Concentration curve an introduction to .

Concentration curve an introduction to .

1. Inequalities in Health and Healthcare
Concentration Curves
Owen O’Donnell
Erasmus University Rotterdam
& Tinbergen Institute
1

2. Outline
•Group mean comparisons (limitations)
•Concentration curves –definitions & properties
•Inference for concentration curves (dominance)

3. Goal: measure socioeconomic-related health inequality
•Degree to which health differs across individuals distinguished by
socioeconomic status (SES)
•As first pass, compare mean health across SES groups

4. How to compare means?
•Relatively,
???? ?????? ?
???? ?????? ?
•Absolutely,
•Bothidentifyinequalityb/wgroupsinsinglepopulation
(ratio≠1difference≠0)
•Butcanidentifydifferentpopulationasexhibitinggreaterinequality
4

5. Did inequality in mortality decrease or increase
in France?

6. Period
1990s2000sChange
Education
low 1099939-160
high 509412-97
difference590527
Age-standardized mortality rate by education
Did inequality in mortality
decrease or increase in France?

7. Period
1990s2000s change
Education
low 1099939 -14.6%
high 509412 -19.1%
ratio2.162.28
Age-standardized mortality rate by education
Did inequality in mortality
decrease or increase in France?

8. Period Change
1990s2000sabsoluteproportionate
Education
low 1099939-160-14.6%
high 509412-97-19.1%
difference590527
ratio2.162.28
Age-standardized mortality rate by education
Did inequality in mortality
decrease or increase in France?

9. Views on inequality
•Absolute invariance: inequality unchanged if health changes by
amountkthroughout the distribution of SES
•Relative invariance: inequality unchanged if health changes by
proportionkthroughout the distribution of SES

10. Limitations of group mean comparisons
•Rate differences and ratios compare two groups
•With categorical SES, ignore middle categories

11. Limitations of group mean comparisons
•Rate differences and ratios compare two groups
•With categorical SES, ignore
•middle categories
•changes in size of categories

12. Life Expectancy At Birth, By Years Of Education At Age 25 For US White Females, 1990–2008.
S. Jay Olshanskyet al. Health Affairs
2012;31:1803-1813
©2012 by Project HOPE -The People-to-People Health Foundation, Inc.

13. But hang on, low education
group is shrinking dramatically

15. Not comparing like with like
•Smaller low education group likely to be different in composition
•Least healthy left behind
•Examine trends in mortality rate in bottom quartile of education
distribution

16. Bound et al. (2015)

17. Limitations of group mean comparisons
•Rate differences and ratios compare two groups
•With categorical SES, ignore
•middle categories
•changes in size of categories
•When make groups from continuous SES, ignore within group inequality
•Do not capture all variation of health across distribution of SES

18. Prevalence of TB by expenditure quintile,
the Philippines, 2002
0 .01 .02 .03
proportion with tb
poorest 20%2nd poorestmiddle2nd richestrichest 20%
Philippines World Health Survey 2002

19. Conclusions
Absolute & relative invariance
different views on what constitutes increase in inequality
Group mean comparisons ignore
•middle of distribution
•size (/composition) of groups

20. Question for discussion
Which do you believe is the more ethically appealing principle
on which to base the measurement of health inequality –
absolute invariance or relative invariance?

21. References
•Bound,J. et al. (2015). Measuring Recent Apparent Declines In Longevity: The
Role Of Increasing Educational Attainment. Health Affairs34:12,2167-2173
•Mackenbach, JP et al (2015). Trends in inequalities in premature mortality: a
study of 3.2 million deaths in 13 European countries. Journal of Epidemiology
and Community Health69:207–217.
•Olshansky, S.J., Antonucci, T., Berkman, L. et al. (2012). Differences in Life
Expectancy Due to Race and Educational Differences are Widening, and Many
May Not Catch Up. Health Affairs31(8):1803-1813
22