ImpactMojoCost Effectiveness 101www.impactmojo.in

ImpactMojo 101 Series · Free Forever

Cost
Effectiveness
101

Getting the Most Good per Rupee — Comparing Interventions on Cost per Outcome, a Foundational Course for Practitioners & Funders in South Asia

Research-BackedSouth Asia Focus100 SlidesFree Access

ImpactMojoCost Effectiveness 101www.impactmojo.in

Agenda

What We Cover

Why Cost-Effectiveness?

Slides 3–11

The Family of Methods

Slides 12–20

Measuring Costs

Slides 21–29

Measuring Effects

Slides 30–38

The Cost-Effectiveness Ratio

Slides 39–47

Cost-Utility & DALYs

Slides 48–56

Time & Discounting

Slides 57–65

Comparing Interventions

Slides 66–75

Uncertainty & Sensitivity

Slides 76–83

Limits & Equity

Slides 84–91

Practice & Tools

Slides 92–99

ImpactMojoCost Effectiveness 101www.impactmojo.in

Section One

Why Cost-Effectiveness?

ImpactMojoCost Effectiveness 101www.impactmojo.in

The Core Reality

Budgets are finite; needs are not

Every development organisation, ministry and funder faces the same wall: there is never enough money to do everything worth doing. Cost-effectiveness is the discipline of choosing well when you cannot fund it all — doing the most good with the resources you have.

The question is not 'Does this programme work?' but 'Does it do more good per rupee than the alternatives we could fund instead?'

The question CEA answers	The question it does not
Which of these does more good per rupee?	Is this programme worth doing at all?
What does the upgrade cost per extra unit?	Who should receive it?
Where would the next rupee do most?	Whether the outcome is the right one to want

The right-hand column is not a gap to be filled by better analysis. Those are value questions, and a ratio cannot settle them — which is why Section 10 exists rather than being an appendix.

The discipline is worth having even where the numbers end up rough. Forcing a team to name the outcome, cost every ingredient and state the comparator usually changes the conversation before any ratio appears.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Opportunity Cost

Every rupee spent is a rupee not spent elsewhere

Opportunity cost

The value of the best alternative you give up when you choose one option. The true cost of funding programme A is the good you could have done with programme B instead.

A health budget spent on a costly tertiary procedure is a budget not spent on the immunisations or bed nets it could have bought. Opportunity cost makes that trade-off visible.

Decision	The visible cost	The opportunity cost
Fund a tertiary cardiac unit	The capital and running budget	The immunisation rounds not run
Extend a pilot in one district	The extension budget	A different district that gets nothing
Hire an M&E specialist	One salary	Two field staff
Do nothing while deciding	Zero on the ledger	A year of outcomes forgone

The last row is the one that never appears in a budget. Delay is a choice with a cost, and because it is invisible in accounting terms it is systematically under-weighted in committees.

Opportunity cost is defined against the best alternative forgone, not the average one. If you have not identified what that alternative is, you have not yet stated the cost of your choice.

ImpactMojoCost Effectiveness 101www.impactmojo.in

The Most Good

'The most good per rupee'

Two programmes can both be genuinely good and still differ enormously in how much good they buy per rupee. Cost-effectiveness analysis surfaces those differences so scarce funds flow where they help most.

MONEY: a fixed budget

→

CHOICE: many worthy programmes compete

→

EVIDENCE: cost per outcome for each

→

DECISION: fund the most outcomes per rupee

Step	The question at that step
A fixed budget	What can we actually spend?
Competing worthy options	What are the real alternatives, including current practice?
Evidence on cost per outcome	Estimated how, from where, with what uncertainty?
Allocation	Who gains and who loses if we shift?

Step two is where most analyses go wrong. If the option set is drawn up before the analysis, the ratio only ranks what someone already thought of — and the best buy may not be on the list.

Include "do nothing" and "do the current thing better" as options. Both are frequently competitive and both are routinely left out of comparisons of new proposals.

ImpactMojoCost Effectiveness 101www.impactmojo.in

A Worked Intuition

Same budget, very different impact

Imagine ₹10 lakh to spend on reducing child illness. Programme A averts one case for ₹5,000; programme B averts one for ₹500. The same money buys 200 averted cases instead of 20 — a tenfold difference in good done.

200 vs 20

cases averted for the same ₹10 lakh

Illustrative

10×

more impact from the cheaper-per-outcome option

Illustrative figures — but the logic is real: differences between interventions are often this large, not marginal.

	Programme A	Programme B
Cost per case averted	₹5,000	₹500
Cases averted for ₹10 lakh	200	2,000
Both are genuinely effective?	Yes	Yes

Both programmes work. That is what makes the comparison uncomfortable. Cost-effectiveness rarely separates good from bad; it separates good from better, and the loser is usually somebody’s well-run project.

Before accepting a tenfold gap, check that the two ratios were built the same way — same perspective, same outcome definition, same time horizon. Differences of this size are as often methodological as real.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Not About Being Cheap

Cost-effective is not the same as cheap

Cheapest

Lowest total cost — but may deliver almost nothing. A programme that does little for very little money is not a bargain.

Cost-effective

Most outcome per rupee. A more expensive programme can be more cost-effective if it delivers far more good.

Cost-effectiveness always holds cost and outcome together. Neither number means much alone.

Programme	Cost	Outcomes	Cost per outcome
Cheapest	₹1 lakh	5	₹20,000
Mid	₹5 lakh	100	₹5,000
Most expensive	₹20 lakh	800	₹2,500

The most expensive programme here is the most cost-effective, by a factor of eight over the cheapest. Total cost tells you what fits the budget; it tells you nothing about value.

The converse trap is just as common in practice: a low unit cost achieved by serving only the easiest cases, which makes the ratio look good and leaves the actual problem untouched.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Who Uses It

Where cost-effectiveness shapes decisions

Health ministries deciding which treatments a public scheme covers
Funders allocating limited grants across many strong applicants
NGOs choosing between programme designs for the same goal
Global bodies (WHO, World Bank) ranking interventions across diseases

From a panchayat health post to a national budget, the underlying question is identical: where does each rupee do the most?

Who uses it	For what decision
Health ministries	Which treatments a public scheme covers
India’s HTA agency	Advice on technologies and their prices
Funders and grant committees	Allocating across strong applicants
NGOs	Choosing between designs for one goal
WHO, World Bank, DCP3	Ranking interventions across diseases

India has a formal health technology assessment process under the Department of Health Research, which appraises interventions and technologies for public financing — the same logic applied at national scale.

At panchayat or district level the arithmetic is simpler and the discipline is the same: name the options, cost them fully, agree the outcome, and be explicit about who each option reaches.

ImpactMojoCost Effectiveness 101www.impactmojo.in

A Tool, Not a Verdict

Cost-effectiveness informs; it does not decide

Cost-effectiveness analysis is one input into a decision, not the whole decision. Equity, rights, feasibility and politics all matter too — themes we return to in Section 10.

The aim is not to put a price on a life, but to be honest about the lives a fixed budget can and cannot save.

— a working principle of health economics

CEA supplies	The decision also needs
Cost per unit of outcome	Whether that outcome is the right goal
A ranking of options	Who each option reaches
A defensible efficiency case	Rights, obligations and law
A basis for reallocation	Whether reallocation is politically possible

Present the ratio alongside the other considerations, not before them. A number produced first tends to anchor the discussion, and everything else then argues against it from a weaker position.

The strongest use of CEA in practice is diagnostic: when a programme looks far worse than its peers, the finding is usually a design problem worth fixing rather than a case for closure.

ImpactMojoCost Effectiveness 101www.impactmojo.in

South Asia Stakes

Why this matters acutely in South Asia

Public health spending per person in much of South Asia is among the lowest in the world relative to need, while disease and deprivation burdens are high. The gap between resources and need makes every allocation choice consequential.

When budgets are tight and needs vast, choosing well is not an academic nicety — it is the difference between reaching many more people or far fewer with the same money.

South Asian condition	Consequence for allocation
Low public health spending per head	Every rupee displaces another rupee of care
High disease and deprivation burden	Many interventions are competitive on paper
High out-of-pocket share of health spending	Participant costs dominate and are rarely counted
Wide district-level variation	A national average ratio fits almost nowhere

The third row is the distinctive one. Where households pay a large share of health costs directly, a provider-perspective analysis omits most of what the intervention actually costs society.

It also means that a programme which reduces out-of-pocket spending can be highly cost-effective from a societal perspective and look like pure cost from a provider one.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Section Two

The Family of Methods

ImpactMojoCost Effectiveness 101www.impactmojo.in

Three Cousins

CEA, CBA and CUA: one family, different units

Economic evaluation comes in three main forms. All compare cost against benefit — they differ in how they measure the benefit, and that choice shapes what you can compare.

CEA: benefit in natural units (cases averted)

→

CUA: benefit in DALYs / QALYs

→

CBA: benefit converted into money

Method	Denominator	Comparison scope	Main hazard
CEA	Natural units	Same outcome only	Cannot compare across outcomes
CUA	DALYs / QALYs	Across health	Weights embed contested values
CBA	Money	Across sectors	Requires pricing health and life

Comparison scope and hazard rise together. Each step towards a more universal denominator buys breadth by importing a value judgement into the measurement itself.

Choose the narrowest method that answers your question. Reaching for CBA when everyone shares one outcome adds contestable monetisation and no additional comparison.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Method 1

Cost-Effectiveness Analysis (CEA)

Cost-Effectiveness Analysis

Compares the cost of interventions per unit of a single natural outcome — e.g. cost per case of malaria averted, per child immunised, per additional year of schooling.

