Section 3: Inequalities and distribution of health outcomes and determinants
Relative contribution of health determinants to health outcomes
One of the challenges to prioritising needs and allocating resources to public health problems is establishing the causes behind identified patterns in disease and ill health. Medical and epidemiological research is very useful in describing the association between various risk factors or exposure and different disease outcomes.
In many ways it is more difficult to model the relative contributions of different broad groups of risk factors on health and wellbeing as a whole. The first ever annual report of the director of public health for South Gloucestershire in 2003 introduced the concept of the wider determinants of health and their influence on inequalities in health and wellbeing.
What are inequalities in health?
The opportunity for a long and healthy life is still linked today, to social circumstances, childhood poverty, where we live, what job we do, how much our parents earned, our race and our gender. Some of the differences, or inequalities in health, are due to factors such as gender that are fixed. But many relate to our social circumstances, our lifestyles and our behaviours: things that can change.
Source: Annual Report of the Director of Public Health 2003 (10)
The relationship between different health determinants is complex. Over the past 20 years, research in this area has provided a range of models to help estimate the relative contribution of different groups of health determinants. Over time, models have become increasingly complex, introducing estimates of the relative contribution of our behaviour, genetics, environmental exposures and socioeconomic circumstance. Figure 2 illustrates a simple model consisting of four main groups of health determinants: healthcare, environmental exposures, genetic determinants and socio-economic factors. Understanding the relative contribution of these different groups of determinants is important because it helps us to prioritise our resources, policies and actions as we look to prevent, rather than treat or manage, health needs.
Figure 2: Simple model of relative contributions of the determinants of health
In a recent influential piece of research from the US, estimates of the relative contributions of the determinants of health were calculated by comparing statistics relating to health outcomes with information about health behaviours and risk factors identified in local surveys. The relative weighting applied to each group of determinants in their model is illustrated in Figure 3 on the following page. There are of course many differences between the US and England which mean that this model cannot be directly or easily applied to South Gloucestershire. There are differences in the provision of healthcare and social security, environmental protection, trends in health behaviours and criminal justice.
Figure 3: Model of relative contributions
Source: Analysis of US County Health Rankings data (11,12)
To find a recent example drawing on information from England, we can look at the recent work undertaken by Public Health England as part of the Global Burden of Disease study (13). They estimated the relative contribution of different groups of health determinants by modelling the proportion of different causes of disease that could be attributed to known risk factors.
One of the advantages of this model is that it is based on national data (England) and can be compared with similar data for all of the other countries included in the Global Burden of Disease study. The analysis undertaken by PHE looked at the data by each of the English regions, comparing and highlighting differences between these populations.
There is no model which can currently be used to estimate the relative contribution of broad determinants of health on the health and wellbeing outcomes in South Gloucestershire. Although individual estimates can be made of the contribution of specific health behaviours on specific outcomes (for example, alcohol consumption on hospital admissions or smoking on mortality), a more complete breakdown is more difficult. This is due in part to the limitations of our understanding of health knowledge, attitudes, beliefs and behaviours at the local authority and neighbourhood level.
Variation between electoral wards in South Gloucestershire
One way in which we can increase our understanding of the relationship between determinants and health outcomes is to explore their distribution and variation between smaller geographic areas or communities. Electoral wards are a useful geographic level to study differences and associations in health determinants and outcomes. They are large enough to be able to pull together information about individuals without compromising confidentiality, but small enough to ensure there are enough data points to allow for some meaningful analysis.
The supplementary health atlas which accompanies this report provides a collection maps that illustrate the geographic distribution of a range a health measures which are available at electoral ward level [a].
There are a range of different ways that public health professionals can assess the level of inequality within and between populations (15–17). Although the JSNA includes short sections on health inequalities for each topic area (6), there has not previously been an attempt to comprehensively assess the scale of health inequalities for each of indicator available at electoral ward level. For this annual report of the director of public health, a suite of summary measures of health inequality were calculated for each health outcome in the atlas to identify which issues made greater or lesser contributions to overall health inequalities in South Gloucestershire [b].
