The Nordic welfare states aim at providing equality of the highest standards for all their citizens. However, numerous studies have demonstrated that socioeconomic inequalities in morbidity and mortality are not among the smallest in these countries as compared with other European regions.1–7 Recently, this has spurred health researchers to evaluate the extent to which the Nordic welfare regime is capable of diminishing socioeconomic health inequalities.8,9 After all, the conclusion that the Nordic welfare regime does not succeed in reducing health inequalities would have serious implications for health policy world wide. In this commentary, we aim at evaluating why the Nordic welfare regime does not completely succeed in reducing socioeconomic inequalities in health, despite its egalitarian nature. Our presentation is divided into three types of explanations: causality, social selectivity and artefacts.

Scholars have generally argued that a distinction should be made between relative and absolute socioeconomic health inequalities. Moreover, the absolute health status of the weakest socioeconomic groups (e.g. manual workers) is considered to be most important as a marker of the ability of welfare regimes to reduce socioeconomic disparities in health. Using this distinction, it was concluded that whereas the Nordic countries perform only intermediately whenever relative inequalities are considered, they have lower absolute inequalities and a higher average absolute health status of the lowest socioeconomic groups as compared with other European societies (although this applies mostly to Sweden and Norway, and only to a lesser extent to Denmark and Finland). Since the authors argue that welfare regime performance should mainly be evaluated through absolute measures of health status and health inequalities, they assert that Nordic welfare regimes do not perform that poorly after all.

Unfortunately, this conclusion has put the issue of identifying why the Nordic countries do not have the smallest socioeconomic health inequalities in the background. We agree that measures of relative health inequalities should be accompanied by an examination of absolute health status and health inequalities,10 but a sole focus on the absolute measures in evaluating the Nordic welfare regimes seems unwarranted. After all, apart from improving levels of welfare and well-being in general, the Nordic welfare regimes explicitly aim at reducing differences between social groups. Thus, even though the weakest socioeconomic groups may be better off in the Nordic countries as compared with other European regions, the mere fact that there is a substantial health gap between those who are most disadvantaged and the higher socioeconomic groups implies that the Nordic welfare regime does not succeed in part of its mission. Therefore, to adequately evaluate the performance of the Nordic welfare regime, relative and absolute measures should be examined jointly, and the conclusion that the Nordic countries perform better than other European societies is premature.

As a result, from a health policy perspective, an appropriate evaluation of why the Nordic welfare regime does not completely succeed in reducing socioeconomic inequalities requires further examination of the exact mechanisms that cause this seemingly paradoxical finding. Although possible mechanisms are suggested in the studies mentioned, a comprehensive overview is still lacking (which may be partly due to these authors considering relative inequalities to be less relevant).

Initially, we would like to address three possible causal influences of the egalitarian Nordic welfare regime type itself. First, it may be the case that adversity is especially damaging when living in a prosperous society, which also claims to be egalitarian and meritocratic. People with a lower socioeconomic position do not only perceive their personal situation as potentially being worse compared with other groups, but they may also feel that there is little chance of improvement, since they have apparently failed to make good use of the possibilities that have been offered to them. Secondly, even in case of no mobility, mortality and morbidity are decreasing more pronouncedly among high status groups compared with low status groups in the Nordic countries; everyone benefits (in an absolute sense) from egalitarianism, but higher social groups are more able to do so than lower social groups, for instance by making better use of the medical system. Thirdly, health differences between social groups may be relatively large in the Nordic countries because of mortality selection. This implies that whereas the frailest part of the population dies during childhood in other European regions, the quality of the medical system in the Nordic countries allows these people to survive. However, they generally do so in poor health. Since these people may be overrepresented in lower as compared with higher social groups, social health inequalities would become larger when frailer people survive.

The unexpectedly strong relative socioeconomic health inequalities in the Nordic countries may also be caused by social selectivity, instead of causation—in at least two ways. First, the socioeconomic gradient in smoking prevalence is much steeper in the Nordic countries as compared with countries in the South. It may be more culturally and normatively accepted to smoke among higher social groups, whereas this may be less true for the Nordic countries. As a result, smokers are a more socially selective group in the Nordic countries as compared with other societies. Secondly, in a related way, like most other countries in Western Europe, the Nordic countries are faced with a relatively large influx of immigrants. In general, these immigrants (especially those from non-Western societies) often have a low socioeconomic position and also encounter problems in obtaining access to health services. Additionally, knowledge on healthy lifestyles is less widespread among immigrants, and immigrants are generally less healthy than the native population. Given that both the average health and socioeconomic position are relatively high in the Nordic countries, immigrants in these societies form a socially more selective group as compared with other countries. As a result, the impact of the immigrant population on socioeconomic inequalities in health may be stronger in the Nordic countries, leading to a relatively steep social gradient in health.

