Equity Analysis

How equity analysis evaluates the distribution of health and resources

Equity analysis examines how health, healthcare access, outcomes, costs and financial burdens are distributed across people and groups and whether those differences are considered unfair or avoidable. This page explains how equity questions are defined, how distributions are measured, and how explicit value judgements connect observed inequalities to conclusions about fairness.

Equity is not established by showing that everyone receives the same resources or achieves the same outcome. People can have different needs, starting positions and barriers, so fair treatment may require different levels or forms of support.

Separating inequality from inequity

A health inequality is an observed difference in health or healthcare between people or groups. A health inequity is a difference judged to be unfair, avoidable or unjust according to an explicit ethical or policy standard.

Data can show that a difference exists and describe its size, but data alone cannot determine whether the difference is fair. An equity analysis should therefore distinguish empirical findings from the values used to interpret them.

Questions that support this distinction include:

• Which difference is being measured?
• Between whom is the comparison made?
• Is the difference avoidable or remediable?
• Does it reflect healthcare need, preference, structural disadvantage or discrimination?
• Which principle of fairness is being applied?
• Who participated in defining the principle and interpreting the result?

The same distribution may be judged differently under different equity principles. Those principles should be made visible rather than hidden within a summary score.

Horizontal and vertical equity

Horizontal equity generally means treating people with equal healthcare needs equally. Vertical equity means treating people with different needs differently in ways that respond to those differences.

For example, equal waiting times for patients with similar clinical urgency may support horizontal equity. Faster treatment for patients with more urgent needs may support vertical equity.

Equal resource allocation across regions may appear fair under a simple equality rule but may be inequitable when regions differ in population health, deprivation, rurality or service-delivery costs. Equity analysis should therefore relate resources and outcomes to relevant need.

Deciding what distribution to examine

Equity analysis can focus on health, access, service use, quality, financial burden or the consequences of a policy. The selected outcome should correspond to the decision rather than being chosen only because data are readily available.

Possible distributions include:

• Health status and life expectancy.
• Disease incidence and prevalence.
• Health-related quality of life.
• Access to prevention, diagnosis and treatment.
• Waiting times and travel burden.
• Healthcare utilisation relative to need.
• Quality and safety of care.
• Health gains produced by an intervention.
• Public expenditure and resource allocation.
• Out-of-pocket payments and insurance contributions.
• Catastrophic or impoverishing health expenditure.
• Patient, family and caregiver time.

Health equity, healthcare equity and financing equity are related but not identical. A policy can improve access without reducing health inequality, or improve average health while increasing the financial burden on a disadvantaged group.

Identifying relevant population groups

Equity analysis requires prespecified characteristics across which distributions will be examined. Group selection should reflect plausible disadvantage, the decision context and the pathways through which inequity may arise.

Relevant characteristics may include:

• Income, wealth or socioeconomic position.
• Education and employment.
• Place of residence, including rurality and neighbourhood deprivation.
• Race, ethnicity, caste or Indigenous identity.
• Sex, gender and sexual orientation.
• Age and life stage.
• Disability.
• Migration, language or citizenship status.
• Religion or culture.
• Housing or institutional status.
• Insurance status.
• Clinical need, severity or comorbidity.

Characteristics can interact. Examining each one separately may conceal disadvantage experienced at their intersection, such as the combined effects of rural residence, disability and low income.

Small sample sizes can make intersectional analysis difficult, but lack of precision should be reported rather than treated as evidence that no difference exists. Privacy and the risk of identifying individuals must also be considered.

Mapping the pathway that produces inequity

An observed gap does not reveal its cause. Equity analysis should examine the pathway through which social position, institutions and healthcare processes influence access, experience and outcomes.

Potential mechanisms include:

• Differences in exposure to health risks.
• Differences in disease onset and severity.
• Financial, geographical or informational barriers.
• Discrimination, stigma or lack of cultural safety.
• Differences in referral, diagnosis or treatment.
• Digital exclusion.
• Variation in service quality.
• Differences in treatment adherence caused by affordability or working conditions.
• Unequal capacity to benefit from an intervention.
• Broader determinants such as housing, education and employment.