Strength: the outcome stays in real, intuitive units. Limit: you can only compare programmes that share the same outcome. Cost per case averted cannot be set against cost per child enrolled.

CEA works when	It breaks when
All options share one outcome	Options produce different kinds of good
The outcome is well defined	The unit shifts — "case detected" vs "case cured"
The outcome is what you care about	It is a convenient proxy for something else
You are choosing within a sector	You must choose between sectors

Unit drift is the quiet killer. Two screening programmes reported as "cost per case" may be counting detection and cure respectively, which are separated by an entire treatment cascade.

Define the outcome in one sentence, with its measurement, before collecting any cost data. Doing it afterwards invites defining it to suit the numbers you already have.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Method 2

Cost-Utility Analysis (CUA)

Cost-Utility Analysis

A special form of CEA that measures benefit in a common health-and-quality unit — the DALY or QALY — so that very different health programmes can be compared on one scale.

Because a DALY captures both length and quality of life, CUA lets you ask whether a rupee does more against blindness, diarrhoea or heart disease — comparisons CEA cannot make. We devote Section 6 to it.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Method 3

Cost-Benefit Analysis (CBA)

Cost-Benefit Analysis

Converts all benefits into money, so cost and benefit share the same unit. The result is a net benefit (benefits − costs) or a benefit-cost ratio.

Power: you can compare a road, a school and a clinic on one ledger. Danger: putting a rupee value on health, education or a life forces uncomfortable — and contested — assumptions.

CBA requires pricing	Method used	Objection
A statistical life	Willingness-to-pay or wage-risk studies	Values scale with income, so poorer lives price lower
A year of schooling	Later earnings	Reduces education to labour-market return
Environmental damage	Stated preference surveys	Hypothetical answers to hypothetical prices

The income objection is the serious one. Willingness-to-pay methods make a life in a rich country worth more than a life in a poor one, and any cross-country CBA inherits that directly.

CBA earns its place where the comparison genuinely spans sectors and no other method can bridge them. It is a poor default choice inside a single sector where CEA would do.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Side by Side

Which method measures what?

Method	Benefit measured in	Lets you compare
CEA	Natural units (cases, children, years)	Programmes with the same outcome
CUA	DALYs or QALYs	Any health programmes, across diseases
CBA	Money (₹)	Anything — health, roads, schools
CMA*	(costs only; outcomes assumed equal)	Options with identical effect

*Cost-minimisation analysis is the special case where two options give the same outcome, so you simply pick the cheaper.

You need to compare	Use
Three designs for raising immunisation coverage	CEA — shared natural outcome
A blindness programme against a diarrhoea programme	CUA — DALYs bridge them
A rural road against a district hospital	CBA — only money spans sectors
Two suppliers of an identical product	Cost-minimisation — effects are equal by assumption

Cost-minimisation is the most misused of the four. It is valid only when effects are genuinely equivalent, and "roughly similar" is not equivalent — it usually means nobody measured the difference.

The choice follows from the comparison you need, not from what data you happen to hold. Fitting the method to available data is how analyses end up answering a question nobody asked.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Choosing

Pick the method that fits the question

Comparing designs for one outcome? → CEA
Comparing across different health conditions? → CUA
Comparing across sectors (health vs infrastructure)? → CBA
Outcomes already known to be identical? → cost-minimisation

The method is not a matter of taste; it follows from what you need to compare and what you are willing to monetise.

Your comparison	Method	What you must then supply
Designs for one outcome	CEA	One clearly defined natural unit
Across health conditions	CUA	Disability or utility weights, and a threshold
Across sectors	CBA	Money values for non-market benefits
Identical effects	Cost-minimisation	Evidence that the effects really are identical

Every step to a broader method adds a requirement, not just a capability. The right-hand column is what you take on, and each entry is contestable in a way natural units are not.

So the default should be the narrowest method that spans your options. Reaching further buys comparison scope you may not need at a price in defensibility.

ImpactMojoCost Effectiveness 101www.impactmojo.in

A Shared Logic

All four share one ratio at heart

Despite the different units, every method asks a version of the same thing: how much do we pay for how much good? The numerator is always cost; only the denominator — the outcome — changes.

value = cost / outcome

ImpactMojoCost Effectiveness 101www.impactmojo.in

A Caution

Different methods, different answers

The same programme can look excellent under one method and mediocre under another, because each weighs outcomes differently. The method is a lens, and lenses shape what you see.

Always report which method you used, and never compare a CEA result for one programme against a CBA result for another. Units must match.

Never compare	Because
A CEA ratio against a CBA ratio	Different denominators; the units are not commensurable
Provider-perspective against societal-perspective costs	One counts participant costs, the other does not
Ratios using different discount rates	The rate alone can change the ranking
An average CER against an ICER	They answer different questions

Each row is a live error in published league tables, not a hypothetical. Entries drawn from different studies almost never share perspective, horizon and discount rate unless someone deliberately harmonised them.

When you cannot harmonise, say so and present the entries separately. A table with a footnote about incomparability is more useful than a ranking that implies comparability it does not have.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Section Three

Measuring Costs

ImpactMojoCost Effectiveness 101www.impactmojo.in

Start Here

Costs are harder to measure than they look

The 'cost' of a programme is rarely the figure on the grant agreement. Real cost includes staff time, volunteers, donated goods, shared overheads and the time of the people you serve. Miss these and you flatter the programme.

A good cost estimate is a careful inventory, not a glance at the budget line.

Cost routinely omitted	Effect on the ratio
Head-office overheads	Understates cost; flatters the programme
Volunteer and community time	Hides the true resource use, often substantial
Donated goods and seconded staff	Makes a subsidised model look replicable
Participants’ time, fares and lost wages	Ignores the largest cost for the poorest
Capital already owned	Treats an existing building as free

Every omission runs the same direction. There is no common error that overstates cost, which means uncorrected estimates are systematically optimistic rather than merely noisy.

This matters most when comparing an NGO pilot with a government programme: the pilot often runs on donated inputs the state would have to buy, so its ratio does not transfer.

ImpactMojoCost Effectiveness 101www.impactmojo.in

The Method

The ingredients (or inventory) method

Ingredients method

Cost estimation that lists every resource an intervention uses — the 'ingredients' — values each at its real price, and adds them up. Also called the inventory method.

LIST every resource used (the ingredients)

→

QUANTIFY how much of each

→

PRICE each at its true value

→

SUM to a total programme cost

Ingredient	Priced at
Staff time	Salary plus benefits, apportioned by time actually spent
Donated goods	Market price — not zero
Volunteer time	The wage they forgo, or a local unskilled rate
A building already owned	Rental equivalent, or annualised capital cost
A vehicle	Annualised over its useful life, plus running costs

"Priced at its true value" means opportunity cost, not what appeared on an invoice. A resource used here could have been used elsewhere, whether or not anyone paid for it this year.

Annualise capital rather than charging it all to year one. A vehicle bought in the pilot year makes that year’s ratio look terrible and every later year look artificially good.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Direct vs Indirect

Direct and indirect costs

Direct costs

Tied straight to delivery: medicines, field staff salaries, training materials, transport to villages.

Indirect costs

Shared support that enables delivery: head-office rent, accounts, IT, management. Must be apportioned, not ignored.

Leaving out indirect costs is the most common way a programme is made to look cheaper than it really is.

Apportioning indirect cost	Basis	Distorts when
By share of direct cost	Simplest	One programme is capital-heavy
By staff headcount	Fair for people-heavy support	Programmes differ in staff intensity
By beneficiary numbers	Intuitive	Reach varies in effort per person
Not at all	Common	Always — it understates cost

Omitting overheads is the single most common way a cost estimate flatters a programme, and it is usually not deliberate: overheads sit in a different budget line and nobody owns the allocation.

State the apportionment basis. Two organisations using different bases will report different unit costs for identical work, and neither is wrong — but they are not comparable.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Fixed vs Variable

Fixed and variable costs

Fixed costs do not change with scale — a training centre, a vehicle, the management team
Variable costs rise with each unit served — vaccines, stipends, per-child materials
Together they shape how cost-per-outcome changes as a programme grows

This is why pilots look expensive: fixed costs are spread over few beneficiaries. Cost per outcome usually falls as you scale — up to a point.

	Fixed	Variable
Examples	Training centre, vehicle, management team	Vaccines, stipends, per-child materials
With scale	Spread thinner per beneficiary	Rise roughly in proportion
At pilot size	Dominate; ratio looks bad	Small share
At scale	Small share	Dominate; ratio flattens out

A pilot’s cost per outcome is not a forecast of the scaled programme’s. Reporting the pilot ratio as if it were is one of the most common overstatements in programme costing.

Project the ratio at plausible scale, separating fixed from variable, and show both. Any funder deciding on scale-up needs the second number, and the pilot only tells them the first.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Economies of Scale

Why cost per outcome falls with scale

Cost per beneficiary as a programme scales (illustrative)

Illustrative

Fixed costs spread over more people, so the curve falls steeply then flattens. Illustrative numbers, but a near-universal shape.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Whose Costs?

The perspective decides what counts

Perspective	Counts costs to…	Often misses
Provider	The implementing organisation	Costs borne by participants
Participant	The beneficiary — travel, time, fees	Provider overheads
Societal	Everyone affected — the fullest view	Hardest to measure fully

State the perspective up front. A programme that is cheap for the provider may be costly for a poor participant who loses a day's wage to attend.

