the four types of evaluation.

The effect side is the only thing that moves. "Effect of A minus effect of B": measured in what? Millimetres? Life-years? QALYs? Pounds? That single choice (the unit in the denominator) is the entire difference between cost-minimisation, cost-effectiveness, cost-utility, and cost-benefit analysis. Four names, one moving part.

The four evaluations aren't four tools. They're four stops on a single axis.

That denominator (how you measure effect) runs along one axis: how universal is the unit? From a unit so specific it only describes one disease, to a unit so universal it can price anything at all. Four stops:

• Cost-minimisation (CMA): the effect is assumed equal, so it drops out of the denominator entirely. The most restrictive stop: it only works when two options do the same thing.
• Cost-effectiveness (CEA): the effect is measured in a natural unit: mmHg, life-years, heart attacks avoided. Clinically direct, but the unit only speaks to its own disease.
• Cost-utility (CUA): the effect is measured in QALYs, a unit engineered to be universal across diseases. This is HTA's workhorse.
• Cost-benefit (CBA): the effect is converted into money, the most universal unit of all. Now health can be weighed against literally anything.

Read left to right and one thing increases at every step: how broad a set of decisions the unit lets you compare. Read it the other way and something else increases: how many hidden value judgements you've had to bake into the unit to make it that universal. That trade (reach versus baked-in assumptions) is the spine of this whole lesson. Let's walk the stops.

Start with the most intuitive: measure the effect in whatever unit the clinic already uses.

Cost-effectiveness analysis keeps the effect in a natural unit, a real clinical measure. Cost per millimetre of mercury lowered. Cost per life-year gained. Cost per heart attack avoided. Cost per stroke prevented.

Its strength is that it's honest and legible. A cardiologist reads "£5,000 per life-year gained" and knows exactly what it means, in units they trust, with no black box in between. For comparing two treatments for the same condition, it's often all you need: our two blood-pressure drugs come out at £250 per extra mmHg, or £5,000 per life-year, and a hypertension specialist can act on that directly.

But now the wall. Suppose the payer has £1 million left and two bids: the blood-pressure drug at £5,000 per life-year, and a new antidepressant at £8,000 per episode-free month. Which buys more health? You can't say. Millimetres don't convert to mood episodes; life-years-in-cardiology don't obviously equal life-years-in-oncology once quality differs. CEA gives you a sharp answer inside one disease and goes silent the moment you need to compare across diseases. That silence is exactly the problem the next stop was invented to solve.

Now the leap that made modern HTA possible: one unit for health, any disease.

Cost-utility analysis sits at the same spot in the fraction (still cost-per-unit-of-effect) but swaps the natural unit for a QALY, the quality-adjusted life year. A QALY fuses the two things every treatment affects (how long you live and how well you live) into a single number. One year in perfect health = 1 QALY; one year at half quality = 0.5 QALY; death = 0.

The consequence is enormous. Because every health outcome (a cancer remission, a hip that no longer aches, a depression lifted, a stroke prevented) can be expressed in QALYs, you can put the cancer drug, the hip replacement, and the antidepressant on one common scale and ask which buys the most health per pound. Our blood-pressure drug comes out at £6,667 per QALY, a number you can line up directly against a cancer therapy at £40,000 per QALY, and know which is the better buy for the system.

That is why CUA is the default in most HTA agencies, NICE included: a payer's whole job is choosing across diseases, and QALYs are the only unit that lets them. But universality was not free. Fusing length and quality into one number requires deciding how bad half-health is compared to full health, a value judgement, made by someone, using methods you'll dissect in Module 6. Hold the discomfort: the QALY's power and its controversy come from the exact same place.

See all four at once. Same two drugs, watch only the denominator change.

Below are our blood-pressure drugs, A versus B. Slide through the four evaluation types and watch two things: the unit in the denominator change, and, underneath, what that unit lets you compare open up or shut down.

Reach isn't free. Each more-universal unit smuggles in a value judgement.

Look back along the axis. Every step toward a more universal unit bought you comparability, and charged you an assumption you may not have noticed:

• CEA (natural unit) bakes in the least: a millimetre is a millimetre. Most transparent, and least useful across diseases.
• CUA (QALY) bought cross-disease comparison by baking in a judgement about how much quality-of-life is worth relative to length-of-life, and whose valuation counts (patients'? the public's?).
• CBA (money) bought comparison against everything by baking in the most contested judgement of all: a price for a year of human life. Put £30,000 on a QALY and you've implicitly priced a life, and any such price tends to value the wealthy and the economically productive more highly, because their ability to pay is higher.
• CMA (equal effect) hides its judgement at the bottom of the axis, not the top: it assumes effects are identical, itself a strong empirical claim that's usually asserted far more often than proven.