The mechanism matters because interventions aimed at the wrong stage may leave the inequity unchanged. Expanding appointments, for example, may not improve access for people who cannot afford transport or unpaid time away from work.

Measuring absolute and relative inequality

Absolute and relative measures describe different aspects of a distribution. Both should be considered when the conclusion could change according to the scale used.

Suppose an adverse outcome occurs in proportion (p_D) of a disadvantaged group and (p_A) of an advantaged group. The absolute inequality is:

$$ Absolute\ difference = p_D - p_A $$

The relative inequality is:

$$ Relative\ ratio = \frac{p_D}{p_A} $$

If an outcome falls from 20% to 10% in one group and from 10% to 4% in another, both groups improve. The absolute gap falls from 10 to 6 percentage points, while the relative ratio rises from 2.0 to 2.5. Reporting only one measure could therefore support an incomplete account of the change.

Measuring ordered socioeconomic inequality

When population groups have a meaningful socioeconomic ranking, analysis can examine whether health or healthcare is concentrated among more or less advantaged people. Measures such as the concentration curve and concentration index use information across the distribution rather than comparing only two groups.

These measures require careful interpretation because their value can depend on whether the outcome is health or ill health, whether it is bounded, and how socioeconomic rank is defined. A summary index should be accompanied by group-level results so the underlying distribution remains visible.

Regression-based measures such as the slope index of inequality and relative index of inequality can estimate absolute and relative differences across an ordered social distribution. Model specification, population composition and uncertainty should be reported.

Adjusting healthcare use for need

Comparing raw service use can be misleading because groups may have different healthcare needs. An equity assessment of utilisation should distinguish variation explained by legitimate need from variation associated with socioeconomic or other non-need characteristics.

Need indicators may include:

• Age and sex when clinically relevant.
• Disease status and severity.
• Comorbidity.
• Functional limitation.
• Predicted capacity to benefit.

The choice of need variables contains assumptions. Adjusting for a variable that is itself a consequence of inequity can remove part of the disparity the analysis is intended to reveal.

Standardisation or regression can estimate use expected on the basis of need and compare it with observed use. Results should explain which differences were treated as legitimate and why.

Equity in healthcare financing

Financing equity examines how payments for healthcare are distributed relative to people's ability to pay and whether obtaining care creates financial hardship. The relevant unit may be the individual, household or population group.

The analysis may examine:

• Progressivity or regressivity of financing contributions.
• Out-of-pocket spending as a share of available resources.
• Catastrophic health expenditure.
• Impoverishment caused by healthcare payments.
• Informal payments.
• Financial barriers that lead people to forgo care.
• Distribution of insurance coverage and benefit protection.

A financing system can raise equal monetary amounts from households while imposing very unequal burdens. Equal payment is therefore not the same as equitable contribution.

Distributional cost-effectiveness analysis

Conventional cost-effectiveness analysis often reports total costs and health gains without showing how those outcomes are distributed. Distributional cost-effectiveness analysis estimates costs and health outcomes across relevant population groups and makes trade-offs between total health and health inequality explicit.

The analysis may include:

• Baseline differences in health.
• Differences in eligibility, uptake and adherence.
• Differences in treatment effect or capacity to benefit.
• Differences in opportunity costs.
• The distribution of health gains and losses.
• Measures of inequality before and after the intervention.
• Alternative social value judgements about health inequality.

Equity weights can assign greater social value to health gains received by disadvantaged groups. The weights are normative inputs and should not be presented as empirical facts. Results should be shown under alternative values when reasonable disagreement exists.

Extended cost-effectiveness analysis

Extended cost-effectiveness analysis examines health gains alongside financial risk protection and the distribution of benefits across population groups. It is particularly useful when interventions affect out-of-pocket spending or protect households from financial hardship.

Possible outcomes include:

• Deaths or illness episodes averted.
• Out-of-pocket expenditure averted.
• Cases of catastrophic expenditure prevented.
• Distribution of health gains by income group.
• Distribution of financial protection by income group.

Health gain and financial protection should be reported separately rather than combined into an unexplained composite. A policy may improve one while having limited or adverse effects on the other.

Equity impact analysis

An equity impact analysis examines how a proposed or implemented policy affects different groups and through which mechanisms. It can be prospective, conducted before implementation, or retrospective, using observed outcomes after a policy change.