Perspective	Counts	Typically omits
Provider	What the implementer spends	Everything borne by participants
Participant	Travel, fees, time, lost wages	Provider overheads
Health system	All public-sector costs	Household and private costs
Societal	All costs to anyone	Nothing — but is hardest to measure

Perspective is the single most consequential methodological choice and the one most often unstated. A programme that shifts cost onto participants looks efficient from a provider perspective and may not be.

State it in the first line of the methods. Two ratios computed from different perspectives are not comparable, and a reader cannot detect the mismatch unless you say which you used.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Hidden to the Poor

Participant costs are real costs

For a daily-wage worker, attending a 'free' health camp can mean lost income, bus fare and childcare. From a societal perspective those are genuine programme costs — and a reason people drop out.

Designs that lower participant costs — doorstep delivery, evening timings, compensation — can improve cost-effectiveness by lifting uptake, not just by spending less.

Participant cost	Typical size
Lost wages for a day	Often the largest single item for a daily-wage worker
Transport to the facility	Doubles for an accompanying family member
Childcare, or children brought along	Rarely counted, always borne
Informal payments	Real, undocumented, and regressive

Participant costs are the main reason "free" services go unused, and they fall hardest on exactly the people a programme is meant to reach. Omitting them makes uptake look like a behaviour problem.

Design changes that cut participant cost — doorstep delivery, evening timings, a single visit instead of three — often improve the ratio more than any efficiency gain on the provider side.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Where the Money Goes

A typical programme cost breakdown

Illustrative share of total cost by ingredient

Illustrative

Notice overheads and participant time together are nearly a quarter of cost — the very items most often left out.

Drop those slices and the programme looks ~23% cheaper than it really is. Illustrative shares.

Cost line	Frequently excluded?
Direct delivery — staff, materials	No
Head-office overheads	Very often
Participant time and travel	Almost always
Capital, annualised	Often

The two lines most often dropped are the two hardest to observe from inside the organisation. Overheads sit in another budget; participant costs never appear in any ledger at all.

Together they can be a quarter of true cost, which means an estimate omitting both reports a programme as roughly a quarter cheaper than it is — and the error is one-directional.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Section Four

Measuring Effects

ImpactMojoCost Effectiveness 101www.impactmojo.in

The Denominator

Effects: the 'good' you are buying

If cost is the numerator, the effect is the denominator — the outcome the programme produces. Getting it right matters as much as costing, and is often harder.

In CEA the effect stays in its natural unit: a case averted, a child immunised, a TB patient cured, a year of schooling added.

Getting the denominator wrong	What happens
Counting outputs instead of outcomes	Cheap per unit, worthless per rupee
Counting reach instead of change	Rewards breadth over depth
Using a proxy nobody validated	A ratio precise about the wrong thing
Counting only outcomes you can measure	The programme optimises towards measurability

The denominator does more damage than the numerator. Costing errors move a ratio by tens of per cent; choosing the wrong outcome can move it by an order of magnitude and reverse a ranking.

Ask what would have to be true for this denominator to be the thing you actually care about. If the answer is a long chain of assumptions, cost the chain or move further down it.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Natural Units

Outcomes in their own real units

Health: cases averted, deaths prevented, patients cured
Education: additional years of schooling, test-score gains
Nutrition: children moved out of stunting
WASH: people with sustained access to safe water

Natural units are intuitive and trusted by practitioners — but they only allow comparison within the same outcome.

Sector	A defensible natural unit	A weak one
Health	Cases averted, deaths prevented	People screened
Education	Additional years of schooling; learning gains	Children enrolled
Nutrition	Children moved out of stunting	Supplements distributed
WASH	People with sustained safe-water access	Taps installed

Every weak unit in the right-hand column is an output, and every one is cheaper and easier to count. That asymmetry is what pulls programme reporting towards them.

"Sustained" is doing real work in the WASH row. A tap that fails in eighteen months and a tap maintained for a decade are the same output and entirely different outcomes.

ImpactMojoCost Effectiveness 101www.impactmojo.in

A Chain

Intermediate vs final outcomes

Intermediate outcome

A step on the causal path — bed nets distributed, children dewormed, people trained. Easy to measure, but not yet the thing you ultimately care about.

Final outcome

The end goal — malaria cases averted, lives saved, income raised. Harder to measure, but it is what actually matters.

Outcome level	Example	Measurable?	Meaningful?
Output	Nets distributed	Easily	Only if used
Intermediate	Nets used nightly	With effort	Closer
Final	Malaria cases averted	Hard	Yes
Ultimate	Deaths prevented, income raised	Hardest	Yes

Measurability and meaning run in opposite directions, which is why programmes drift up the chain towards outputs. It is not dishonesty; it is what the monitoring system can deliver.

Where you must cost an intermediate outcome, state the assumed link to the final one and test it in sensitivity analysis. That converts a hidden assumption into a visible one.

ImpactMojoCost Effectiveness 101www.impactmojo.in

The Results Chain

From inputs to impact

INPUTS: money, staff, nets

→

OUTPUTS: nets distributed

→

OUTCOMES: nets used nightly

→

IMPACT: malaria cases averted

Costing per output (nets distributed) is cheap to measure but can mislead — a net unused averts nothing. Push toward outcomes and impact where you can.

Cost per…	Cheap to measure?	Can mislead because
Net distributed	Yes	An unused net averts nothing
Net used nightly	Harder	Usage may not be sustained
Case averted	Hard	Needs a counterfactual
Death prevented	Hardest	Rare events; large samples needed

Move one step further down the chain than your monitoring system offers. If you can only cost outputs, at least model the step to outcomes explicitly and test the assumed conversion rate.

The conversion rate between output and outcome is usually the largest uncertainty in the whole analysis, and it is usually assumed rather than measured.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Attribution

Did the programme really cause the effect?

Cost-effectiveness needs the effect caused by the programme — not the total change, some of which would have happened anyway. That requires a credible counterfactual: what would have occurred without it.

Crediting a programme with outcomes it did not cause overstates its cost-effectiveness. This is why rigorous evaluation (RCTs, quasi-experiments) underpins the best cost-effectiveness estimates.

Effect estimate	Counterfactual	Risk
Before-and-after change	None — assumes nothing else changed	Credits the programme with the trend
Comparison with non-participants	Weak — they differ by selection	Selection bias, usually upward
Quasi-experimental	Plausible, assumption-dependent	Fails if the assumption fails
RCT	Strong	May not transfer to your setting

Without a counterfactual you are dividing cost by the total change, not the programme’s effect. In a setting where the outcome was already improving, that can overstate cost-effectiveness several-fold.

Monitoring data is the most available and the least suitable source. It records what happened to participants, which is not the same as what the programme caused.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Cost per Outcome

Putting cost and effect together

Divide total cost by total effect and you have the headline number: cost per outcome. ₹6 lakh spent to avert 300 cases is ₹2,000 per case averted.

₹6,00,000

total programme cost

Illustrative

₹2,000

per case averted (cost ÷ 300 cases)

Element	Value	What to check
Total cost	₹6,00,000	Perspective; overheads; participant time
Effect	300 cases averted	Against what counterfactual?
Cost per outcome	₹2,000	Average or incremental?
Comparator	Unstated	This is the missing piece

A cost-per-outcome figure with no stated comparator is incomplete. ₹2,000 per case averted compared with nothing is a different claim from ₹2,000 compared with the existing programme.

Run the four checks above on any ratio you are handed. Three of them will usually be unanswerable from the document, and that itself tells you how much weight it can bear.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Measuring Effect

Where effect estimates come from

Source	Strength	Caution
RCT / trial	Strong causal estimate	May not transfer to your context
Quasi-experiment	Real-world, plausible counterfactual	Assumptions can fail
Programme monitoring	Cheap, available	No counterfactual
Published meta-analysis	Pools many studies	Settings differ from yours

An effect estimate is only as good as the evidence behind it. Borrowed effect sizes need a sanity check against your reality.

Source	Use it when	Adjust for
Your own RCT	It exists — rare	Nothing; this is the best case
RCT from elsewhere	Context is broadly similar	Baseline burden, prices, delivery quality
Quasi-experiment	No trial is feasible	State and test the identifying assumption
Meta-analysis	Several studies exist	Heterogeneity; the pooled estimate may fit nobody
Monitoring data	Only for costs, not effects	It has no counterfactual

A pooled meta-analytic effect can be an average of settings none of which is yours. Look at the spread across studies as well as the central estimate, and use the spread in sensitivity analysis.

Whatever the source, carry its uncertainty into the ratio. An effect estimate with a wide confidence interval produces a cost per outcome with a correspondingly wide range, and reporting only the midpoint hides that.

ImpactMojoCost Effectiveness 101www.impactmojo.in

A Warning

Cheap outputs can buy nothing

It is easy to be cost-effective at producing outputs nobody uses. The discipline is to cost what changes lives, not what is easy to count.

— a programme-evaluation maxim

Always ask whether your denominator is a real outcome or a convenient proxy. The choice can swing the ratio by an order of magnitude.

Denominator chosen	Cost per unit	What it means
Leaflets printed	Very low	Almost nothing
People trained	Low	Attendance, not change
Practice changed	Higher	The intended outcome
Health outcome improved	Highest	The thing worth buying

The same programme can report four ratios differing by orders of magnitude, all arithmetically correct, depending only on which row it chose to divide by.

When comparing two published ratios, check the denominator first. Differences in the outcome definition explain more apparent variation between studies than any real difference in efficiency.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Section Five

The Cost-Effectiveness Ratio

ImpactMojoCost Effectiveness 101www.impactmojo.in

The Headline Number

The cost-effectiveness ratio

Cost-effectiveness ratio (CER)

Total cost divided by total effect — the average cost of one unit of outcome. Lower is better: less money for the same good.

