Health Policy Evaluation

Health policy evaluation is the branch of health economics that tries to answer a deceptively simple question: when a government, insurer, or health system spends money on care, does it buy health worth the cost, and who gets it?

7 frameworks, from 1950 to today · Health Economics · Economics

still a working view ended · each leads to its entry below

Overview

Health policy evaluation is the branch of health economics that tries to answer a deceptively simple question: when a government, insurer, or health system spends money on care, does it buy health worth the cost, and who gets it? The stakes are unusually high. Health budgets are finite, and every dollar spent on one treatment is a dollar unavailable for another. Evaluation supplies the evidence that lets decision-makers choose between competing uses of scarce resources—and increasingly, between competing claims about fairness.

The field's history is a sequence of tools built to answer that question under real-world constraints. Each tool emerged because the previous one could not handle a problem: how to compare treatments with different outcomes, how to know whether a policy actually caused the change it seemed to cause, how to model decisions before the data exist, and how to ask who benefits and who pays.

From cost accounting to cost-benefit analysis

The earliest systematic evaluations of health programmes grew out of public health and welfare economics in the mid-twentieth century. Governments funding large programmes—vaccination campaigns, tuberculosis control, hospital construction—wanted to know whether the money was well spent. The first tool was cost-benefit analysis (CBA), which values both costs and outcomes in money. Its appeal was comparability: if a programme's benefits exceed its costs in dollar terms, it is worth doing, and it can be ranked against any other public investment.

The classic application was in water and sanitation. In the 1950s and 1960s, economists evaluating water supply and sewerage projects in low-income countries valued the health gains—reduced typhoid, cholera, and diarrhoeal disease—by the economic output saved from fewer deaths and fewer sick days. The method worked where outcomes could plausibly be monetised as productivity. It struggled where they could not. What is the dollar value of a year of life lived with dignity, or of a child's survival that never enters the labour market? CBA's supporters found its discipline convincing; its critics found the monetisation of health ethically and practically strained.

The quality-adjusted life year and the cost-effectiveness turn

The decisive break came in the late 1960s and 1970s, when health economists began measuring outcomes in health itself rather than money. The key innovation was the quality-adjusted life year (QALY): a single index combining length of life with its quality, where one year in perfect health equals 1 and death equals 0, with states in between weighted by how people value them. The idea was developed in the late 1960s and formalised in the 1970s by researchers including Milton Weinstein and William Stason, who used it to compare treatments for hypertension and other conditions. The QALY made it possible to compare a cancer drug, a hip replacement, and a screening programme on one scale.

This gave rise to cost-effectiveness analysis (CEA) and its close relative cost-utility analysis (CUA), which express results as cost per unit of health gained—cost per life-year saved, or cost per QALY12. The practical payoff was a threshold: if a health system is willing to pay, say, $50,000 per QALY, then interventions below that line are good value and those above are not. The approach was institutionalised most visibly in the United Kingdom's National Institute for Health and Care Excellence (NICE), founded in 1999, which used cost-per-QALY thresholds to decide which drugs the National Health Service would fund. Similar bodies followed in Australia, Canada, and elsewhere.

The method's supporters found it transparent and comparable. Its critics raised two persistent objections. First, the QALY assumes that a year of life is worth the same regardless of who receives it, which conflicts with the common intuition that saving a young person's life matters more than extending an elderly person's by the same amount. Second, the weights used to value health states are contested—whose preferences count, and how are they elicited? These debates shaped the next generation of methods, including distributional cost-effectiveness analysis, which explicitly incorporates equity concerns into the cost-per-QALY framework rather than treating all QALYs as equal13.

Randomised policy trials: the experimental turn

Cost-effectiveness analysis tells you whether an intervention is good value if it works. The prior question—does it work?—was long answered by observational comparison, which is vulnerable to confounding: sicker patients may receive a treatment precisely because they are sicker, making it look ineffective. The randomised controlled trial (RCT) solved this for clinical interventions by randomly assigning patients to treatment or control, so that the groups differ only by chance. The RCT's rise in medicine is well known; its extension to policy questions is the distinctive contribution of health policy evaluation.

The landmark case was the RAND Health Insurance Experiment (1974–1982), which randomly assigned families to different levels of cost-sharing and measured their health care use and health outcomes. It showed that higher cost-sharing reduced spending but, for most people, did not harm health—a finding that reshaped insurance design debates for decades. Randomisation was later applied to policy questions directly: the Oregon Health Insurance Experiment (2008 onward) used a lottery to assign Medicaid coverage among low-income adults, allowing researchers to estimate the causal effects of insurance on utilisation, financial strain, and health. Randomised policy trials are expensive and sometimes ethically fraught, but they provide the strongest evidence available when well executed.

Quasi-experimental causal inference

Randomisation is not always possible or ethical. The response was a set of quasi-experimental methods that exploit natural variation to approximate random assignment. The most influential is difference-in-differences, which compares the change in outcomes in a group affected by a policy to the change in an unaffected comparison group, netting out common trends. A classic application was the evaluation of Medicaid expansions in the United States in the 1980s and 1990s, where researchers compared states that expanded eligibility to those that did not. Other tools include instrumental variables (using a factor that affects treatment but not outcomes directly), regression discontinuity (exploiting thresholds, such as an age cutoff for eligibility), and synthetic control (constructing a weighted comparison group from untreated units).