A structured process includes:

1. Define the equity objective and the decision the analysis will inform.
2. Identify relevant groups and justify why they may experience different effects.
3. Describe the baseline distribution of need, access, outcomes and financial burden.
4. Map the intervention pathway from eligibility through uptake, implementation and outcome.
5. Estimate group-specific effects including benefits, harms, costs and opportunity costs.
6. Assess implementation barriers that may change who receives or benefits from the intervention.
7. Compare the distribution before and after the policy.
8. Test alternative assumptions and ethical judgements.
9. Identify mitigation or redesign options when inequitable effects are expected.
10. Plan monitoring to determine whether the predicted distribution occurs in practice.

The analysis should not stop at identifying a disparity. It should connect the disparity to modifiable mechanisms and decision options.

Data quality and missing populations

Equity analysis is limited when disadvantaged groups are absent from datasets, grouped into broad categories or recorded inconsistently. Missing group identifiers can itself reflect exclusion within data systems.

Important checks include:

• Whether relevant characteristics are collected using appropriate categories.
• Whether people can self-identify.
• Whether categories have changed over time.
• Whether missingness differs between groups.
• Whether institutional or administrative data omit people outside formal services.
• Whether small groups are suppressed or combined in ways that conceal disadvantage.
• Whether linkage methods perform equally across groups.

Statistical adjustment cannot recover experiences that were never measured. Quantitative findings may need to be combined with qualitative evidence and community knowledge.

Participation and procedural equity

Equity concerns both the distribution of outcomes and the fairness of the process used to make decisions. People affected by a policy should have meaningful opportunities to influence which outcomes, groups and fairness principles are considered.

Participation should be designed to avoid reproducing existing power differences. Providing an opportunity to comment is not sufficient when language, time, accessibility or trust prevents meaningful involvement.

Procedural analysis may consider transparency, representation, reasons given for decisions, mechanisms for challenge and accountability for implementation.

A simplified example

Suppose a screening programme reaches 80% of people in higher-income areas and 50% in lower-income areas. A reminder intervention raises uptake to 88% and 62%, respectively.

Before the intervention, the absolute gap is:

$$ 80% - 50% = 30\ percentage\ points $$

After the intervention, the gap is:

$$ 88% - 62% = 26\ percentage\ points $$

The programme improves uptake in both groups and narrows the absolute gap by four percentage points. However, a substantial difference remains. The analysis should investigate whether transport, appointment flexibility, trust or other barriers limit the intervention's effect in lower-income areas.

Efficiency and equity

Equity and efficiency may align or conflict. Removing barriers to high-value care can improve both population health and fairness, while some equity-promoting policies may reduce the maximum total health achievable from a fixed budget.

The trade-off should not be assumed. Analysis should estimate total outcomes and their distribution, state the social value judgement and show whether a redesign can improve equity with less loss of total benefit.

Equity should also be considered on the opportunity-cost side. Funding a programme may displace services used by other groups, so the distribution of foregone benefits can change the equity conclusion.

Common misunderstandings

Equity analysis is not a demographic subgroup table added after the main analysis. It requires a defined fairness question, an appropriate distribution and an account of the mechanisms producing differences.

Common misunderstandings include:

• Equal resources do not necessarily produce equitable outcomes.
• Any observed difference is not automatically an inequity.
• Statistical adjustment does not determine which differences are fair.
• Population averages do not show who benefits or bears costs.
• Improving outcomes for every group does not guarantee that inequality decreases.
• A relative gap and an absolute gap can move in different directions.
• Equity weights are value judgements rather than measured clinical parameters.
• Subgroup analysis alone does not explain the causes of inequity.
• Lack of statistical significance does not prove that an equity difference is absent.

Interpreting an equity analysis

An equity conclusion should state the outcome, population groups, measure of distribution and principle of fairness being applied. It should report uncertainty and show whether alternative reasonable definitions or value judgements change the result.

A useful equity analysis connects distributions to mechanisms and decisions. It makes visible who gains, who loses, which barriers are avoidable and what trade-offs are being accepted, enabling fairness to be considered explicitly rather than assumed.