CER = total cost / total effect

CER tells you	It assumes
The average cost of one unit of outcome	That units are interchangeable
A basis for comparison	That other entries were built the same way
Whether a programme is worth running	That the comparator is doing nothing
Nothing about the next unit	Constant returns — usually false

Lower is better only within a valid comparison. A lower CER achieved by counting an easier outcome, or by excluding participant costs, is not better value — it is a different measurement.

The fourth row is why the CER cannot guide scale-up decisions. Average cost across everyone reached says nothing about what reaching the next group would cost.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Average CER

The average ratio compares to doing nothing

The average CER compares a programme against a baseline of doing nothing. It answers: across everyone reached, what did each unit of outcome cost on average?

Useful for a first sense of a programme — but real decisions usually compare options against each other, not against nothing. That needs the incremental ratio.

	Average CER	ICER
Compares against	Doing nothing	The next-best option
Answers	What did the whole programme cost per unit?	What does the upgrade cost per extra unit?
Right for	Whether to run a programme at all	Choosing between options
Misused as	Justification for an upgrade	Rarely misused — more often absent

"Doing nothing" is almost never the real alternative. There is usually a current practice, and the honest comparator is that practice rather than an empty counterfactual nobody would choose.

Naming the comparator is half the analysis. Two studies of the same intervention with different comparators will produce different ratios and both can be correct.

ImpactMojoCost Effectiveness 101www.impactmojo.in

The Key Concept

The Incremental Cost-Effectiveness Ratio (ICER)

ICER

The extra cost of switching from option B to option A, divided by the extra effect gained. It measures the cost-effectiveness of the upgrade, not the whole programme.

ICER = cost_A − cost_B / effect_A − effect_B

The ICER is incremental, not average. It is the single most important — and most misread — number in the field.

ICER component	What to get right
Cost A − Cost B	Both costed on the same perspective and horizon
Effect A − Effect B	Both effects from comparable evidence
The comparator B	Current practice, not "nothing"
The threshold	Stated, justified, and varied in sensitivity analysis

An ICER is a ratio of two differences, so it inherits the uncertainty of both. Small differences in the denominator produce wildly unstable ratios — and when the effect difference is near zero, the ICER is meaningless.

Report the incremental cost and incremental effect separately alongside the ratio. A reader can then see whether the ICER rests on a substantial difference or a rounding error.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Why Incremental

Why the increment, not the average, matters

You are rarely choosing between a programme and nothing. You are choosing whether the extra spend on a better option is worth the extra good it buys. The ICER isolates exactly that trade-off.

A new treatment may be excellent on average yet a poor upgrade — if it costs far more than the current one for only a little extra benefit. The ICER catches what the average hides.

	Cost	Effect	Average CER	ICER vs previous
Do nothing	0	0	—	—
Option B	₹5,00,000	200	₹2,500	₹2,500
Option A	₹8,00,000	260	₹3,077	₹5,000

Option A’s average ratio is ₹3,077 and its incremental ratio is ₹5,000. The decision is about the upgrade, so ₹5,000 is the number that matters — and it is 62% higher than the average suggests.

This gap widens the more effective the baseline option already is. A strong existing programme makes every improvement on it expensive at the margin, which averages conceal entirely.

ImpactMojoCost Effectiveness 101www.impactmojo.in

A Worked ICER

Comparing two TB-screening strategies

Strategy	Cost	Cases detected
B: standard screening	₹5,00,000	200
A: enhanced screening	₹8,00,000	260
Difference (A − B)	₹3,00,000	60 extra

ICER = ₹3,00,000 ÷ 60 = ₹5,000 per extra case detected. The question for the funder: is an extra case worth ₹5,000? Illustrative figures.

Question the funder should then ask	Why
Is an extra detected case worth ₹5,000?	This is the threshold question
Detected, or treated and cured?	Detection without treatment averts nothing
Who are the extra 60 cases?	Enhanced screening may reach a different group
Does the ₹5,000 hold at scale?	Marginal cost usually rises

The second question is the one that most often changes the answer. A screening upgrade that detects more cases and does not improve linkage to treatment has bought a statistic.

The third matters for equity: if the extra 60 are the hardest to reach, a ratio that looks poor may be exactly the spending a fairness criterion would prioritise.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Reading the ICER

Is the ICER 'worth it'?

An ICER only means something against a threshold — the most you are willing to pay for one unit of outcome. Below the threshold, the upgrade is judged worthwhile; above it, not.

Thresholds are themselves value judgements, often tied to a country's income. We return to this in Sections 6 and 10 — there is no neutral 'correct' threshold.

Threshold basis	Problem with it
1–3× GDP per capita	Arbitrary; unrelated to what the budget can afford
What was funded last year	Entrenches historical choices
A published benchmark from elsewhere	Reflects another country’s budget and burden
Opportunity cost of the actual budget	Correct in principle, demanding to estimate

The last row is what a threshold is supposed to represent: the health displaced elsewhere when you fund this. Empirical estimates of that displacement come out well below the GDP-based rules of thumb.

WHO has moved away from blanket GDP multiples for exactly this reason. If you use one, say it is a convention rather than a finding, and show the result at other thresholds.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Dominance

When the choice is obvious

Dominant

An option that costs less and does more dominates — always choose it. No ICER needed.

Dominated

An option that costs more and does less is dominated — drop it from the comparison entirely.

ICERs only matter for the hard middle cases: more cost and more effect, where you must weigh the trade-off.

Option	Cost	Effect	Verdict
Costs less, does more	Lower	Higher	Dominant — adopt, no ICER needed
Costs more, does less	Higher	Lower	Dominated — drop from the comparison
Costs more, does more	Higher	Higher	Compute the ICER; judge against a threshold
Costs less, does less	Lower	Lower	ICER of the downgrade — a savings question

The fourth row is the uncomfortable one and it is a real decision. Dropping a costly service frees money; whether that is right depends on what the freed money buys, which is again an ICER.

Eliminate dominated options before computing any ICERs. Leaving them in produces ratios against an option nobody should choose, which makes the whole comparison misleading.

ImpactMojoCost Effectiveness 101www.impactmojo.in

A Common Slip

Don't confuse average with incremental

Reporting a programme's average cost per outcome when the decision is about an upgrade is a frequent and serious error. It can make a poor-value addition look like a bargain.

Rule: if you are comparing two options, you need the ICER between them — the difference in cost over the difference in effect. Never the two averages.

Situation	Correct number
Should we run this programme at all?	Average CER against the status quo
Should we upgrade from B to A?	ICER of A versus B
Which of five options?	Rank by cost, drop dominated, ICERs along the frontier
Is this a good buy in general?	No single number answers this

With more than two options, compute ICERs sequentially along the efficiency frontier — each option against the next cheapest surviving one, not all of them against the cheapest.

Reporting an average where an incremental ratio was needed makes a poor-value addition look like a bargain. It is the most consequential arithmetic error in the field, and it is easy to miss on reading.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Section Six

Cost-Utility & DALYs

ImpactMojoCost Effectiveness 101www.impactmojo.in

The Problem

How do you compare blindness with diarrhoea?

CEA cannot compare a programme that prevents blindness with one that prevents child deaths — the outcomes are in different units. To choose across all of health, you need a common currency of health itself.

That common currency combines two things people care about: how long you live and how well you live. Enter the DALY and the QALY.

Comparison	Possible with
Two malaria programmes	CEA — shared unit
Blindness prevention vs child mortality	CUA only
A treatment that extends life vs one that improves it	CUA — DALYs combine both
Health vs a rural road	CBA only

The third row is what the DALY was built for. Length of life and quality of life are different goods, and a single measure combining them is the only way to trade one against the other explicitly.

Combining them requires deciding how much a year of impaired health is worth relative to a year of full health — which is precisely the disability weight, and precisely what is contested.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Two Measures

QALYs and DALYs: length × quality of life

QALY

Quality-Adjusted Life Year — a year of healthy life. Higher is better; programmes aim to gain QALYs.

DALY

Disability-Adjusted Life Year — a year of healthy life lost. Higher is worse; programmes aim to avert DALYs.

QALYs count health gained; DALYs count health lost. They are mirror images — do not mix them in one calculation.

	QALY	DALY
Counts	Health gained	Health lost
Direction	Higher is better	Lower is better
Weights	Utility, 1 = full health, 0 = death	Disability, 0 = full health, 1 = death
Used mostly by	HTA agencies — NICE and similar	Global burden of disease work, WHO

They are near mirror images and are not interchangeable in a calculation. The weights are elicited differently and from different populations, so a QALY figure cannot be read across as DALYs.

India’s health technology assessment work generally uses QALYs and cost per QALY; global disease-burden comparisons use DALYs. Know which your source used before quoting a threshold at it.

ImpactMojoCost Effectiveness 101www.impactmojo.in

The DALY

What a DALY actually is

DALY

Disability-Adjusted Life Year = Years of Life Lost to early death (YLL) + Years Lived with Disability (YLD). One DALY is one lost year of full health.

DALY = YLL (years of life lost) + YLD (years lived with disability)

Because a DALY is a loss, lower is better and interventions seek to avert DALYs — not accumulate them.

Component	Calculation	Sensitive to
YLL	Reference life expectancy − age at death	Which reference life table is used
YLD	Years with the condition × disability weight	The weight, and duration estimates
DALY	YLL + YLD	Both, plus any age-weighting or discounting

Using a single global reference life expectancy is deliberate. It means a death at 30 counts the same everywhere, rather than counting for less in countries where life expectancy is already low.

Interventions are said to "avert" DALYs because a DALY is a loss. A programme that averts 500 DALYs has prevented 500 years of healthy life from being lost — not added 500 years to anyone.

