Equity Analysis

Equity Analysis of Health, Health Care and Health Financing

Equity analysis asks how the benefits and burdens of a health system or a programme are shared across groups of people, and then judges that pattern against a stated principle of fairness. In health economics it brings together two tasks that are easy to blur, measuring a difference between groups and deciding whether that difference is unfair. The methods are used to monitor health systems and to set priorities, including in health technology assessment, where NICE now considers how a technology affects health inequalities. Choices about the variable, the groups, the need adjustment and the scale of measurement can change the verdict, so a sound analysis states each one openly.

Health, health care and financing as separate questions

The World Bank guide by O'Donnell and colleagues frames equity questions in three spheres. The first is health itself, such as whether gaps in health outcomes between poorer and better-off people have grown. The second is the delivery of care, where the usual principle is horizontal equity, defined in the guide as equal treatment for equal medical need irrespective of other characteristics such as income, race or place of residence. The third is financing, where analysts ask whether payments rise or fall as a share of ability to pay and whether health spending pushes households into catastrophe or poverty.

Each sphere has its own toolkit. Health outcome gaps are summarised with measures of health inequality. Delivery is judged by comparing actual use with the use expected from need. Financing is judged by progressivity, often summarised with the Kakwani index, and by the incidence of catastrophic health expenditure, which the guide describes as medical spending above some fraction of household income or total expenditure, usually over one year. Benefit incidence analysis adds a fourth angle by asking which income groups receive public health spending.

A programme can move these spheres in different directions, for example widening access to care while leaving the health gap unchanged, so an equity analysis names its sphere before its measure.

Measured gaps and judgements of unfairness

Whitehead's paper for the WHO Regional Office for Europe gave the field its working distinction. Health differences can be measured from routine statistics, but only some of them count as inequities, which she describes as differences that are unnecessary and avoidable and also considered unfair and unjust. To call a situation inequitable, the cause has to be examined and judged unfair in its social context. The health inequity page treats that judgement in depth.

The same caution applies to care. Whitehead's working definition of equity in health care covers equal access to available care for equal need, equal use for equal need and equal quality of care for all. She warns that a difference in use rates between social groups does not automatically make the difference inequitable and is a signal that further study is needed. Some people choose not to use a service, and more activity is not always better care.

Cookson and colleagues show that every equity metric embeds value judgements about equality of what, equality between whom and how the inequality is indexed. Harper and colleagues add that the values implicit in health inequality measures can lead to radically different interpretations of the same data. For that reason the fairness principle and the measure are best fixed before the results are seen, and the report shows whether a different defensible choice would change the conclusion.

Steps in a health equity analysis

The WHO handbook on health inequality monitoring breaks monitoring into five steps, which are identifying indicators, obtaining data, analysing them, reporting the results and implementing changes where warranted. An equity analysis of a decision follows a similar path with added attention to need and opportunity cost.

1. State the decision and the equity concern. Name the options and the fairness principle, such as equal use for equal need or reducing the gap in lifetime health.
2. Choose the sphere and the variable. Pick health, use of care or financing, and the indicator that matches the decision, rather than the one that happens to be available.
3. Define the equity-relevant groups. Use socioeconomic position, area deprivation, ethnicity, sex, disability or another characteristic that the decision context justifies, ranked where a ranking exists.
4. Describe the baseline distribution. Report the indicator for each group, which the WHO handbook says should always accompany summary measures.
5. Adjust for need where the question concerns care. Estimate the use each group would have on the basis of need and compare it with actual use.
6. Estimate the effect of each option by group. Include benefits, harms, costs and the distribution of the opportunity cost of funding the option.
7. Summarise on both scales. Report absolute and relative inequality, as the WHO handbook advises.
8. Test the value judgements. Show whether a different equity principle, need adjustment or group definition changes the conclusion.

Need standardisation for health care use

For delivery questions the O'Donnell guide describes standardising use for need before testing for inequity. Need is usually proxied by the use expected from characteristics such as age, gender and measures of health status, and the need-standardised use is then summarised with the concentration index. The choice of need variables is itself a value judgement. Non-need correlates of use, such as insurance cover, are included in the regression but not standardised for, because leaving them out would bias the estimated effect of need.

Absolute and relative measures

The simplest summaries compare two groups, often the most and least advantaged. The gap can be expressed as a difference or as a ratio.

$$D = p_d - p_a \qquad R = \frac{p_d}{p_a}$$

where $D$ is the absolute difference in percentage points, $R$ is the rate ratio, $p_d$ is the rate in the disadvantaged group and $p_a$ is the rate in the advantaged group. For a favourable outcome such as uptake, as in the worked example below, the two rates are swapped, so that both measures show the shortfall of the disadvantaged group as a positive gap and a ratio above one.

The two can move in opposite directions. Harper and colleagues note that when both groups improve, relative inequality rises if the rate of improvement is smaller for the group with the worst initial health, even while the absolute gap narrows. For ordered groups such as income quintiles, measures that use the whole distribution, including the concentration index and the slope index of inequality, avoid discarding the middle groups.

Equity analysis inside economic evaluation

Standard cost-effectiveness analysis reports total costs and total health. Cookson and colleagues describe two ways of making it equity-informative. Equity impact analysis quantifies the distribution of costs and effects by equity-relevant variables, while equity trade-off analysis quantifies the trade-off between improving total health and other equity objectives.

Their health equity impact plane places cost-effectiveness, read as net total health impact, on one axis and net equity impact on the other. Trade-offs arise only when the two axes disagree, as when a cost-effective preventive programme succeeds mainly in advantaged communities. The net equity impact counts who bears the opportunity cost as well as who gains, and a programme funded by cutting other public services may impose its losses mainly on poorer people who rely more on them.

