Abstract:
Population ageing is one of the largest contemporary health challenges. Not only do shifting population
age structures place significant burden on both individuals and social systems, they also threaten to
widen health inequities in old age. An effective health response to population ageing requires high
quality research of older adults. Yet, area-based deprivation measures that are derived from the
working-age population dominate ageing policy and research. This thesis challenges the acceptance of
conventional measures of deprivation for older populations and proposes the use of an alternate
deprivation tool, the Older People’s Index of Multiple Deprivation [OPIMD].
The OPIMD is a new and novel individual-based deprivation index that is constructed from indicators
that are most relevant to older New Zealand adults. This thesis is one of first pieces of research to
evaluate the OPIMD for a census-derived population aged 65 years and over. Logistic regression
modelling and regression tests were completed for three outcome variables (Parkinson’s disease,
glaucoma, and no access to a motor vehicle) to determine the influence that the OPIMD has on health
and social inequalities and to compare this influence to two area-based deprivation indexes, the New
Zealand Index of Deprivation [NZDep] and the New Zealand Index of Multiple Deprivation [NZIMD].
The OPIMD was a significant predictor for all outcome variables and produced a stronger association
with Parkinson’s disease and motor vehicle access than the NZIMD or NZDep.
The findings of this thesis support the use of the OPIMD as a deprivation tool in older adult populations.
The OPIMD accurately captures the experience of deprivation in old age and would therefore be useful
for measuring health inequalities in health research and intervention planning. The OPIMD should also
be considered for use alongside area-based measures in health funding decisions, such as the New
Zealand capitation funding formula. However, application of the OPIMD at the population-level would
be supported by a robust method of data imputation.