ImpactMojoCost Effectiveness 101www.impactmojo.in

The Two Halves

Dying early, and living unwell

YLL

Years of Life Lost: a death at 30 against a life expectancy of 70 loses 40 years. Captures the burden of dying early.

YLD

Years Lived with Disability: years spent in ill-health, each weighted by how disabling the condition is. Captures suffering, not just death.

Together they let a single number reflect both a fatal disease and a chronic, disabling but non-fatal one.

Component	What it measures	Driven by
YLL	Years lost to premature death	Age at death against a reference life expectancy
YLD	Years lived in ill health	Duration × disability weight

Because YLL is anchored to a reference life expectancy, a death in childhood counts for far more than a death in old age. That is a deliberate value choice built into the arithmetic.

Earlier versions also age-weighted years and discounted them, which weighted mid-life above infancy and old age. The Global Burden of Disease study dropped both, and the change moved disease rankings substantially.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Disability Weights

Not all years of illness weigh the same

Each condition carries a disability weight from 0 (full health) to 1 (equivalent to death). A year with a mild condition counts as a small fraction of a DALY; a year with a severe one counts as much more.

These weights come from surveys of how people value health states — they are useful, standardised, and genuinely contested. Whose values, and whose lives, the weights reflect is a real ethical question.

Disability weight questions	Why it is contested
Whose judgements?	General-public surveys, not people with the condition
Do people with a condition agree?	Often not — they rate their own state higher
Is a year with disability worth less?	Disability-rights scholars argue the premise is wrong
Are they culturally stable?	Weights vary between populations surveyed

The disability-rights critique is not a technical objection to be refined away. It says the framework assigns lower value to lives lived with disability, and that this is a moral claim disguised as measurement.

You can use DALYs and hold the critique. What you cannot do is present a cost-per-DALY figure as a neutral fact — the weights are contested values, and honest reporting says so.

ImpactMojoCost Effectiveness 101www.impactmojo.in

The CUA Ratio

Cost per DALY averted

Cost-utility analysis divides cost by DALYs averted: cost per DALY averted. Lower means more health bought per rupee, and it works across completely different diseases.

cost / DALYs averted = cost per DALY averted

To compute cost per DALY averted, you need	Usually the weak link
Full programme cost	Overheads and participant costs
Cases averted, causally attributed	The counterfactual
DALYs per case	Disability weights and duration
A time horizon and discount rate	Both frequently unstated

Four estimates multiply into one figure, and the result is quoted as though it were a measurement. Each carries its own uncertainty, and they compound rather than cancel.

This is why a published cost-per-DALY figure should be read as an order of magnitude. Differences of two- or three-fold between studies are common and often methodological.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Orders of Magnitude

Some interventions avert DALYs very cheaply

Cost per DALY averted — illustrative ranking (lower = better)

Illustrative; relative ordering only, not exact figures

Bars show relative ordering only. Bed nets and basic vaccines are commonly cited as among the most cost-effective health buys — but treat any precise '$ per DALY' figure with care.

Commonly cited as low cost per DALY	Caveat
Childhood vaccination	Depends on existing delivery infrastructure
Insecticide-treated bed nets	Depends on local malaria burden
Vitamin A supplementation	Effect size debated in recent trials

Every entry is conditional on its setting, and a ranking strips the conditions out. Bed nets are an excellent buy where malaria is endemic and a poor one where it is not.

Treat any precise per-DALY figure with suspicion, especially one quoted with no date, source or setting. The ordering is more robust than the numbers — and it shifts too.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Thresholds

When is a cost per DALY 'good value'?

A widely used rule of thumb judged an intervention cost-effective if its cost per DALY averted was below one to three times a country's GDP per capita. It is convenient — and increasingly criticised as arbitrary.

WHO has moved away from blanket GDP-based thresholds toward context-specific judgement. A threshold is a value choice dressed as a number — treat it as one input, never a verdict.

Threshold claim	Status
"Below 1× GDP per capita = highly cost-effective"	A rule of thumb, never an empirical finding
"Below 3× GDP per capita = cost-effective"	The same; widely used and widely criticised
"WHO recommends these thresholds"	No longer — WHO moved away from blanket multiples
"The threshold reflects what we can afford"	Only if derived from the actual budget constraint

A GDP-based threshold rises with national income and has no connection to the health budget. Two countries with the same budget per head and different GDPs would judge the same intervention differently.

Empirical work estimating the health actually displaced by spending has generally produced thresholds well below 1× GDP per capita — meaning the conventional rule approves interventions that displace more health than they add.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Section Seven

Time & Discounting

ImpactMojoCost Effectiveness 101www.impactmojo.in

Money Has Timing

A rupee today is worth more than a rupee next year

Costs and benefits arrive at different times. A rupee in hand today can be invested, can meet an urgent need, and is certain — so we value it more than the same rupee a decade away. Discounting makes future and present values comparable.

This is not about inflation. Even with stable prices, sooner is worth more than later.

Flow	Year 0	Year 10	Discounted at 3%
₹1,00,000 cost	₹1,00,000	—	₹1,00,000
₹1,00,000 benefit	—	₹1,00,000	About ₹74,400

Discounting is separate from inflation. These are real rupees in both years; the reduction reflects timing alone, not any change in prices.

Get this distinction right before anything else in the section. Analyses that apply a discount rate to nominal figures double-count, and it is a surprisingly common error in programme budgets.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Why Discount

Time preference and opportunity cost

Time preference: people generally prefer benefits now to benefits later
Opportunity cost of capital: money tied up could have earned a return
Uncertainty: the further ahead, the less sure the benefit will arrive

Discounting is how we honour all three — putting flows that occur at different times on one comparable footing.

Reason to discount	Applies to money?	Applies to health?
Time preference	Yes	Contested
Opportunity cost of capital	Yes	Not directly
Uncertainty about the future	Yes	Yes
Inflation	No — handle separately in real terms	No

Discounting is not inflation adjustment, and conflating the two is a common error. Work in real terms first, then discount; doing both in one rate double-counts.

The middle column is why discounting money is uncontroversial and discounting health is not. Health cannot be invested at a return, so the strongest argument for the rate does not carry across.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Present Value

Bringing the future into today's terms

Present value (PV)

What a future cost or benefit is worth today, after discounting. A benefit of X in year t is worth X ÷ (1 + r)^t now, where r is the discount rate.

PV = future value / (1 + r)^t

Symbol	Meaning	Where mistakes happen
r	Discount rate per year	Real, not nominal
t	Years from the present	Mid-year vs end-year conventions
PV	Present value	Must be applied to costs and benefits alike

Apply the same rate and convention to both sides. Discounting benefits but not costs, or using different rates for each, produces a ratio that means nothing and is easy to do by accident.

For long horizons some analyses use a declining rate, on the argument that uncertainty about the far future justifies weighting it less harshly than constant compounding implies.

ImpactMojoCost Effectiveness 101www.impactmojo.in

See the Decay

How a discount rate erodes future value

Present value of ₹1,000 received in year t (at 3% and 8%)

Illustrative, standard discounting

At 8%, ₹1,000 in 30 years is worth under ₹100 today. The rate you pick dramatically changes how much the future counts.

₹1,000 received in	Worth today at 3%	At 8%
Year 10	About ₹744	About ₹463
Year 20	About ₹554	About ₹215
Year 30	About ₹412	About ₹99

At 8%, a benefit thirty years out is worth a tenth of its face value; at 3% it retains over 40%. The same programme can be excellent or worthless depending on a number chosen in a spreadsheet.

This is why prevention, education and climate work fares badly under high rates: costs are immediate, benefits decades away — exactly what discounting penalises.

ImpactMojoCost Effectiveness 101www.impactmojo.in

The Rate

Choosing the discount rate

Common rates in health and development run around 3% per year, though some analyses use higher. The choice is consequential and partly ethical: a high rate makes long-term benefits almost vanish.

This matters acutely for the climate, for children, and for prevention — programmes whose payoffs land decades away look far worse under a high discount rate. The rate encodes how much we value the future.

Rate	Effect	Typically favours
0%	Future counts equally	Prevention, children, climate
3%	Common convention in health	Balanced
8–12%	Future nearly vanishes	Immediate treatment over prevention

The rate is partly an ethical parameter, not only a financial one. A high rate says that a life saved in forty years counts for a fraction of a life saved now, which is a position rather than a calculation.

Report the base case at 3%, and show the result at 0% and at a higher rate. If the decision flips between them, the finding is that the choice depends on the rate — and that should be said plainly.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Discount Effects Too?

Should we discount health, not just money?

Convention often discounts both costs and health effects (DALYs, QALYs) — otherwise a calculation could justify delaying every benefit forever. But discounting future lives is philosophically uncomfortable.

Whatever you choose, be explicit and consistent, and test how sensitive your result is to the rate — which is exactly what Section 9 is about.

Position	Argument	Problem
Discount health at the same rate as money	Consistency; avoids paradoxes	Devalues future lives
Discount health at a lower rate	Health cannot be invested	Can justify indefinite delay
Do not discount health	All lives count equally	Postponing every programme looks costless

The "indefinite delay" paradox is the standard argument for discounting health: if future health is not discounted, waiting a year is free, and so is waiting forever.

There is no settled resolution. Convention discounts both at the same rate; the defensible practice is to say which you did and show the result under the alternatives.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Time Horizon

How far ahead do you count?

Alongside the rate, you must choose a time horizon — how many years of costs and benefits to include. A prevention programme whose payoffs land over a lifetime looks worthless on a three-year horizon.