These methods transformed the field's evidentiary standards. Where earlier evaluations often relied on before-and-after comparisons that could not separate a policy's effect from secular trends, quasi-experimental designs made causal claims credible without randomisation. The approach's supporters found it pragmatic and widely applicable; its critics noted that its validity depends on assumptions—such as parallel trends—that cannot always be tested.

Structural modelling: evaluating policies that do not yet exist

All the methods above evaluate something that happened or was tried. But policymakers often need to know what would happen under a policy that has never been implemented. Structural modelling addresses this by building an explicit mathematical model of behaviour—how individuals respond to prices, insurance, and health states—and then simulating the effects of alternative policies. The approach descends from the demand-for-health model of Michael Grossman (1972), which treated health as a capital stock that people invest in, and from the labour-supply and tax-policy models of public economics.

A prominent example is the work of Jonathan Gruber and colleagues on health insurance mandates, which used structural models to predict how employer mandates would affect wages, employment, and coverage before such policies were enacted. Structural models are demanding: they require assumptions about preferences and market structure that are hard to validate. Their supporters argue that only structural models can answer counterfactual questions and incorporate behavioural responses that reduced-form methods miss. Their critics argue that the assumptions are doing too much work and that reduced-form evidence is more credible where it exists. The two approaches now coexist, often complementarily.

Distributional and equity analysis

The most recent turn asks not just whether a policy is efficient but who gains and who loses. Distributional cost-effectiveness analysis extends the cost-per-QALY framework by tracking how health gains and costs are distributed across groups—by income, age, ethnicity, or region—and by applying equity weights that reflect societal preferences for reducing health inequality. The approach was formalised in the 2010s by researchers including Richard Cookson and colleagues, building on earlier work on equity in health economics.

The motivation is practical. A policy that improves average health but widens the gap between rich and poor may be unacceptable even if it is cost-effective. Distributional analysis makes that trade-off explicit rather than implicit. It remains methodologically unsettled—how to weight equity, whose preferences to use, and how to handle the fact that health and income are correlated—but it has become a standard component of health technology assessment in several countries.

The present

Today's health policy evaluation draws on all of these traditions. Cost-effectiveness and cost-utility analysis remain the workhorses of reimbursement decisions, institutionalised in bodies like NICE and its counterparts4. Randomised policy trials and quasi-experimental methods supply the causal evidence that feeds those analyses. Structural models answer counterfactual questions where experiments are impossible. Distributional analysis increasingly accompanies standard evaluations, reflecting a broad shift from asking only "is it worth it?" to asking "worth it for whom?"

The open questions are substantial. How should health systems value outcomes that are hard to capture in QALYs—such as dignity, caregiver burden, or the option value of future treatment? How should equity weights be set, and by whom? Can structural models be validated well enough to trust their counterfactuals? And as health budgets face pressure from ageing populations and expensive new technologies, the field's central question—whether spending buys health worth the cost—becomes more urgent, not less.

References

  1. Cost-effectiveness analysis, Wikipedia
  2. Incremental cost-effectiveness ratio, Wikipedia
  3. Distributional cost-effectiveness analysis, Wikipedia
  4. National Institute for Health and Care Excellence, Wikipedia

The frameworks of Health Policy Evaluation

Each framework is a way of working the field has used: when it held, and the question it answers.

  1. 1950-1985

    Cost-benefit analysis

    Is a programme's benefit worth its cost in money?

    Fly to its tower in Nousfarer

  2. 1970-Present

    Cost-effectiveness and cost-utility analysis

    How much health does a treatment buy per dollar spent?

    Fly to its tower in Nousfarer

  3. 1972-Present

    Structural estimation

    What would happen if we tried a new policy?

    Fly to its tower in Nousfarer

  4. 1974-Present

    Randomised policy trials

    Does a policy cause the outcomes it seems to?

    Fly to its tower in Nousfarer

  5. 1985-Present

    Quasi-experiment

    Can we infer cause without an experiment?

    Fly to its tower in Nousfarer

  6. 1995-Present
  7. 2010-Present

    Distributional cost-effectiveness analysis

    How can equity be weighed against efficiency in health spending?

    Fly to its tower in Nousfarer

How the ideas connect

Each arrow runs from an earlier idea to the later one it shaped. The list below says what passed between them.

Cost-benefit analysisCost-effectiveness and cost-utility a…Structural estimationRandomised policy trialsQuasi-experimentDistributional and equity analysisDistributional cost-effectiveness ana…

Beyond the field

Where Health Policy Evaluation's ideas came from in other fields, and what they led to there: the strongest 12 of 48 links.

A city in Nousfarer: a red-carpeted stair climbing terraces to a crowned tower on an acropolis, more towers round it, forest and snowy mountains beyond.
A city in Nousfarer, the game played in the browser.

Health Policy Evaluation is a planet in Nousfarer

Its city has a tower for each of these 7 frameworks. Land, walk in, and climb one concept at a time.

Land on Health Policy Evaluation

Related fields

Fields its ideas are linked to

More in Health Economics

See where Health Policy Evaluation sits on the Map of Human Knowledge