The paradoxical position of the Nordic countries may be the result of a statistical artefact. First, it has been argued that variations in health inequalities can be explained by a mathematical rule rather than by substantial interpretations. However, it should be confirmed by other cross-national studies whether or not it would be an overestimation to entirely attribute cross-national health differences to this type of artefact. Secondly, it has been argued that variations between countries in the level of self-assessed social health inequalities may be due to cross-cultural variations in the self-report of health. However, little is known at this stage about the extent to which cross-cultural variation in reporting styles also affects cross-national differences in social health inequalities.

These three types of explanations would have radically different implications for the evaluation of the Nordic welfare regime. Finding support for causation mechanisms would imply that the Nordic welfare regime itself is responsible for socioeconomic health inequalities being only intermediate. From a policy perspective, this would mean that, although the universalistic and redistributive approach taken in the Nordic countries makes positive overall health outcomes, it does not sufficiently reduce the relative inequalities in health. If the counterintuitive position of the Nordic countries is mainly caused by selectivity issues, however, changing welfare policy would not result in any improvement. In case of artefacts being most prominent as an explanation, policy modifications would not only be futile, but also unnecessary. Therefore, after this brief systematization of explanations, we call for an elaborate further examination of this issue in empirical research. In doing so, specific attention should be paid to different indicators of people's socioeconomic position (i.e. education, occupation or class, and income), and both relative and absolute inequalities should be considered, as well as the absolute health status of both the weakest and strongest socioeconomic groups.

Conflicts of interest: None declared.

References

1
Cavelaars
AEJM
Kunst
AE
Geurts
JJM
, et al.  . 
Differences in self reported morbidity by educational level: a comparison of 11 western European countries
J Epidemiol Commun Health
 , 
1998
, vol. 
52
 (pg. 
219
-
27
)
2
Eikemo
TA
Health inequalities in European welfare states. Evidence from the European Social Survey
 , 
2009
Saarbrücken
VDM Publishing House
3
EUROTHINE
2007
Rotterdam
Department of Public Health, University Medical Centre Rotterdam
4
Kunst
AE
Bos
V
Lahelma
E
, et al.  . 
Trends in socioeconomic inequalities in self-assessed health in 10 European countries
Int J Epidemiol
 , 
2005
, vol. 
34
 (pg. 
295
-
305
)
5
Mackenbach
JP
Bos
V
Andersen
O
, et al.  . 
Widening socioeconomic inequalities in mortality in six Western European countries
Int J Epidemiol
 , 
2003
, vol. 
32
 (pg. 
830
-
7
)
6
Mackenbach
JP
Kunst
AE
Cavelaars
AEJM
, et al.  . 
Socioeconomic inequalities in morbidity and mortality in western Europe
Lancet
 , 
1997
, vol. 
349
 (pg. 
1655
-
9
)
7
Mackenbach
JP
Stirbu
I
Roskam
AJ
, et al.  . 
Socioeconomic inequalities in health in 22 European countries
N Engl J Med
 , 
2008
, vol. 
358
 (pg. 
2468
-
81
)
8
Dahl
E
Fritzell
J
Lahelma
E
, et al.  . 
Siegrist
J
Marmot
M
Welfare state regimes and health inequalities
Social inequalities in health
 , 
2006
Oxford
Oxford University Press
(pg. 
193
-
222
)
9
Lundberg
O
Åberg Yngwe
M
Kölegård Stjärne
M
Björk
L
Fritzell
J
The Nordic Experience: Welfare States and Public Health (NEWS)
 , 
2008
Stockholm
Centre for Health Equity Studies
10
Vågerö
D
Erikson
R
Socioeconomic inequalities in morbidity and mortality in western Europe
Lancet
 , 
1997
, vol. 
350
 (pg. 
516
-
7
)

Comments

1 Comment
Understanding the forces driving cross-national variations in relative differences in outcome rates
23 January 2010
James P. Scanlan

In exploring why socioeconomic inequalities in health are not smallest in Nordic welfare states, Huijts and Eikemo [1] note that "it has been argued that variations in health inequalities can be explained by a mathematical rule rather than by substantial interpretations," but that "it should be confirmed by other cross-national studies whether or not it would be an overestimation to entirely attribute cross-national health differences to this type of artefact." The reference to a mathematical rule is apparently an allusion to the pattern I have described in various places whereby the rarer an outcome, the greater tends to be the relative difference in experiencing it and the smaller tends to be the relative difference in avoiding it, which Eikemo has discussed more directly elsewhere.[2]

