Exploring the validity of using the English index of Multiple Deprivation as a proxy for individual or household income

OBJECTIVES: To explore the validity of using the English Index of Multiple Deprivation (IMD) as a proxy for individual or household income; for total IMD and subdomains, and for different individual and household types. STUDY DESIGN: Cross-sectional analysis of survey and administrative data. METHODS: Income and contextual data were obtained from the Understanding Society survey, alongside the Lower Super Output Area (LSOA) of residence to enable linkage to the 2019 English IMD. The relationship between income and IMD (including subdomains) was explored using descriptive analysis and univariable linear regression models. Multivariable regression models were then employed to determine the relationship between income and IMD for different individual and household characteristics. RESULTS: Greater levels of income heterogeneity were observed in the least deprived LSOAs. R-squared values from univariable models suggest that around five percent of the variation in household income and around one percent of individual income can be attributed to IMD. Also, IMD was a particularly poor predictor of income for certain groups: young people under 25, females, those not in good health, White British individuals, single person households without children, and households in social rented accommodation. CONCLUSIONS: IMD is a poor proxy for individual or household level income, particularly for some population subgroups. Using IMD as a proxy potentially introduces large amounts of measurement error. IMD should be treated as an area level measure, both conceptually and statistically. Some IMD subdomains may be more useful for certain studies.