Data Insight: Assessing the reliability and validity of the Northern Ireland Multiple Deprivation Measure 2017

Authors: Enya Redican, Stein Gerrit Paul Menting, Jamie Murphy & Mark Shevlin

Small-area deprivation measures help identify the most deprived areas, supporting decisions about where resources are most needed and informing efforts to reduce health inequalities. The Northern Ireland Multiple Deprivation Measure (NIMDM) 2017 is Northern Ireland’s official measure of area-level deprivation and measures deprivation across seven domains: Income, Employment, Health & Disability, Education, Skills & Training, Access to Services, Living Environment and Crime & Disorder. This Data Insight explores the research utility of the NIMDM 2017. 

Read publication

Given that the Northern Ireland Multiple Deprivation Measure (NIMDM) 2017 is widely used by administrative data researchers, we wanted to investigate whether the measure and its seven domains accurately reflect what they were designed to measure, and whether the measure is associated with health outcomes in the ways one would expect. Accessing NIMDM 2017 domain-indicator data via the NI Neighbourhood Information Service, and linking NIMDM data to hospital admissions data, we conducted our analyses in two-phases. 

Phase 1 revealed that indicators within some domains were weakly associated and, therefore, did not seem to be measuring the same construct. Some indicators were also more highly associated with indicators in other domains rather than with those in their own domain. Finally the inferred seven-domain structure of the measure was not supported by the data. 

Phase 2 revealed that associations between deprivation level and health outcomes were extremely weak. While higher deprivation scores were associated with an increased risk of hospital admission for chronic obstructive pulmonary disease (COPD), substance related mental health disorder and depression, associations with the other physical and mental health conditions were negligible.

Overall, our findings suggest that the current NIMDM may not be performing optimally, and that the statistical methods used to create the measure may not be entirely consistent with how deprivation is most commonly conceptualised. We hope that these findings may be useful for the development of future iterations of the measure. 

What we found

Phase 1 results 

• Most indicators within domains were satisfactorily correlated with one another; however, several indicators in the Living Environment domain showed weak, statistically non-significant or negative correlations.
• Only the Education, Access to Services and Crime and Disorder domains comprised indicators that were most highly correlated within those domains. 
• Internal consistency testing indicated that most domains were reliably identified by indicators, however, the Income and Living domains were not.
• Indicators within the Living Environment domain were more closely associated with the overall NIMDM score than to the overall Living Environment score. 
• The indicator data did not support the inferred dimensional structure of the NIMDM 2017. Individual, unidimensional domains, and a hierarchical structure which recognised seven correlated domains could not be identified with the data.

Phase 2 results

• Although most health conditions were associated with deprivation, the correlations were extremely weak.
• Six of the ten health outcomes – including coronary heart disease (CHD), dementia, diabetes, hypertension and stroke – showed little or no relationship with deprivation. By contrast, hospital admissions for COPD, depression and mental and behavioural disorders related to substance use increased as deprivation worsened.
• There was no increased likelihood of being admitted to hospital for six of the ten health outcomes (CHD, dementia, diabetes, hypertension, stroke, and other conditions) as overall and domain deprivation worsened.
• There was an increased likelihood of being admitted to hospital for COPD, depression and mental and behavioural disorders due to psychoactive substance use as overall and domain deprivation worsened.

Why it matters

The NIMDM 2017 has been widely used by researchers and policymakers to understand and attend to patterns of deprivation and need across the Northern Irish population. Following the introduction of new statistical geographies after the 2021 Census, the current measure is now outdated, creating a need for a revised index aligned with the new spatial framework.

Small area-level deprivation measures play a central role in informing policy decisions, resource allocation and the monitoring of health and social inequalities. However, validation has been identified as an underutilised step in their development. By conducting a data-driven investigation of the NIMDM 2017, this study highlights indicators and domains that may require attention and provides evidence on how well the current measure captures variation in health risk. These findings have direct implications for the development of the next iteration of the NIMDM, and may aid in the development of a more psychometrically robust and policy-relevant measure for use by both researchers and policymakers alike.