In many places, I have argued that the perception that health inequalities in Norway and Sweden are comparatively large among western European nations (a perception significantly based on the comparatively large relative socioeconomic differences in mortality and morbidity in these countries found in a 1997 study by Mackenbach et al.[3]) failed to consider that large relative difference in mortality and other adverse outcomes are to be expected in Norway and Sweden simply because adverse health outcome rates are low in these countries. In addition to references 4-9, discussion of the bearing of the above-described statistical pattern on the interpretation of Nordic health inequalities may be found in Section E.2 of the Measuring Health Disparities page of jpscanlan.com [10] and Section B.6 of the Scanlan's Rule page the same site.[11]

The point, however, is not that patterns of relative differences in mortality or other adverse outcome rates in Norway and Sweden are solely a function of a mathematical rule or that there is any realistic possibility that cross-national variations in relative differences in adverse outcome rates generally might be entirely attributed to a statistical artefact. Rather, as perhaps best explained in references 5 and 12, differences in outcome rates – whether measured in terms of relative differences in experiencing the outcome, relative differences in avoiding the outcome, absolute differences between rates, or odds ratios – will invariably be a function of (a) the difference between the means of the underlying distributions of risks of experiencing the outcome and (b) the overall prevalence of the outcome. The point can be well illustrated with hypothetical test score data. In many places, I have shown how the lowering of a cutoff will tend to increase relative differences in failure rates and reduce relative differences in pass rates (with absolute differences and odds ratios affected in a more complicated, but still systematic, manner). Consider the situation where in setting A the means scores of an advantaged group and a disadvantaged group differ by 0.5 standard deviations and in setting B the means scores differ by 0.4 standard deviations. At every point identified by a particular failure rate for the advantaged group, the relative difference in pass rates and the relative difference in failure rates (as well as the absolute difference between rates and the difference measured by the odds ratio) will be greater in setting A than in setting B. But lowering the cutoff in setting B will tend to increase the relative difference in failure rates in that setting (perhaps, but not necessarily, enough to make such difference larger than in setting A) while reducing the relative difference in pass rates (further increasing the extent to which the relative difference in pass rates is greater in setting A than in setting B). (While the directions of changes in absolute differences between rates and differences measured by odds ratios will vary depending on the rates at issue, such directions will similarly tend to be the opposite of one another.[11]) A useful analysis of the comparative size of health inequalities must involve determining the comparative size of the difference between the underlying means in the two settings (whether the settings are different countries, the same country at different points in time, or subpopulations within a country). The Solutions sub-page of the Measuring Health Disparities page of jpscanlan.com [13] provides a method for attempting to make that determination when the only information available is the outcome rates for the groups being studied.

The existence of the pattern described in the first paragraph above does not resolve the issue of the comparative size of health inequalities in Norway and Sweden vis a vis other developed countries. Table 4c of reference 8 and Table 8 (slide 16) of reference 14, which apply the approach just mentioned to data from the article to which reference 8 responds, do not necessarily show the smallest mortality inequalities in Norway and Sweden. Assuming the validity of the approach, as discussed in reference 8, there still would exist questions about whether the advantaged and disadvantaged groups being compared in one setting sufficiently correspond to the advantaged and disadvantaged groups being compared in another setting.

Depending on how inequalities in Norway and Sweden compare with those in other countries once such issue is resolved, it might well be useful to explore the potential role of various factors discussed by Huijts and Eikemo. But it is only when one has a sound basis for appraising the size of inequalities in the first place – either through the methodology I have proposed or one like (or superior to) it – that there will be much value in exploring the reasons for perceived cross-national variations in health inequalities. And, of course, the exploration of such reasons must as well be undertaken with an understanding of the extent to which such things as perceived differences in the socioeconomic gradients in smoking (or non-smoking) are influenced by the overall smoking level.[8]

As to the further exploring the role of the above-described statistical tendencies in additional cross-national studies (on the order of studies that have so far been done by Eikemo at al.[2] and Houweling et al.[15]), it is inconceivable that such studies will show such tendencies to explain all variation. Invariably the statistical forces I have described will influence observed patterns of rate differences in each setting, as will variations in the magnitude of differences between the risk distributions of the groups being compared (sometimes, possibly, with a systematic interaction between the two influences [16]). But whatever resources are devoted to such studies, they should be undertaken with a sound understanding of the forces (statistical and otherwise) that will drive observed patterns of rate differences.

