Background
The associations between occupational status, spatial concentration and health have been extensively researched. However, changes in patterns of employment suggest that established measures might not be wholly representative of modern socioeconomic conditions. The study examined if the Index of Concentration at the Extremes (ICE) derived from the neighbourhood concentration of occupational classifications in the UK census described geographical changes in self rated health (SRH). A comparison was made with the commonly used Index of Multiple Deprivation (IMD) to determine the relative utility of the ICE and its contribution to modelling health inequalities in comparison to the IMD.

Research questions
A systematic review examined the association between SRH, occupational status and use of the ICE in public health research. Quantitative analysis assessed associations between the concentration of advantage, disadvantage and SRH in the North West England region of the UK, including geographic correlations between SRH, IMD and ICE in 2001 and 2011 census data.
The research questions were: what is the utility of using ICE metrics derived from employment relations compared to more traditional measures of deprivation represented by the IMD, for explaining relative spatial inequalities in SRH? Are employment relations as operationalised by occupational status better at explaining variations in SRH than more traditional measures of social deprivation?

Methods
Data on SRH, IMD, occupational status (NS-SeC) age and ethnicity in the North West England region of the UK was extracted at the Lower Super Output Level (LSOA) level from the UK national census datasets for 2001 and 2011. The association with SRH was examined for IMD and a novel ICE derived from census returns enumerating the occupational categories of the National Statistics Socioeconomic Classifications (NS-SeC).
Bivariate analysis determined the relationship between ICE, IMD and SRH and to test for any significant relationship that varied geographically. Hot spot analysis identified statistically significant spatial clusters of high and low values. Geographically Weighted Regression (GWR) provided a local model by fitting a regression equation between dependent and independent variables in each neighbourhood of the region.

Results
Analysis of 2001 and 2011 census data found better SRH in more rural areas of the North West region such as Cumbria and Cheshire. Poorer SRH was found mainly in East Lancashire, between Liverpool and Greater Manchester, and in coastal communities in Cumbria and the Fylde coast.
Concentrations of working-class occupations were associated with poorer SRH and concentrations of higher-status occupations were associated with better SRH. The ICE derived from the NS-SeC made a greater contribution to the variance explained by a geographically weighted regression (GWR) model than the IMD.
Average age and age groups best predicted SRH for the 2001 census, while the ICE combined with age and ethnicity was the best predictors for SRH in 2011.

Discussion
ICE measures derived from NS-SeC data demonstrated that the ICE is a useful adjunct to conventional measures of material deprivation, as it may capture neighbourhood conditions not represented by the IMD. Incorporating extremes of socioeconomic status allowed the examination of neighbourhood inequalities that do not rely on a single disadvantaged group. The ICE improves on the IMD in that occupational classifications represent qualities of employment not captured by absolute measures of deprivation. There is potential for the ICE to be used with other measure and geographies.