Health inequalities

Because an improving average can hide a widening gap: a population's mean indicator can rise while the distance between the most and least advantaged groups grows. It applies when you want to know not only how much health improves overall, but whether that improvement is shared or concentrated. The mean dissolves the distribution: if advantaged groups improve faster than disadvantaged ones, the average rises and inequality grows at the same time. Health inequalities and equity analysis measures that gap with indices that capture the full social gradient, so that you can see whether a policy is closing or widening the distance between groups, which the average alone cannot show with enough precision. It is the difference between knowing that the population is improving and knowing for whom.

Comparing only the two extremes ignores everything in between and depends on how those extremes are defined, so indices that take in the full social gradient are used instead. This applies to any measurement of socioeconomic inequality in health. The difference between the most and least advantaged groups discards the information from the groups in between and changes with how many groups are defined or where the cut-offs fall, which makes it hard to compare across populations or over time. Indices such as the Slope Index of Inequality (SII), the Relative Index of Inequality (RII) and the concentration index summarise inequality across the whole social distribution, weighting every group by its position and size. That gives a measure of inequality that reflects the population as a whole and can be compared across areas and periods on a consistent basis, instead of a comparison that depends on two extreme groups and how they were defined.

It can happen because absolute and relative inequality tell different stories: a gap can shrink in relative terms while it grows in absolute terms, or the other way round. This applies to any reading of how an inequality is evolving, where choosing a single metric biases the conclusion. Relative inequality looks at the ratio between groups, how many times worse off one is than another; absolute inequality looks at the difference in number of cases, years or rate. Depending on the starting values and the size of the change in each group, the relative and absolute differences can move in different directions. That is why both dimensions are reported explicitly: an equity policy is judged differently depending on whether you look at the proportional or the absolute difference, and presenting only one can give an incomplete picture of whether things are getting better or worse.

You need a health indicator that is comparable across groups and a social stratification variable—socioeconomic status, education or income—to rank the population, or a grouping variable such as gender, ethnicity or geography when the comparison is not ordinal, plus strata large enough to estimate the indices precisely. This applies to any equity analysis, where the quality of the stratification determines the quality of the result. The client provides the health data and the information that defines the social groups; the service defines the stratification variable and its ordering, checks that the health indicator is comparable across groups and confirms that the stratum sizes allow the indices to be estimated with enough precision. The variable that defines the social stratification or the comparison group is a sensitive methodological decision: how the population is ranked and grouped determines which gradient or which contrast is measured. The data do not need to arrive perfect, but the social dimension along which you want to measure the gap must be represented in a usable form.

Decomposition quantifies how much of the gap is associated with each factor, such as income, education or geography, and how much is associated with potentially modifiable factors, but it does so in terms of contribution, not causal mechanism. It applies when, as well as measuring inequality, you want to know where an intervention would have scope to reduce it. The analysis splits the size of the gap among the factors that accompany it, estimating the relative contribution of each under the model used. That helps with targeting, but it does not explain why each factor produces the inequality or how the mechanisms link together: it answers "how much of the gap is associated with this factor?", not "through what processes does it generate it?". If the question is about mechanisms, how some determinants lead to others and produce the outcome, that belongs to social determinants of health analysis, not to measuring the gap.

Health inequalities and equity analysis answers ‘How large is the gap, between whom, and how is it changing?’: it measures the gap with absolute and relative indices such as the SII, RII, concentration index or Gini index, and can break down which factors contribute to the observed difference. Social determinants of health analysis asks a different question, ‘What processes might create or sustain that gap?’, and examines how individual, material and contextual factors relate to the health outcome using multilevel or structural equation models. The two complement one another, but neither replaces the other: health inequalities analysis does not model causal mechanisms. If you need to quantify and monitor a gap, you need health inequalities analysis; if the gap is already established and you want to investigate the pathways that may explain it, you need social determinants analysis. The natural sequence is to measure first and investigate possible mechanisms second.

Health inequalities analysis measures the gap between social groups defined by income, education, gender or ethnicity; geospatial analysis describes how the problem is distributed geographically. The dimension is different: inequalities analysis asks how health is distributed across social groups or positions on the social gradient; geospatial analysis places the indicators on the map and analyses spatial patterns, hot spots and accessibility. If geography works as a social marker or as a dimension of access, it can appear in both analyses, but with a different question. The two can intersect, since geography can be one of the dimensions of inequality and a social gap can have a geographical expression, but they answer different questions: social inequality asks between which social groups health is distributed; geospatial analysis asks where it is concentrated in space. If the question is about equity between social strata, this is the service; if it is about geographical distribution and access, it is geospatial analysis of population health.

We deliver the estimated inequality indices, both absolute and relative, across the full social gradient; the decomposition of the gap by contributing factors, with the estimated contribution of potentially modifiable factors; and, where time series are available, the evolution of the indicators on a constant methodological basis, so that comparisons capture real change rather than differences in method between time points. It is for those who need to justify where to target resources with a comparable measure of inequity, or to monitor whether a policy narrows or widens the gap. The deliverable reports the absolute and relative dimensions explicitly, documents the stratification variable and its assumptions, and is built reproducibly so that it can be recalculated with each new wave of data. The client provides the data and decides the equity policy; the service estimates the inequality indices, decomposes the gap and monitors how it evolves, providing the basis on which the decision is made.