Access the full text.
Sign up today, get DeepDyve free for 14 days.
(1988)
Income equivalences and the selection uf family expenditures on children
(1985)
Wellington: Department of Health
J. Eachus, M. Williams, Philip Chan, G. Smith, M. Grainge, J. Donovan, S. Frankel (1996)
Deprivation and cause specific morbidity: evidence from the Somerset and Avon survey of healthBMJ, 312
P. Morrison (1994)
Housing occupancy and the changing size of households and dwellings in New Zealand 1951-1991.New Zealand Population Review, 20
(1993)
Deprivation and resource allocation in the West Midlands regional health authority
A. Sloggett, H. Joshi (1994)
Higher mortality in deprived areas: community or personal disadvantage?BMJ, 309
Townsend (1987)
125J Soc Pol, 16
Margaret Eames, Yoav Ben-Shlomo, M. Marmot (1993)
Social deprivation and premature mortality: regional comparison across England.British Medical Journal, 307
C. Morphet (1992)
The interpretation of small area census data.Area, 24
Sally MACINTYREt, Sheila MACIVERij, A. Sooman (1993)
Area, Class and Health: Should we be Focusing on Places or People?Journal of Social Policy, 22
(1978)
Minimum income levels und income equivalence scales
J. Michaelis, R. Rahn, V. Carstairs, Russell Morris (1989)
Deprivation: explaining differences in mortality between Scotland and England and Wales.British Medical Journal, 299
(1998)
NZDep96 index of deprivatiun: Research Report No. 8. Wellington: Health Services Research Centre
(1992)
Immunisation coverage in New Zealand: Results of the regional immunisation coverage surveys
Roger Anderson, P. Sorlie, E. Backlund, N. Johnson, G. Kaplan (1997)
Mortality Effects of Community Socioeconomic StatusEpidemiology, 8
Townsend Townsend (1987)
DeprivationJ Soc Pol, 16
S. Schwartz (1994)
The fallacy of the ecological fallacy: the potential misuse of a concept and the consequences.American journal of public health, 84 5
Russell Morris, V. Carstairs (1991)
Which deprivation? A comparison of selected deprivation indexes.Journal of public health medicine, 13 4
Department of Public Health, Wellington School of Medicine, Wellington, New Zealand Abstract In New Zealand, existing area-based indices of deprivation were inadequate because of lack of theoretical underpinning and use of comparatively large areas resulting in masking of variation within them. There is growing demand for small area based indices of deprivation for the purposes of resource allocation, research, and community advocacy. This paper describes a new Census-based index of deprivation based on the smallest possible geographical areas using existing Census boundaries. The index uses deprivation variables selected according to established theory, and derived from the 1991 New Zealand Census. Ten age and gender standardised variables were combined using principal components analysis. Each variable is a standardised proportion of people in a small area with a lack of a defined material or social resource. Age/gender standardisation is important to avoid confounding and to improve the performance of indices in resource allocation formulae. The index correlates highly with mortality, hospital discharges, lung cancer registrations and childhood immunisation status. (Aust N Z J Public Health 1998; 22: 835-7) Peter Crampton Health Services Research Centre, Wellington, New Zealand Frances Sutton Wellington, New Zealand eprivation refers to relative disadvantage and consists of material deprivation, such as housing and living conditions, and social deprivation, such as social support and education factors.' Deprivation may be measured at the individual level and at the group level. In New Zealand, the most widely used area based measure of deprivation is the Health and Equity index,2which is based on areas with population of around 2,000. Such comparatively large areas result in masking of variation within them. We planned to develop a theoretically-grounded robust Census-based area measure of deprivation specific to New Zealand, based on the smallest area unit possible. Methods Small areas were defined based on Statistics New Zealand's (SNZ) smallest geographical units, meshblocks, which have a median population about 9 0 . Small meshblocks were agglomerated where necessary, but only within known SNZ primary sampling unit boundaries, to make connected small areas with population at least 100 (as far as possible). About 35,000 meshblocks were agglomerated into 20,166 small areas, 94.6% of which had populations of at least 100 persons. Larger area sizes were also explored. Variables in the index are proportions of persons or households in an area with a lack of something and reflect seven dimensions of deprivation: income, transport, living 1998 VOL. 22 NO. 7 space, home ownership, employment, qualifications and support. Two of these, income and support, are captured by more than one Census variable. Two variables, household income and occupancy, were adjusted for household size and composition, using standard equivalence scale^^-^ to avoid bias created by difference in family size.6 For example, unadjusted household income will not adequately reflect resources available to family members. All variables are related to age and gender to some extent so we produced agelgender standardised proportions