Cut the horizon too short and you systematically undervalue prevention and anything slow-burning. State the horizon, and match it to when the real effects actually occur.

Horizon	Systematically favours	Systematically penalises
3 years (a project cycle)	Treatment, immediate delivery	Prevention, education, infrastructure
10 years	Most health programmes	Long-latency benefits
Lifetime	Prevention and early-childhood work	Nothing — but forecasts get shakier

The horizon is usually set by the grant, not by the intervention. A three-year evaluation window on a programme whose payoff lands in adulthood is measuring the wrong period and will report failure.

Choose the horizon from the causal chain and state it. If the funder’s window is shorter, model both and show what the short window omits.

ImpactMojoCost Effectiveness 101www.impactmojo.in

A Caution

Discounting is an assumption, not a fact

The discount rate is the most powerful, least visible assumption in any long-horizon analysis — small changes swing the conclusion.

— a maxim of cost-benefit practice

Always report your discount rate, justify it, and show the result at alternative rates. Hiding it is a red flag in any analysis you read.

Report	Because
The base-case rate	A reader cannot interpret the result without it
Why that rate	Convention, guideline, or funder requirement
Results at alternative rates	Shows whether the conclusion is rate-dependent
The time horizon alongside it	Rate and horizon interact strongly

Rate and horizon must be reported together. A high rate over a long horizon and a low rate over a short one can produce similar present values by entirely different reasoning.

An analysis with a long horizon and no stated discount rate cannot be evaluated at all. Treat the omission as a reason to ask for the model rather than to accept the headline.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Section Eight

Comparing Interventions

ImpactMojoCost Effectiveness 101www.impactmojo.in

The Pay-Off

Why we did all this: to compare

The whole point of standardising costs and effects is to lay interventions side by side and ask where each rupee does most good. Done honestly, the comparison can redirect funding to far higher impact.

But comparison is only as trustworthy as the consistency behind it — same method, same units, same perspective, same discount rate. Compare like with like, or not at all.

For a comparison to be valid, all entries share	Or else
The same method	Denominators are not commensurable
The same perspective	Some entries count participant cost, others do not
The same discount rate and horizon	Long-payoff options are penalised inconsistently
Comparable outcome definitions	"Case detected" and "case cured" are not the same unit
Comparable contexts, or stated adjustment	You are comparing settings, not interventions

Harmonising entries is most of the work in building a league table, and it is the part that gets skipped when ratios are lifted from published papers.

If you cannot harmonise, report the entries with their assumptions attached rather than in a ranked column. A table that looks like a ranking will be read as one.

ImpactMojoCost Effectiveness 101www.impactmojo.in

League Tables

Ranking interventions by cost per outcome

League table

A ranked list of interventions by their cost-effectiveness ratio — cheapest per unit of outcome at the top — used to guide where to spend next.

League tables are powerful and seductive. Their honesty depends entirely on whether the entries were estimated the same way. A table mixing methods is worse than no table.

League table question	What to check
Who built it, and for what decision?	The framing shapes what was included
Were entries estimated the same way?	Method, perspective, horizon, rate
What is the uncertainty on each?	Overlapping ranges make the order meaningless
What is missing from the table?	Unmeasured outcomes count as zero

Ranks imply a precision the underlying estimates rarely support. Where confidence intervals overlap heavily, the difference between third and eighth place may be noise.

The fourth question is the most important and the least asked. Anything nobody measured — dignity, cohesion, resilience — is absent from the table and therefore, in effect, valued at zero.

ImpactMojoCost Effectiveness 101www.impactmojo.in

$ per Outcome

GiveWell-style 'cost per outcome' comparison

Illustrative cost per equivalent outcome across programmes

Illustrative; relative comparison only

Funders like GiveWell rank charities by cost per comparable outcome and direct money to the top. The same logic scales from a global fund to a single grant committee. Figures here are illustrative.

Trustworthy when	Check
Uncertainty shown per entry	Overlapping ranges make the order meaningless

Ranks compress away what would let you disagree. A table of positions and nothing else is an argument, not evidence.

GiveWell publishes its models with the judgement calls visible — the practice to imitate, whatever you make of its conclusions.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Reallocation

Moving money up the league table

If programme A averts an outcome for ₹400 and programme E for ₹7,500, shifting funds from E toward A — where appropriate and feasible — can multiply total impact for the same budget.

But 'shift the money' is rarely so simple: programmes serve different people, places and needs. Cost-effectiveness tells you where to look harder, not always where to cut.

Before reallocating, ask	Because
Do they serve the same people?	Efficiency gains can come from dropping the hardest to reach
Can the efficient one absorb the money?	Absorptive capacity is a real constraint
Does the ratio hold at larger scale?	Marginal cost rises as the easy cases are used up
What is lost when the other closes?	Staff, trust and reach are not instantly rebuilt

"Shift the money up the table" assumes the top entry can scale at the same ratio. That assumption is usually false, and it is exactly what diminishing returns describes.

Reallocation also has transition costs nobody puts in the model: closing a programme destroys relationships and institutional knowledge that took years to build and will not return if the decision is reversed.

ImpactMojoCost Effectiveness 101www.impactmojo.in

A Cautionary Tale

The deworming debate

Mass school deworming became a famous 'best buy' — cheap pills, and one influential study linked treatment to better schooling and later earnings. It topped cost-effectiveness league tables for years.

Then re-analyses and replications questioned the size and durability of those long-run effects. The pills are cheap and safe; the magnitude of the benefit — the denominator — turned out far more uncertain than the rankings implied.

Stage of the deworming debate	What it established
Original trial and follow-ups	Large long-run schooling and earnings effects claimed
Replication and re-analysis	Coding and analytic choices materially changed results
Cochrane-style reviews	Little consistent effect on nutrition, haemoglobin or school performance
Current position	Genuinely contested; cheap enough that some still recommend it

The honest summary is that the effect size is disputed, not that deworming was debunked. Both overstatements circulate, and the second is now as common as the first once was.

What is not disputed is the mechanism of the failure: an intervention topped league tables for years on an effect estimate that could not bear the weight the ranking put on it.

ImpactMojoCost Effectiveness 101www.impactmojo.in

The Lesson

What deworming teaches about league tables

A ranking is only as solid as the effect estimate beneath it
A single influential study can move millions — and may not replicate
Cheap cost (numerator) cannot rescue an uncertain effect (denominator)
Precise-looking ratios can hide enormous uncertainty

The deworming story is not 'CEA is useless' — it is 'never read a league-table rank without its uncertainty'. Which is Section 9.

Lesson from deworming	What to do about it
A ranking is only as solid as its effect estimate	Report the evidence grade beside the ratio
One influential study can move millions	Wait for replication before large reallocation
A cheap numerator cannot rescue an uncertain denominator	Sensitivity-test the effect, not just the cost
Precise ratios hide wide uncertainty	Publish the range, not the point

This is not an argument against cost-effectiveness analysis. It is an argument against reporting a point estimate as settled, which is a reporting failure rather than a methodological one.

The same structure recurs whenever a single strong result drives allocation. Ask what would have to be false for the ranking to reverse, and how confident anyone is that it is not.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Transferability

A ratio from elsewhere may not hold here

Costs, wages, disease burden and delivery systems differ across districts and countries. A cost-effectiveness ratio estimated in Kenya or Bihar will not automatically hold in Odisha or Sindh.

Before borrowing a ratio, ask what would change in your context: input prices, baseline burden, uptake, capacity. Adapt, don't copy.

What differs across settings	Effect on a borrowed ratio
Baseline disease burden	Fewer cases to avert → worse ratio
Wage levels	Staff-intensive programmes cost differently
Existing delivery infrastructure	Marginal cost is far lower where a system exists
Programme quality at scale	Trial-quality delivery rarely survives scale-up

Baseline burden is the largest single driver and the most forgotten. An intervention that averts many cases where prevalence is high averts few where it is low, at similar cost.

Recost the numerator with local prices and re-estimate the denominator with local burden before borrowing any ratio. Transferring a headline figure unadjusted is not evidence use.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Diminishing Returns

The best buy changes as you scale

A top-ranked intervention does not stay top-ranked forever. As you fund the easiest-to-reach first, the cost of reaching each additional person tends to rise — so even the best buy eventually becomes less cost-effective at the margin.

This is why portfolios beat single 'winners': the right answer is usually to fund the best buy up to a point, then move down the table — not to pour everything into one line.

Coverage reached	Who remains	Marginal cost
First 50%	Accessible, willing, near a facility	Low
Next 30%	Further away, harder to persuade	Rising
Last 20%	Remote, mistrustful, hardest to serve	Often several times the average

Average cost-effectiveness at 50% coverage says nothing about the cost of the next 10%. Decisions about scale-up need the marginal ratio, and it is almost never the one reported.

This is also where efficiency and equity collide most directly: the last 20% is disproportionately the poorest and most remote, and they are exactly whom the marginal ratio argues against reaching.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Good Practice

Comparing interventions honestly

Use the same method, units, perspective and discount rate for every entry
Show the uncertainty around each ratio, not just the point estimate
State the context each estimate came from
Pair the ranking with equity and feasibility, never efficiency alone

Practice	Prevents
Same method, units, perspective, rate for all entries	Comparing incommensurable numbers
Show uncertainty per entry	A spurious ordering
State each estimate’s context	Transferring a ratio that does not hold
Pair the ranking with equity data	Efficiency achieved by not reaching the hardest
Report what could not be quantified	Unmeasured benefits valued at zero

These five take a page and answer most of the objections in the chapter. None requires additional data collection — only that what was already decided is written down.

