For whom is income inequality most harmful? A multi-level analysis of income inequality and mortality in Norway
Introduction
The relationship between inequalities in income distribution and health outcomes has been explored for almost 30 years. Rodgers (1979) was probably the first to empirically assess the hypothesis that there was a relationship between country-level income inequality and population health indicators such as infant mortality and life expectancy. Several others addressed the topic during the 1980s (e.g., Flegg, 1982; Legrand, 1987; Wilkinson, 1986), and research proliferated in the 1990s. Studies from a growing number of countries and sites have addressed the question whether or not income inequality in a society is associated with population health, over and above the effects of average income and individual-level health predictors. Recent reviews (Deaton, 2003; Lynch et al., 2004; Macinko, Shi, Starfield, & Wulu, 2003; Subramanian & Kawachi, 2004; Wilkinson & Pickett, 2006) demonstrate that numerous studies have found associations between income inequality and health outcomes such as all-cause mortality, cause-specific mortality, and self-perceived health, but there are also many unsupportive findings and a number of issues are unresolved.
In regards to methodology, however, there seems to be a consensus that the question ideally should be addressed by simultaneously analysing individual and contextual variables by means of multilevel statistical techniques. Most studies to date have analysed data aggregated to the level of counties, states, or countries. Among the nearly 170 studies reviewed by Wilkinson and Pickett (2006), less than a third had a multilevel design. Results from aggregate-level studies are not necessarily misleading, but the ecological approach does not allow for disentangling the effects of individual-level characteristics such as education and individual income from what is at the centre of interest, namely the contextual effect of income inequality on health outcomes (Lynch et al., 2004; Subramanian & Kawachi, 2004; Wagstaff & van Doorslaer, 2000).
In regards to other issues, questions remain. One issue is the prevalence of the association: does it appear only under certain circumstances or only within particular national contexts? Within USA, both aggregate-level (e.g. Kaplan, Pamuk, Lynch, Cohen, & Balfour, 1996; Kawachi & Kennedy, 1997) and multilevel studies (e.g., Kennedy, Kawachi, Glass, & Prothrow-Stith, 1998; Subramanian, Blakely, & Kawachi, 2003; Subramanian & Kawachi, 2006) have reported findings in line with the hypothesis, contrary to studies in for instance Canada (Ross et al., 2000), New Zealand (Blakely, Atkinson, & O’Dea, 2003), and Japan (Shibuya, Hashimoto, & Yano, 2002). This has been interpreted as indication that the association is only present in inegalitarian countries such as USA (or Chile, see Subramanian, Delgado, Jadue, Vega, & Kawachi, 2003) where income and other social resources such as access to health care are comparatively unequally distributed (Lynch et al., 2004; Subramanian & Kawachi, 2004). Findings from the Nordic countries seem to agree with this proposition. In the Nordic countries, there are comprehensive welfare states and relatively egalitarian income distributions (Fritzell & Lundberg, 2005; Lynch et al., 2004), and neither studies from Denmark (Osler et al., 2002), Sweden (Gerdtham & Johannesson, 2004; Ross et al., 2005), nor Finland (Blomgren, Martikainen, Makela, & Valkonen, 2004; Martikainen, Maki, & Blomgren, 2004), have detected significant associations between income inequality and health.
Another issue is what the appropriate geographic units are for demonstrating associations between area-level income inequality and health outcomes. Wilkinson and Pickett (2006) have strongly argued that the hypothesis cannot be properly tested when income inequality is measured in relatively small areas, because such communities are too small to reflect the scale of social class differences in a society. They substantiate this by showing that a clear majority of studies at the level of countries, US states, and larger regions, corroborate with the hypothesis, while studies at the level of for instance counties or parishes often produce unsupportive results.
A third issue is whether contextual income inequality affects the health of different population subgroups in similar ways. Those relatively few studies which have addressed this topic are mainly from USA, and results are inconsistent. Some suggest that the rich have better health when living in more inegalitarian areas (e.g., Kahn, Wise, Kennedy, & Kawachi, 2000; Subramanian, Kawachi, & Kennedy, 2001), while others find more detrimental health effects of income inequality among those with below-medium or low income (e.g., Kennedy et al., 1998; Lochner, Pamuk, Makuc, Kennedy, & Kawachi, 2001). A recently published US study found few signs of different effects of state income inequality on self-rated health among various population subgroups, and the small observed associations implied more adverse effects of income inequality for relatively advantaged socioeconomic groups (Subramanian & Kawachi, 2006). However, a recent Italian study of mortality, using aggregate-level data, found that the negative effects of higher income inequality occurred primarily in low-income provinces, suggesting that income inequality was more detrimental for disadvantaged population categories (Materia et al., 2005).