via indirect standardisation, with the New Zealand population as the standard.6 We used four age bands. The youngest age group, 0-17, reflects non-voting status and, in general, dependency, while the oldest, 60 and over, reflects 1991 entitlement to state retirement income, and vulnerability to changing living arrangements, income levels, employment status and health status. The remaining adults have been split into two roughly equal size groups, 18-39 and 40-59. Variables based on household characteristics were indirectly standardised by ascribing the household characteristic to each individual in the household. Some variables included in the index are also age group restricted as not all combinations of age group and dimension are meaningful.6 The NZDep91 index is the score on the first principal component from a principal components analysis; that is, it is a weighted sum of 10 standardised proportions in an area. Each component meshblock in a small Correspondence to: Us Clare Salmond, Department of Public ilealth, Wellington School of Medicine, PO 3ox 7343, Wellington South, New Zealand. Fax: (64-4) 389 5319; e-mail: $almond@ wnmeds.ac.nz AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH Salmond, Crampton and Sutton Brief Report area, typically consisting of one or two meshblocks, is assigned the small area score. Table 1: Weights for variables included in NZDep91. Standardised proportion of persons Weight (coefficient in first principal component) 0.375 0.367 0.364 0.356 0.342 ResuI ts The first principal component of the 10 variables accounts for 47.3% of the overall variance. The factor structure of the first two (major) components using varimax rotation does not suggest clear subscales: the index appears to measure a single underlying construct of deprivation.The weighting coefficients for the first principal component are shown in Table 1 . Separate indices were created based on standardised and unstandardised variables. The differences in weights between the two first principal components are generally slight, and the scores from the two indices are very highly correlated (r=0.997). However, we have used the standardised variables in NZDep91 because some very deprived small areas are affected non-negligibly. The weights for these variables are consistent with expectations from other indexes2.â Careful examination of both the smallest areas and those at either extreme of the deprivation scale showed neither anomalies nor large pockets of missing data. An alternative index based on agglomeration sizes of at least 200 persons had weights very similar to the aggregations to 100 so this version was not used. We also validated the index by correlation with health outcomes well known to be associated with deprivation. Three datasets were explored: mortality from all causes in the Wellington region, 1990-93; hospital discharge ratios for the same region and period; and national registrations for lung cancer for the same period. Each dataset was explored in four broad age bands. An ordinal 40-point deprivation scale was created from the distribution of deprivation scores for each area by assigning a rank to each ordered 2.5% of the distribution. Rates of events were calculated within each of the 40 groups of areas. Rank correlations of event rates and deprivation show clearly that areas of increased deprivation experienced increased mortality rates, increased hospital discharge ratios, and increased registrations for lung cancer (Table 2). The lack of correlation between lung cancer and deprivation among those under 40 years of age, particularly those under age 18, is to be expected. Lung cancer in this latter age group is very uncommon.The correlation between deprivation and discharge ratios is least in those 60 years and over, and may be due to confounding by length of stay, and residual confounding by age. In single parent families and aged under 60 On means tested benefit and aged 18-59 Below threshold for equivalised household income Without access to a car and aged 18 or over Without qualifications and aged 18-59 Unemployed and aged 18-59 Separated or divorced and aged 18-59 Not in owned home Below threshold for equivalised occupancy Separated or divorced and aged 60 or over A 1992 survey of 706 two and three-year-old children throughout New Zealand provides an example of the use of the index in research. Data included the meshblock identification of the childâs address (except for six children) and an indication of whether the child was fully immunised according to the contemporary immunisation schedule.* The proportion of these 706 children fully immunised by age two was 56.4% rising to 64.3% by the time of the interview when some were aged three. Children not fully immunised by age two were more likely to be living in more deprived areas (KruskallWallis z=2.26, p=0.024). In time, however, the incompletely immunised children in the more deprived areas completed immunisation at a greater rate than partly immunised children elsewhere abating the relationship with deprivation by the time of the interview (z=1.44, p=O. 15). This change may reflect efforts of health professionals to increase immunisation uptake, and/or partly immunised children in the more deprived areas having delayed immunisation for social or health reasons. Discussion Indices of deprivation have application in needs-based, population-based funding formulae and can be used in research in a Table 2: Rank correlations between âvalidationâ data and 40-point deprivation scale values, by age group. 