Finally, allowing that the term "artefact" can be variously interpreted, I question whether it has an appropriate place in this context. As discussed, lowering a cutoff will tend to increase relative differences in failure rates while reducing relative differences in pass rates. Similarly, reducing poverty will tend to increase relative differences in poverty rates while reducing relative differences in rates of avoiding poverty.[4,12] Such changes are the standard consequence of reducing the prevalence of one outcome while increasing the prevalence of the opposite outcome. The existence of this pattern - or, put another way, that each relative difference tends to be influenced by the prevalence of the outcome - is a reason why changes in either relative difference ought not to be interpreted as necessarily indicating a meaningful change in the comparative situation of the groups being studied. But I am uncertain that it would make sense to regard the failure of either of these measures - or absolute differences or odds ratios - to provide sound information on whether there have been meaningful changes in the comparative situation of the subject groups as involving an artefact. The same, I think, applies to discussions of comparative size of health inequalities in different nations.

References:

1. Huijts T, Eikemo TA. Causality, social selectivity or artefacts? Why socioeconomic inequalities in health are not smallest in the Nordic countries. Eur J Pub Health 2009;19:452-53.

2. Eikemo TA, Skalicka V, Avendano M. Variations in health inequalities: are they a mathematical artifact? International Journal for Equity in Health 2009;8:32: http://www.equityhealthj.com/content/pdf/1475- 9276-8-32.pdf

3. Mackenbach JP, Kunst AE, Cavelaars AEJM, et al. Socioeconomic inequalities in morbidity and mortality in western Europe. Lancet 1997;349:1655–9.

4. Scanlan JP. Can we actually measure health disparities? Chance 2006:19(2):47-51: http://www.jpscanlan.com/images/Can_We_Actually_Measure_Health_Disparities.pdf

5. Scanlan JP. The Misinterpretation of Health Inequalities in the United Kingdom, presented at the British Society for Populations Studies Conference 2006, Southampton, England, Sept. 18-20, 2006: http://www.jpscanlan.com/images/BSPS_2006_Complete_Paper.pdf

6. Scanlan JP. The Misinterpretation of Health Inequalities in Nordic Countries, presented at: 5th Nordic Health Promotion Research Conference, Esbjerg, Denmark, June 15-17, 2006: http://www.jpscanlan.com/images/Esbjerg_Oral.pdf

7. Scanlan JP. Explanation for large health inequalities in Nordic countries. Eur J Public Health Nov. 1, 2006 (responding to Hemmingsson T, Lundberg I. Can large relative mortality differences between socioeconomic groups among Swedish men be explained by risk indicator- associated social mobility? Eur J Public Health 200515:518 -522): http://eurpub.oxfordjournals.org/cgi/eletters/15/5/518#22

8. Scanlan JP. Comparing the size of inequalities in dichotomous measures in light of the standard correlations between such measures and the prevalence of an outcome. Journal Review Jan. 14, 2008 (responding to Boström G, Rosén M. Measuring social inequalities in health – politics or science? Scan J Public Health 2003;31:211-215): http://journalreview.org/v2/articles/view/12850975.html

9. Scanlan JP. Reconsidering a landmark study. Lancet Feb. 25, 2008 (responding to Mackenbach, JP, Kunst, AE, Cavelaars, et al. Socioeconomic inequalities in morbidity and mortality in western Europe, Lancet 1997; 349: 1655-59): http://journalreview.org/v2/articles/view/9186383.html

10. Measuring Health Disparities page of jpscanlan.com: http://jpscanlan.com/measuringhealthdisp.html

11. Scanlan’s Rule page of jpscanlan.com: http://jpscanlan.com/scanlansrule.html

12. Scanlan JP. Race and mortality. Society 2000;37(2):19-35 (reprinted in Current 2000 (Feb)): http://www.jpscanlan.com/images/Race_and_Mortality.pdf

13. Solutions sub-page of Measuring Health Disparities page of jpscanlan.com: http://www.jpscanlan.com/measuringhealthdisp/solutions.html

14. Scanlan JP. Measures of Health Inequalities that are Unaffected by the Prevalence of an Outcome, presented at the 16th Nordic Demographic Symposium, Helsinki, Finland, June 5-7, 2008: http://jpscanlan.com/images/Scanlan_JP_NDS_Presentation_2R.ppt

15. Houweling TAJ, Kunst AE, Huisman M, Mackenbach JP. Using relative and absolute measures for monitoring health inequalities: experiences from cross-national analyses on maternal and child health. International Journal for Equity in Health 2007;6:15: http://www.equityhealthj.com/content/6/1/15

16. Scanlan JP. The relationship between overall prevalence and measures of differences between outcome rates. International Journal for Equity in Health __________ (responding to Eikemo TA, Skalicka V, Avendano M. Variations in health inequalities: are they a mathematical artifact? International Journal for Equity in Health 2009;8:32: http://www.equityhealthj.com/content/pdf/1475-9276-8-32.pdf): http://jpscanlan.com/images/Comment_on_Eikemo_et_al..pdf

Conflict of Interest:

None declared

Submitted on 23/01/2010 7:00 PM GMT