Section snippets
Purpose and design of the present study
These issues are addressed by the present study, which analyses register-based population data from Norway by means of multilevel statistical techniques. Norway can be classified as a social-democratic welfare state with a fairly egalitarian social structure (Kautto, Heikkilä, Hvinden, Marklund, & Ploug, 1999), and evidence from this country can be particularly interesting in view of the claim that income inequality effects primarily emerge in societies with relatively inegalitarian income
Data and the classification of regions
The database “FD-Trygd” used for these analyses was constructed by Statistics Norway by linkages of various administrative registers (Akselsen, Dahl, Lajord, & Sivertstøl, 2000). The database includes all inhabitants in Norway registered at 1 January 1993. Data from the population register have been linked to the taxation authority's income register, Statistics Norway's educational register, and other administrative registers, by means of the personal identification number. Also, vital
Results
Model 2 (Table 2) shows that when adjusting for sex and age, the inclusion of the region-level variables adds significantly to the fit of the model, compared to Model 1 (−2LL improvement=12.9, DF=2). The risk of mortality was significantly reduced with higher mean income in the regions, and there was moreover a clear association between higher income inequality and higher mortality (OR=1.034, 95% CI=1.030–1.038) in Model 2. Model 3 includes the full set of individual-level mortality predictors,
Main results
The results of these analyses indicate that when area-level income inequality is calculated with respect to Norwegian economic regions, there was a marked tendency during the 1990s that overall mortality increased with higher regional income inequality. These results were obtained in multilevel analyses after adjustment for region-level mean income and various individual-level mortality predictors. The income inequality effect was higher when adjustment was made only for sex and age than when
Conclusion
This study indicates that when contextual income inequality is measured at the level of Norwegian economic regions, higher levels of regional income inequality were associated with higher mortality, and this was most marked among those located in disadvantaged social positions. These results deviate from the prevailing view that contextual income inequality is unrelated to population health in comparatively egalitarian countries. They suggest that in Norway, neither a fairly egalitarian social
Acknowledgements
We acknowledge Norwegian Social Science Data Services for their assistance in providing data from the database “FD trygd”. Neither Norwegian Social Science Data Services nor Statistics Norway are responsible for the analysis or for the interpretations provided in the paper. The paper is part of the project “Income Inequalities and Mortality”, which was funded by the Norwegian Research Council.
References (46)
- et al.
The effects of regional characteristics on alcohol-related mortality—A register based multilevel analysis of 1.1 million men
Social Science & Medicine
(2004) - et al.
The relationship of income inequality to mortality—Does the choice of indicator matter?
Social Science & Medicine
(1997) Inequalities in health—some international comparisons
European Economic Review
(1987)Income inequality and health: Expanding the debate
Social Science & Medicine
(2000)- et al.
Whose health is affected by income inequality? A multilevel interaction analysis of contemporaneous and lagged effects of state income inequality on individual self-rated health in the United States
Health & Place
(2006) - et al.
Does the state you live in make a difference? Multilevel analysis of self-rated health in the US
Social Science & Medicine
(2001) - et al.
Income inequality and population health: A review and explanation of the evidence
Social Science & Medicine
(2006) - et al.
FD-Trygd. Variabelliste. [“FD-Trygd”—database on social security, etc. List of variables] Notater 2000/70
(2000) - et al.
No association of income inequality with adult mortality within New Zealand: A multilevel study of 1.4 million 25–64 year olds
Journal of Epidemiology and Community Health
(2003) Health, inequality, and economic development
Journal of Economic Literature
(2003)
Associations between relative income and mortality in Norway: A register-based study
European Journal of Public Health
Inequality of income, illiteracy, and medical care as determinants of infant mortality in underdeveloped countries
Population Studies
Fighting inequalities in health and income: One important road to welfare and social development
Absolute income, relative income, income inequality, and mortality
Journal of Human Resources
Standard for økonomiske regioner [Classification of economic regions]. Rapporter 99/6
State income inequality, household income, and maternal mental and physical health: cross sectional national survey
British Medical Journal
Inequality in income and mortality in the United States: Analysis of mortality and potential pathways
British Medical Journal
Social capital, income inequality, and mortality
American Journal of Public Health
Income distribution, socioeconomic status, and self-rated health: A US multi-level analysis
British Medical Journal
State-level income inequality and individual mortality risk: A prospective, multilevel study
American Journal of Public Health
Cited by (67)
The crime rate of five Latin American countries: Does income inequality matter?
2023, International Review of Economics and FinanceRelative deprivation, inequality and the Covid-19 pandemic
2023, Social Science and MedicineRising childhood income inequality and declining Americans’ health
2022, Social Science and MedicineAn exploration of solutions for improving access to affordable fresh food with disadvantaged Welsh communities
2018, European Journal of Operational ResearchShifts in subjective well-being of different status groups: A longitudinal case-study during declining income inequality
2018, Research in Social Stratification and Mobility