Published: 06/05/2004, Volume II4, No. 5904 Page 26 27

Government guidance on measuring health inequalities is rife with inconsistencies.

A system that reveals percentage health gaps should make the task more effective.

Allan Low explains

The need for primary care trusts and local strategic partnerships to quantify levels of health inequality across their populations is growing. Both are expected to set local health inequality targets and demonstrate measurable progress in reducing local inequalities over the next three years under the Department of Health's priorities and planning framework. Service planning is to be informed by equity audits, performance-managed by strategic health authorities.Measuring levels of health inequality becomes an essential requirement when choosing topics for equity audit, setting local targets against relevant benchmarks and assessing progress towards reduced inequalities in health.

But despite all this, very little advice has been given on how to measure levels of health inequality locally.

Sadly, there is no clarity to be gained by examining how the government's own health inequality targets and indicators have been constructed, as there are three inconsistencies between government policy on health inequalities and the construction of targets and indicators to measure progress.

First, of the government's two headline health inequality targets, one - on life expectancy - measures health variability and is constructed without reference to socio-economic factors. The other - on infant mortality - measures socio-economic inequality in health and is constructed with reference to socioeconomic factors.

Similarly, the national health inequality indicators published in Tackling Health Inequalities: a programme for action are a mix of those constructed with and without reference to socio-economic factors. Indicators constructed with reference to socio-economic factors and which do measure socio-economic inequalities include road accidents or the number of people who smoke; increasing the number of people having influenza vaccinations and reducing teenage pregnancies are not. Nor is the mortality rate from major diseases.

The second inconsistency between government policy and quantification of targets and indicators relates to statements by public health minister Melanie Johnson and advisers about 'the importance of not just considering the two extreme ends of the spectrum when examining [inequalities in] life expectancy, but the gradient across the whole population'.Yet targets are set in terms of the average and the worst 20 per cent.

The third inconsistency relates to the requirement for PCTs to take the lead in addressing health inequalities and to demonstrate progress in reducing local inequalities in the next three years, yet the only assistance with assessing progress are the 70 indicators.

These indicators will only be useful for assessing and tracking health inequalities if differential data is gathered by area or socio-economic group, measures of levels of inequality across the groups are reported, and changes in inequality levels are monitored over time.

No accompanying guidance on how to use the indicators basket to construct summary measures of socio-economic inequalities is provided.

Sunderland teaching PCT has taken a different approach. Assessment of socio-economic inequalities in health within the trust are taken by using a metric which measures them across the whole population, uses routine data sources and can be calculated on a spreadsheet. Results are provided in easily understood percentage health gaps that enable comparisons between illness types and over time, and enables the estimation of levels of inequity in service provision.

The metric employed is the slope index of inequality, which is calculated as the gap in the rate of mortality or morbidity between the most and least disadvantaged groups. It takes account of all the groups in between, as well as their population sizes.

The index provides an estimate of the absolute health gap across all groups. This has limitations for making comparisons between conditions and over time since the size of the gap will depend on the measurement scale being used. The requirement for making valid comparisons over time is that the measure stays constant if the health status of all groups changes by the same proportion between two time periods. The requirement for making valid comparisons across conditions is that the measure should be scale-neutral and therefore not influenced by the units used to measure any health condition.

Both these requirements can be easily addressed by calculating the relative gap. This merely represents the absolute health gap in terms of a percentage of the average level of health across all groups.

The relative gap can be used to compare levels of inequality for different causes of death. In Sunderland, the percentage gap for coronary heart disease is 50 per cent. For stroke, the percentage gap is only 22 per cent.

For breast cancer, it is only 5 per cent. Comparisons can also be made with indicators of morbidity. For example, the percentage gap for admissions for accidents and accidental injury is 50 per cent. For teenage pregnancy, the level of inequality is 82 per cent.

They can provide an assessment of current levels of inequality, so inform the setting of realistic local targets for reducing inequalities. They can be compared over time, so inform performance assessments.

In the absence of clear guidance from government on how to quantify health inequalities locally, the practical advantages and weaknesses of the SII metric deserve to be explored further by more widespread use within local service providers and PCTs.

In practice: the slope index of inequality

Implementing the slope index of inequality has enabled Sunderland teaching primary care trust to look more closely at local reasons for rates of deprivation, writes Emma Forrest.

'We wanted a way to see what was different from what we expected.Since carrying out the work we have asked staff if they have found it informative and the response has been positive, ' says PCT director of public health Dr Judy Thomas.Work carried out by local strategic partnerships is expected to benefit from the system particularly.

'It illustrates the discussions we need to have. It shows us that sometimes you have to look underneath the data.For example, we have high rates of teenage pregnancy in a particular ward - is this because a lot of young mothers are being rehoused in one area? It has also enabled us to look at how the population is accessing primary care and what that really means.'

Allan Low is a freelance health economist

Key points

National guides for assessing inequality at local level are inconsistent.

A simple tool, the slope index of inequality, measures inequalities in health across the whole population of a primary care trust.

The index enables PCTs to measure inequality by health outcome and service provision.

Further information

www. sunderland. nhs. vh/hyromo