So the axis has a value judgement at both ends and in the middle: there is no neutral, assumption-free unit. Choosing the analysis is choosing which judgement you're willing to live with.

The one where effect drops out, powerful when true, dangerous when assumed.

Cost-minimisation analysis is the simplest stop: if two options produce the same effect, the effect cancels out of the fraction, and the decision collapses to one question: which is cheaper? Clean and correct… when the equal-effect claim is actually true.

Here's a legitimate case. Two versions of the same biologic (an originator and a biosimilar) with equivalence formally demonstrated (a non-inferiority trial, a pre-specified clinical margin, the difference statistically and clinically ruled out). Effect proven equal, so effect leaves the equation:

The evaluation that can compare anything, and the reason HTA keeps it at arm's length.

Cost-benefit analysis takes the final step: it converts the health effect itself into money, so both sides of the ledger are in pounds and you compute a straight net benefit: benefit-in-£ minus cost-in-£. Positive, fund it; negative, don't. It's the only evaluation whose answer is a single figure in one currency, and the only one that can weigh a health programme against a non-health one.

The machinery for pricing health is willingness-to-pay: surveys and revealed behaviour estimating what people will pay to reduce risk, which cashes out as a "value of a statistical life" or a value per life-year. Regulators outside health use it constantly: transport agencies pricing road-safety schemes, environmental agencies pricing clean-air rules.

So why does HTA mostly avoid its most complete tool? Because pricing a year of human life in pounds is politically radioactive, and because willingness-to-pay is entangled with ability-to-pay: value a life by what someone would pay for it and you systematically value the rich above the poor, the earning above the retired. HTA's escape hatch is elegant: CUA. By measuring benefit in QALYs and comparing against a threshold, HTA gets almost the decision-usefulness of CBA without ever writing down a price for life, the threshold does that work implicitly, one step removed. CBA still shows up where health spills beyond the clinic: vaccination with productivity effects, public-health programmes competing with other government budgets, broad regulatory decisions. But as the engine of a reimbursement decision, HTA reaches for cost-utility and lets the threshold price the QALY quietly.

The other chair

Same skill from both chairs, reading the choice of denominator as the value-laden decision it is, not the technical formality it pretends to be.

Why this matters for HTA

When it lands on your desk: the type of economic evaluation isn't a stylistic choice a manufacturer makes freely: it's largely dictated by the agency's reference case, the mandated recipe for how a submission must be built. Knowing the four types is knowing what you're allowed to submit, what you must challenge, and what a deviation is trying to tell you.

• You expect CUA as the default and treat departures as signals. Most agencies (NICE, and AOTMiT in Poland) require cost-per-QALY as the reference case precisely because they allocate across diseases. A submission that leads with CEA in a natural unit, or with CMA, is making a choice you need to explain before you evaluate.
• You treat CMA as a claim to be audited, not a shortcut to be accepted. Because it deletes the effect from the analysis, an unproven CMA is the single easiest way to smuggle an inferior treatment past a busy committee. Equal effect is a conclusion that must be earned, never a premise.
• You recognise that the choice of unit pre-loads the answer. The denominator decides what's comparable and whose values count, so by the time you're checking the arithmetic, the most important decision has already been made. Your leverage is in questioning the ruler, not just the measurement.

By the time the numbers arrive, the ruler has already been chosen, and the ruler decides the game. Read it first.

Types of evaluation, in one breath.

• Every economic evaluation is one fraction: cost ÷ effect, compared across two options. The cost side is always pounds. Only the effect unit changes.
• That unit runs along a single axis of universality: CMA (effect assumed equal, drops out) → CEA (natural unit) → CUA (QALYs) → CBA (money).
• The more universal the unit, the wider the set of decisions you can compare, and the more value judgement you've baked in to get there.
• CEA is transparent but silent across diseases. CUA is HTA's workhorse because QALYs compare any disease, at the cost of building quality-of-life judgements into the unit.
• CBA is the most complete and the most avoided: it can weigh health against anything, but only by pricing life in pounds, which HTA sidesteps by using QALYs and a threshold instead.
• CMA is the most misused: valid only when equal effect is proven, and a way to hide an inferior treatment when it's merely claimed.

Choosing the analysis is choosing whose values get built into the denominator. There is no view from nowhere.

You now know the four shapes an evaluation can take and how to read the choice of ruler. But we've been treating "cost" as a single settled number in the numerator, and it isn't. Whose costs count? The payer's only, or society's: lost work, informal care, travel? And a pound spent in ten years' time: does it weigh the same as one spent today? The next stretch of the module opens up that numerator: the perspective of the analysis, what counts as a cost, and how time changes its value.