If a table cannot satisfy the first row, present the entries as separate case studies rather than as a ranking. The visual form of a ranked column makes a claim the data does not support.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Section Nine

Uncertainty & Sensitivity

ImpactMojoCost Effectiveness 101www.impactmojo.in

The Honest Truth

Every ratio is an estimate, not a fact

A cost-effectiveness ratio is built from uncertain inputs — uncertain costs, uncertain effects, an assumed discount rate. The single number that results looks precise but is not. Sensitivity analysis asks how much it could move.

A point estimate with no uncertainty attached is not a finding — it is an illusion of precision. Demand the range.

Uncertain input	Typical range	Usually the biggest driver?
Effect size	Wide — from the confidence interval	Yes, almost always
Unit costs	Moderate	Sometimes
Discount rate	A stated set of alternatives	Yes, for long horizons
Coverage achieved	Wide at scale-up	Often

A point estimate with no uncertainty attached is not a result; it is a claim. Every input above is estimated, and the ratio inherits all of their uncertainty at once.

Sensitivity analysis is also a research-prioritisation tool: whichever input swings the answer most is where better evidence would be worth buying.

ImpactMojoCost Effectiveness 101www.impactmojo.in

One-Way

One-way sensitivity analysis

Vary one input at a time across a plausible range, holding the rest fixed, and watch how the result moves. It reveals which assumptions the answer hinges on.

If halving the assumed effect doubles the cost per outcome and flips your decision, you have found the input that deserves the most scrutiny — and better data.

Vary	Plausible range from	If the decision flips
Effect size	The confidence interval	Buy better effect evidence before deciding
Unit costs	Observed variation across sites	Cost the specific setting properly
Coverage achieved	Pilot versus realistic scale-up	The scale-up assumption is the decision
Discount rate	0%, base case, higher	Say the conclusion is rate-dependent

Use real ranges, not arbitrary ±20%. A range taken from the confidence interval or from observed site variation is defensible; a round percentage chosen for tidiness is decoration.

The right-hand column is the point of the exercise. Sensitivity analysis is a decision tool, not a robustness ritual: it tells you what to do next when the answer is not stable.

ImpactMojoCost Effectiveness 101www.impactmojo.in

The Tornado

Tornado diagram: what moves the answer most

One-way sensitivity: swing in cost per outcome by input (illustrative)

Illustrative

Sorted widest-to-narrowest, the bars form a 'tornado'. Here the effect size dominates — so that is where to invest in better evidence. Illustrative.

Tornado bar	Reading it
Widest bar at the top	The input that drives the answer
Bar crossing the threshold line	That input alone can flip the decision
All bars narrow	The conclusion is robust — report that

Effect size sits at the top of most tornado diagrams. Better outcome evidence usually buys more decision quality than better cost data.

A tornado where no bar crosses the threshold is a strong result. Robustness is a finding, and it is under-reported because it feels like the absence of one.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Probabilistic

Probabilistic sensitivity analysis (PSA)

Probabilistic sensitivity analysis

Assign each uncertain input a probability distribution, then simulate the model thousands of times to produce a distribution of cost-effectiveness results, not a single number.

PSA answers the question decision-makers really have: not 'what is the ratio?' but 'how likely is this option to be the best buy, given everything we don't know?'

	One-way	Probabilistic (PSA)
Varies	One input at a time	All inputs together, from distributions
Output	A range per input; a tornado diagram	A distribution of results
Answers	Which input matters most?	What is the probability this is good value?
Effort	A spreadsheet	A simulation, thousands of runs

PSA answers the question decision-makers actually have. Not "what is the ratio" but "how likely is it that this is worth funding at our threshold" — which one-way analysis cannot deliver.

One-way analysis understates total uncertainty because inputs vary together in reality. Both are worth running: one-way to find the drivers, PSA to size the overall risk.

ImpactMojoCost Effectiveness 101www.impactmojo.in

The CE Plane

The cost-effectiveness plane

Plot extra cost (vertical) against extra effect (horizontal) relative to the comparator. The origin is the comparator; an option's position decides whether it is an easy yes, an easy no, or a trade-off.

Lower-right

More effect, less cost — dominant. Adopt it.

Upper-left

Less effect, more cost — dominated. Reject it.

Quadrant	Cost	Effect	Decision
Lower right	Less	More	Dominant — adopt
Upper left	More	Less	Dominated — reject
Upper right	More	More	Depends on the threshold
Lower left	Less	Less	A savings trade-off — depends on the threshold

Most real comparisons land in the upper right, which is why the threshold matters so much: it is the line on the plane that separates yes from no, and it is a value judgement.

When a PSA cloud straddles more than one quadrant, no single ICER describes it. Report the proportion of simulations falling below the threshold instead.

ImpactMojoCost Effectiveness 101www.impactmojo.in

See the Plane

An ICER on the cost-effectiveness plane

Simulated results vs a comparator (origin = comparator)

Illustrative PSA cloud

The scatter (one dot per simulation) shows how wide the uncertainty is. A line from the origin is the willingness-to-pay threshold: dots below it are 'worth it'. Illustrative.

PSA cloud shape	Interpretation
Tight cluster below the threshold line	Confident good value
Wide cloud straddling the line	Genuinely uncertain — report the probability
Cloud spanning quadrants	Even the direction of effect is unclear

The useful output is a probability, not a ratio: "at a threshold of ₹X, this is good value in 78% of simulations" is a statement a committee can act on.

A cost-effectiveness acceptability curve plots that probability across a range of thresholds, which removes the need to commit to one and is often the single most useful chart in an analysis.

ImpactMojoCost Effectiveness 101www.impactmojo.in

The Takeaway

Why point estimates mislead

Two interventions can have the same headline ratio while one is a near-certain good buy and the other a coin-toss. The point estimate alone cannot tell them apart; the spread can.

Treat any single '₹X per outcome' as the centre of a range, never the truth. The width of that range is often the most decision-relevant fact in the whole analysis.

	Intervention X	Intervention Y
Point estimate	₹5,000 per outcome	₹5,000 per outcome
Plausible range	₹4,500–5,500	₹900–40,000
Probability below a ₹10,000 threshold	Near certain	Roughly a coin toss

Identical headline ratios, entirely different decisions. The point estimate is the least informative number in the analysis and the only one that usually gets quoted.

When a range is wide, the useful action is often to buy better evidence rather than to decide. Sensitivity analysis tells you which piece of evidence would narrow it most.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Section Ten

Limits & Equity

ImpactMojoCost Effectiveness 101www.impactmojo.in

Not the Whole Story

Efficiency is one value among several

Cost-effectiveness measures efficiency — outcomes per rupee. But societies also care about fairness, rights, dignity and reaching the worst-off. A perfectly efficient choice can be deeply unjust.

Not everything that counts can be counted, and not everything that can be counted counts.

— commonly attributed to William Bruce Cameron

Value	Can CEA weigh it?
Efficiency — outcomes per rupee	Yes — this is what it measures
Fairness in who receives	Only if you add equity weights
Rights and legal obligation	No — obligations do not depend on ratios
Dignity in how people are treated	No — process is invisible to the ratio
Political feasibility	No

A perfectly efficient allocation can be unlawful. Where a right exists — to emergency care, to school — the cost per outcome does not determine whether it must be provided.

The useful posture is that CEA settles one question well and is silent on the others. Trouble starts when silence is read as the others having no weight.

ImpactMojoCost Effectiveness 101www.impactmojo.in

What CEA Ignores

The blind spots of cost per outcome

Distribution: a total of outcomes hides who gets them
Rights & dignity: some things are owed regardless of efficiency
Process: how people are treated, not just the result
The hard-to-reach: serving them is costlier — and easily deprioritised

Blind spot	Concrete example
Distribution	A total of outcomes achieved entirely in accessible districts
Rights and dignity	Emergency care owed regardless of the ratio
Process	Two programmes, same outcome, one treats people badly
The hard to reach	Higher cost per outcome; systematically deprioritised

Each blind spot is structural, not an oversight to be fixed with better data. The ratio is a total divided by a total, and totals cannot express distribution.

Reporting outcomes by group alongside the aggregate ratio costs almost nothing and closes the first and fourth rows. It is the cheapest equity correction available.

ImpactMojoCost Effectiveness 101www.impactmojo.in

The Equity Trap

Cheapest-to-reach is not the same as fairest

Reaching a remote Adivasi hamlet or a person with disability often costs far more per outcome than reaching an accessible town. Pure cost-effectiveness quietly pushes resources away from exactly those who need them most.

Left unchecked, 'value for money' can entrench the very inequities development is meant to undo. This is the central ethical tension of the whole field.

Group	Cost per outcome	Need
Accessible urban households	Lowest	Lower
Rural, road-connected	Moderate	Moderate
Remote Adivasi hamlets	Highest	Highest
People with disabilities	Highest	Highest

Cost per outcome and need run in opposite directions, which means unweighted cost-effectiveness is not neutral about equity — it has a systematic direction, and it points away from the worst-off.

This is not an argument against measuring efficiency. It is an argument for never presenting the ranking without the distribution alongside it.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Equity Weighting

Building fairness into the numbers

One response is equity weighting: counting an outcome for a worse-off person as worth more than the same outcome for someone advantaged, so the analysis rewards reaching the marginalised.

It makes the value judgement explicit rather than pretending efficiency is neutral — but choosing the weights is itself a political and ethical act, not a technical one.

