To calculate the AGI we first calculate the outcome of interest for each neighbourhood in the country (for example the rate of emergency hospital admissions per 100,000 population). The calculated outcome may need to be standardised to allow us to compare neighbourhoods with different population age and sex structures. We then rank all the neighbourhoods on a zero to one scale based on the index of multiple deprivation rank with the richest (least deprived) neighbourhood being ranked zero and the poorest (most deprived) neighbourhood being ranked one. Next we select all the neighbourhoods in the CCG for which we want to calculate the AGI and plot the outcome for each neighbourhood against the index of multiple deprivation rank. Finally we draw a line of best fit through the points we have plotted - when fitting our line we give those neighbourhoods with larger populations more weight than those neighbourhoods with smaller populations. The AGI is the gradient of this line. The steeper the line the greater the difference between the richest and poorest neighbourhoods and hence the greater the inequality and the larger the AGI value.
The similar CCG group members are selected based on how closely matched they are on the following criteria: IMD (deprivation) score, IMD health domain score, total registered population, % of population 0-4, % of population 5-14, % of population 15-24, % of population 75+, ratio of registered population to ONS estimates ("list inflation"), population density, slope variation in population density, % of population black ethnic groups, % of population Asian ethnic groups. Further details can be found here.
The scatter plot shows LSOAs within CCGs with the size of the point representing the population that the LSOA contributes to the CCG. The caterpillar plot show CCG AGI values and their 95% confidence intervals with the average across all CCGs plotted represented by the red dashed line
Contains National Statistics data © Crown copyright and database right 2016. Contains OS data © Crown copyright and database right 2016. Data licensed under the Open Government Licence v3.0