18-39 ~~ 0-17 ~~ ~ Age group (years) 40-59 60 and over Dataset Lung cancer (New Zealand 1990-93) Discharae ratios (Wellinaton 1990-93) Mortality (Wellington 1990-93) :r -0.05 pâ ~0.001 r s 0.26 0.94 0.41 r s <0.001 <0.001 0.009 <0.001 <0.001 r s 0.91 0.81 0.73 P <0.001 <0.001 <0.001 Note: (a) rs = Spearman rank correlation coefficient; p = probabi/ity AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 1998 VOL. 22 NO. 7 Brief Report New Zealand index of deprivation variety of settings such as health and other social services. They are also useful for advocacy at a community level. Composite indices provide a powerful basis for explaining differences in health status,â-â although the relative importance of area level effect on health status versus individual level effects is currently being debated.â.â Furthermore, they are stable over time (during intercensal periods), robust (able to withstand small data input errors), and focus on physical and social environment as well as p e r ~ o n . However, their derivation is complex, their ~~.~~ their input variables use is prone to the ecological falla~y,â~,lâ may be selected rather arbitrarily, and there are statistical limitations when applying indices to small In the case of NZDep91, we were careful to select our variables from a theoretical perspectiveâ and to use a standard statistical procedure to derive the index. Our index, while complex, has been created for all meshblocks in New Zealand, and is readily available. As far as we know, NZDep9 1 is the only index to incorporate variables equivalised for household composition, and to use age and sex standardisation of the input variables. The latter is important for two reasons: In the resource allocation context age and gender are usually taken into account directly, being such important determinants of need for health services in their own right. By using age and sex standardised variables we avoid including the influence of age and sex twice in a resource allocation formula. In the research context, age and sex are commonly strong confounders in the relationship between deprivation and health or other outcome measures. Standardisation by eight age/gender groups will remove much of the confounding between area deprivation and any outcome measure, although effect modification is possible. A similar argument applies to ethnicity. The largest ethnic minority in New Zealand is Maori, nearly 13%of the population in 1991 and unevenly distributed across the country. The resulting small numbers in many areas precluded standardisation for ethnicity. The income variables used reflect a lack of resources and are likely to be highly correlated with all sorts of deprivation variables, not just the selected few available from the Census. Thus, personal and household income thresholds in the NZDep91 index are used as surrogates for factors such as inadequate diet, inadequate clothing or inadequate household facilities - elements of material deprivation cited by Townsendâ but not available from Census variables. A wide range of organisations and groups have now used the index and many are beginning to use the updated index based on 1996 Census data.â Acknowledgments We acknowledge the very generous support and assistance provided by staff of Statistics New Zealand (SNZ). To protect against disclosure of information supplied by individual respondents, SNZ practice is to randomly round all aggregated Census output. In this case, access was granted to unrounded aggregate Census data, under a special contract between SNZ and Clare Salmond and Peter Crampton, so that the scarcely populated meshblocks could be sensibly agglomerated to form larger areas. The access was granted in a strictly protected environment on SNZ premises, under supervision of SNZ staff, and no unrounded Census aggregations were removed from the site. Both researchers are bound by the same provisions of the Statistics Act 1975 which bind staff of SNZ to preserve the confidentiality of individual respondent data. We are grateful to the Ministry of Health for providing the childhood immunisation data. This research was funded by the Health Research Council of New Zealand.
Australian and New Zealand Journal of Public Health – Wiley
Published: Dec 1, 1998
You can share this free article with as many people as you like with the url below! We hope you enjoy this feature!
Read and print from thousands of top scholarly journals.
Already have an account? Log in
Bookmark this article. You can see your Bookmarks on your DeepDyve Library.
To save an article, log in first, or sign up for a DeepDyve account if you don’t already have one.
Copy and paste the desired citation format or use the link below to download a file formatted for EndNote
Access the full text.
Sign up today, get DeepDyve free for 14 days.
All DeepDyve websites use cookies to improve your online experience. They were placed on your computer when you launched this website. You can change your cookie settings through your browser.