Equity weighting decision	What it requires you to state
Who counts as worse-off	Income, caste, disability, geography — on what basis
How much extra weight	A number, chosen and defended
Whose judgement sets it	A committee, a public process, or an analyst alone
Sensitivity to the weight	Whether the ranking survives a different weight

Equity weighting does not remove the value judgement; it makes it visible and arguable. That is its advantage over unweighted analysis, where the judgement is implicit and cannot be challenged.

Distributional cost-effectiveness analysis formalises this by reporting outcomes by group rather than in a single total. It is more work and it answers "for whom" as well as "how much".

ImpactMojoCost Effectiveness 101www.impactmojo.in

Whose Values?

Numbers carry hidden value judgements

Disability weights embed whose view of a 'good life'
Discount rates embed how much we value future generations
Thresholds embed what a life or a healthy year is 'worth'
The choice of outcome embeds what we decided to count at all

Every one of these is a values choice wearing the costume of arithmetic. Data-literate practice means reading the choices, not just the digits.

Technical choice	The value it encodes
Disability weights	Whose view of a life worth living
Discount rate	How much future people count
Threshold	What a year of health is worth
Choice of outcome	What was worth counting at all
Perspective	Whose costs are real

None of these is derivable from data. Each is a position, defended by convention, and each can flip a ranking — which is why an analysis that reports results without reporting its assumptions is not checkable.

The remedy is not to abandon the method. It is to publish the assumptions prominently and show how the conclusion moves when a reader substitutes their own.

ImpactMojoCost Effectiveness 101www.impactmojo.in

The Counted & the Counted-Out

What gets measured gets funded

Cost-effectiveness can only weigh outcomes that someone chose to measure. Benefits that are hard to quantify — dignity, social cohesion, women's voice, ecological resilience — risk being treated as if they were worth zero.

'Not measured' is not the same as 'not valuable'. An analysis that silently drops the unmeasurable will systematically under-fund it. Name what your ratio leaves out.

Hard to quantify	Treated by the ratio as
Dignity in the encounter	Worth zero
Women’s voice and agency	Worth zero
Social cohesion	Worth zero
Ecological resilience	Worth zero
Institutional capacity built	Worth zero

Omission is not neutrality. A benefit left out of the denominator is treated in the arithmetic exactly as if it had been measured and found to be nothing.

List what you could not quantify, explicitly, next to the ratio. A named omission can be weighed by a decision-maker; an unnamed one has already been decided.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Use It Well

A servant, not a master

Cost-effectiveness is most useful as a discipline of attention — forcing you to be explicit about costs, outcomes and trade-offs — and most dangerous when it becomes the sole arbiter of who gets help.

Best practice: present cost-effectiveness alongside equity, rights and feasibility, and let deliberation — not a single ratio — make the final call.

CEA as a discipline of attention forces you to	Which helps even if the ratio is rough
Name the outcome you are buying	Ends vague goals
Cost every ingredient	Reveals hidden and donated resources
Name the comparator	Makes the real alternative explicit
State assumptions	Converts disagreement into something arguable

Most of the value arrives before the division. Teams that go through this process usually change the programme design before they ever compute a ratio.

Present the result alongside equity, rights and feasibility, and let deliberation weigh them. A ratio that arrives as a verdict rather than an input has exceeded what the method can support.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Section Eleven

Practice & Tools

ImpactMojoCost Effectiveness 101www.impactmojo.in

What You Need

The data a cost-effectiveness study needs

Cost data: a full ingredients inventory, including overheads and participant costs
Effect data: a credible, causal estimate of the outcome
The comparator: what you are measuring against
Assumptions: discount rate, time horizon, perspective — stated openly

Most disputes over a result trace back to one of these four. Pin them down before you compute anything.

Data element	Where disputes usually arise
Cost inventory	What was left out — overheads, donated inputs, participant time
Effect estimate	The counterfactual, and whether it transfers
Comparator	Whether "nothing" or current practice was used
Discount rate and horizon	Whether prevention was fairly treated
Perspective	Whose costs were counted

Most arguments about a cost-effectiveness result are arguments about these five choices, not about the arithmetic. Stating all five up front converts a dispute about numbers into a dispute about assumptions, which is more tractable.

Write them as a table at the front of any analysis you produce. It takes ten minutes and it is the single most useful thing you can do for a reader.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Build the Model

Spreadsheets are enough to start

Most cost-effectiveness analyses begin life in a spreadsheet: costs in one sheet, effects in another, the ratio computed transparently, and a tab for sensitivity analysis. Clarity beats sophistication.

Keep every assumption in a labelled, single cell so a reviewer can change it and watch the result move. An auditable spreadsheet is worth more than a black-box model.

Model discipline	Why
One assumption per labelled cell	A reviewer can change it and watch the result move
No hard-coded numbers in formulas	Hidden constants are the commonest source of error
Costs, effects and results on separate sheets	Keeps the logic followable
A sensitivity tab from the start	Retrofitting one means rebuilding the model
Every source cited in the cell comment	Provenance survives staff turnover

Clarity beats sophistication in this field. A transparent spreadsheet a funder can interrogate is worth more than an elegant model nobody outside the team can open.

Build the sensitivity tab first, before the results are known. Doing it afterwards invites choosing the ranges that keep the answer where you already put it.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Good Practice

Habits of a trustworthy analysis

State the perspective, comparator, horizon and discount rate explicitly
Use the ingredients method — cost everything, including the hidden
Report the ICER for choices between options, not just averages
Always run sensitivity analysis and show the range
Present equity alongside efficiency

Habit	Takes
State perspective, comparator, horizon, rate	Four lines
Use the ingredients method	A day of inventory work
Report ICERs for choices between options	One extra column
Run and show sensitivity analysis	A spreadsheet tab
Present equity alongside efficiency	Disaggregating what you already have

None of these requires new data collection. They are disclosure and arithmetic discipline, which is why their absence is a signal about the analysis rather than about the budget behind it.

The CHEERS reporting statement covers all five and more. Running it over your draft before circulation catches most of what a reviewer would otherwise raise.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Watch-Outs

Red flags when you read a CEA

A single ratio with no uncertainty or sensitivity analysis
An average ratio where an incremental one was needed
No stated perspective or discount rate
An effect size from one study, treated as settled
A ranking that ignores who is reached and who is missed

Red flag	What it usually conceals
A single ratio, no range	A wide uncertainty the author would rather not show
An average where an ICER was needed	A poor-value upgrade presented as a bargain
No stated perspective	Participant costs excluded
One study, treated as settled	An effect estimate that has not replicated
A ranking with no distributional data	Efficiency gained by not reaching the hardest

These are reading instructions, not accusations. Any one of them can be innocent; what they share is that the missing information is exactly what would let you disagree with the conclusion.

Ask for the model, not the summary. An analyst who will share the spreadsheet is making a different claim from one who will only share the headline figure.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Reading List

Where to go deeper

Resource	What it offers
Drummond et al., Methods for the Economic Evaluation of Health Care Programmes	The standard textbook of the field
WHO-CHOICE	WHO's database and tools for comparing health interventions
Disease Control Priorities (DCP3)	Cost-effectiveness evidence across diseases for low/middle-income settings
GiveWell & the Disease Control Priorities reviews	Worked 'cost per outcome' analyses you can learn from
iDSI Reference Case	A shared standard for credible economic evaluation

Source	Best for
Drummond et al.	The standard reference; methods in full
WHO-CHOICE	Comparable estimates across health interventions
Disease Control Priorities (DCP3)	Cost-effectiveness evidence for low- and middle-income settings
CHEERS reporting statement	A checklist for what an analysis must disclose
GiveWell’s published models	Worked examples with the assumptions exposed

CHEERS is the most immediately useful item here for a practitioner. It is a reporting checklist, so you can run it over an analysis you are reading as easily as one you are writing.

GiveWell publishes its spreadsheets, including the judgement calls. Whatever you make of its conclusions, the models are among the clearest available demonstrations of how these choices are actually made.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Connect the Course

Pair this with the rest of the 101 Series

Cost-effectiveness sits inside a wider toolkit. It is strongest when fed by good measurement, sound evaluation and honest data.

Pair this deck with ImpactMojo's Data Literacy, Impact Evaluation and Theory of Change 101 courses for the full picture.

ImpactMojoCost Effectiveness 101www.impactmojo.in

The Takeaways

If you remember five things

Most good per rupee — because budgets are finite and choices have opportunity costs
Match the method — CEA (natural units), CUA (DALYs), CBA (money)
Use the ICER — the incremental, not average, ratio when comparing options
Show the uncertainty — a point estimate alone is an illusion of precision
Efficiency is not the only value — weigh equity, rights and dignity too

Takeaway	The mistake it prevents
Most good per rupee	Treating "cheap" as "cost-effective"
Match the method	Comparing a CEA ratio with a CBA one
Use the ICER	Justifying an upgrade with an average
Show the uncertainty	A point estimate read as a fact
Efficiency is not the only value	A ranking that quietly abandons the hardest to reach

Four of the five are about how a result is reported rather than how it is computed. That is where most of the damage in this field is done, and where most of it can be prevented.

If you take one habit from the deck, take the comparator. Asking "compared with what?" of any ratio you are shown will expose more weak analysis than any other single question.

ImpactMojoCost Effectiveness 101www.impactmojo.in

Cost Effectiveness 101 · Complete

Now ask: the most
good per rupee?

You don't need to be a health economist — you need to hold cost and outcome together, demand the increment and the uncertainty, and never let a ratio outrank equity. Explore the rest of the ImpactMojo 101 Series, free forever.

More 101 Courses Explore ImpactMojo Dataverse

Continue your learning

Impact Evaluation Studio MEL Flagship Opportunity Cost Game

CC BY-NC-ND 4.0·Free Forever·ImpactMojo 101 Series