Table 1 below summarises a single measure of relative inequalities for the selection of health indicators included in the atlas. A higher relative index of inequality (RII) score suggests a greater degree of inequality between wards within South Gloucestershire and a higher association with the locally ranked measure of deprivation (IMD 2015). As the reduction of health inequalities is a key public health function for local authorities, a higher rank on this list indicates a greater priority for addressing inequalities [c].
Table 1: Rank order of health outcomes by measure of relative inequality
|Indicator measure||Relative Index of Inequality (95% CI)|
|Mothers with record of smoking at time of delivery||3.55 (2.73 – 4.37)|
|Alcohol-specific hospital admissions||3.22 (2.53 – 3.9)|
|Emergency admissions for substance misuse||2.8 (2.22 – 3.38)|
|Emergency admissions for Chronic Obstructive Pulmonary Disorder (COPD)||2.72 (2.12 – 3.32)|
|Bad health (census)||2.34 (1.87 – 2.81)|
|Very bad or bad health (census)||2.03 (1.71 – 2.35)|
|Emergency hospital admissions for mental/behavioural issues||2.03 (1.76 – 2.29)|
|Alcohol related admissions||1.64 (1.39 – 1.89)|
|Rate of years of life lost – all causes||1.64 (1.32 – 1.96)|
|Emergency admissions for diabetes||1.51 (1.26 – 1.76)|
|Continuation of breastfeeding||1.37 (1.12 – 1.76)|
|Emergency admissions for Coronary Heart Disease (CHD)||1.36 (1.13 – 1.58)|
|Children overweight or very overweight at Year 6||1.3 (1.1 – 1.5)|
|Emergency hospital admissions for injury in children under 5||1.24 (0.92 – 1.57)|
|Children overweight or very overweight at reception||1.1 (0.9 – 1.3)|
|Male life expectancy at birth||1.07 (1.03 – 1.12)|
|Female life expectancy||1.07 (1.01 – 1.13)|
|Low birth weight||0.97 (0.74 – 1.2)|
|Teenage pregnancy (small area data suppressed)||Estimate not available|
The data in Table 2 is ranked by the RII measure. It shows the indicators with the highest measures of inequality at the top of the table. Whilst the table shows a social gradient across nearly all of the indicators, the ranking suggests priority areas for the council and its partners to make gains in reducing in equality.
It is possible to use the RII measure to track progress in reducing health inequalities over time and the public health and wellbeing division will consider how it can use these measures to monitor the impact of our work in this area.
Using data to identify where the greatest health inequalities lie only tells half of the story. Measures of inequality show the gap or difference in health outcomes experienced by the most and least deprived communities – in the example given above, this was the difference by electoral ward areas. Another approach to tackling inequalities is to look at which local communities appear to have the consistently highest level of health needs when looking at the information available at this level.
Although ward level health data has previously been published by the council, the English indices of multiple deprivation (IMD) are often used as a proxy indicator for local needs and a way of prioritising local resources. For this report, we used the ward level health indicator data listed in Table 1 to rank each ward area by the level of need and assigned them to five equal sized groups or ‘quintiles’. By counting the number of indicators that each ward appears in each indicator quintile, the wards were then ranked by a measure of overall health need.
Figure 4 shows the comparison of the different approaches to ranking electoral wards by a summary measure of health need. The first column uses the IMD 2015 score to show the local rank order; the second column shows the wards ranked by only the health domain of the IMD 2015 score; the final column is ranked by the mean rank value using the suite of health indicators available at ward level as described above. Ward names highlighted in red are those which include priority neighbourhood areas.
This figure shows that the relative level of overall health need for each electoral ward is highly dependent on the indicators used to generate the index. This is a strong argument in support of local authority access to health data at individual level and small area geographies as this enables an improved understanding the spatial distribution local health need and levels of health inequality across a wider range of issues.
Figure 4: Comparison of ward ranking by different measures of health need
This summary analysis supports the prioritisation of public health resources and activity by ward area to address the health inequalities identified earlier in the report. The principle of ‘proportionate universalism’ requires to the council to address the social gradient in health through universal action, but with a scale and intensity which is proportionate to the level of disadvantage (18).
[a] A list of the indicators included in the atlas is included in Appendix A of this report.
[b] Summary measures included: absolute and relative gap, odds ratios, and slope and relative indices of inequality.
[c] Further information about how the RII has been calculated for this report is included in Appendix A of this report.