Extended cost-effectiveness analysis, developed within the Disease Control Priorities project, adds financial risk protection to health gains by income group. Distributional cost-effectiveness analysis, developed at the University of York, models the social distribution of health gains and of health opportunity costs within a fixed budget. In the bowel cancer screening case study described by Cookson and colleagues, a targeted reminder strategy gave a smaller gain in total health but a larger reduction in health inequality than a universal one. Equity constraint analysis counts the total health given up by choosing a fairer but less cost-effective option, and equity-weighted analysis asks how much concern for equity would justify that choice, the question examined on the equity-efficiency trade-off page.

How NICE treats health inequality evidence

The NICE manual for technology appraisal and highly specialised technologies (PMG36) accepts a distributional cost-effectiveness analysis in section 4.12 only as supporting evidence, and only where there is clear evidence of a significant burden of health inequalities in the eligible population. At the decision stage, sections 6.2.35 to 6.2.38 say that where reliable evidence shows a technology substantially affects health inequalities, the committee will consider how this bears on whether it is an effective use of NHS resources. The committee can then apply flexibility to the range normally considered cost-effective, but only when the inequality impact is substantial and after considering displacement and opportunity cost. Section 6.2.36 keeps this separate from NICE's legal duties on equality and human rights, including under the Equality Act 2010.

Worked example of screening uptake in two deprivation groups

The figures below are illustrative. A screening programme reaches 80% of invited people in the least deprived areas and 50% in the most deprived areas. A reminder letter raises uptake to 88% and 62% respectively. The question is whether the reminder narrowed the inequality in uptake.

The absolute gap in uptake falls by 4 percentage points.

$$80 - 50 = 30 \qquad 88 - 62 = 26$$

where each figure is uptake in percentage points, least deprived first, before and then after the reminder.

The ratio of uptake also falls.

$$\frac{80}{50} = 1.60 \qquad \frac{88}{62} = 1.42$$

where each ratio is uptake in the least deprived areas divided by uptake in the most deprived areas.

The picture changes when the same data are framed as the adverse outcome, non-uptake. Non-uptake is 20% in the least deprived areas and 50% in the most deprived areas before the reminder, and 12% and 38% after it, so the absolute gap still falls by 4 points. The ratio of non-uptake rises.

$$\frac{50}{20} = 2.50 \qquad \frac{38}{12} = 3.17$$

where each ratio is non-uptake in the most deprived areas divided by non-uptake in the least deprived areas.

The reason is the rate of improvement. Non-uptake fell by 8 points out of 20 in the least deprived areas, a 40% fall, and by 12 points out of 50 in the most deprived areas, a 24% fall. This is the pattern Harper and colleagues describe, a smaller proportional improvement for the worse-off group.

One set of data shows the absolute gap and the uptake ratio narrowing and the non-uptake ratio widening. An analyst who had stated in advance that the concern was the absolute shortfall would report a narrowing, with the non-uptake ratio alongside it. A distributional analysis would go on to model health gains by group and who bears the cost of the letters.

Limits and common misreadings

A table of results by subgroup is a description of who gets what. It becomes an equity analysis only when a fairness principle, a need standard and an account of cause are added. Subgroup analysis for clinical heterogeneity answers a different question.

Decomposition of an inequality index into contributing factors is often read causally. The O'Donnell guide cautions that without estimated causal effects the decomposition explains inequality only in the statistical sense and may be a poor guide to policy.

Cookson and colleagues note that equity-informative evaluation needs social distributions of key parameters rather than population averages, and that data limitations are particularly severe in low-income countries without basic health information systems. A null finding in a small group may reflect low precision and does not show that no inequity exists.

Equity-informative evaluation is an input into deliberation. Cookson and colleagues describe it as an input into decision-making processes and not an algorithm that settles the decision. They also list concerns that these methods do not address, including orphan diseases and non-discrimination, so a fair process needs more than a distributional model. The wider principle sits on the equity and health equity pages.

Sources

• Cookson R, Mirelman AJ, Griffin S, Asaria M, Dawkins B, Norheim OF, Verguet S, Culyer AJ. Using cost-effectiveness analysis to address health equity concerns. Value in Health. 2017;20(2):206-212. https://pmc.ncbi.nlm.nih.gov/articles/PMC5340318/
• Harper S, King NB, Meersman SC, Reichman ME, Breen N, Lynch J. Implicit value judgments in the measurement of health inequalities. Milbank Quarterly. 2010;88(1):4-29. https://pmc.ncbi.nlm.nih.gov/articles/PMC2888011/
• National Institute for Health and Care Excellence. NICE technology appraisal and highly specialised technologies guidance: the manual (PMG36). Published 31 January 2022, last updated 13 October 2026. Sections 4.12 and 6.2.35 to 6.2.38. https://www.nice.org.uk/process/pmg36
• O'Donnell O, van Doorslaer E, Wagstaff A, Lindelow M. Analyzing Health Equity Using Household Survey Data: A Guide to Techniques and Their Implementation. Washington, DC: World Bank; 2008. Chapters 1, 15, 16 and 18. https://openknowledge.worldbank.org/handle/10986/6896
• Whitehead M. The concepts and principles of equity and health. International Journal of Health Services. 1992;22(3):429-445. Also published in Health Promotion International. 1991;6(3):217-228. Originally a WHO Regional Office for Europe discussion paper, 1990 (EUR/ICP/RPD 414). https://doi.org/10.2190/986L-LHQ6-2VTE-YRRN
• World Health Organization. Handbook on health inequality monitoring: with a special focus on low- and middle-income countries. Geneva: World Health Organization; 2013. https://iris.who.int/handle/10665/85345