scispace - formally typeset
Search or ask a question
Journal Article•DOI•

On the measurement of inequalities in health

Adam Wagstaff1, Pierella Paci2, Eddy van Doorslaer3•
University of Sussex1, City University London2, Erasmus University Rotterdam3
01 Jan 1991-Social Science & Medicine (Soc Sci Med)-Vol. 33, Iss: 5, pp 545-557
TL;DR: It is suggested that only two methods--the slope index of inequality and the concentration index--are likely to present an accurate picture of socioeconomic inequalities in health.
About: This article is published in Social Science & Medicine.The article was published on 1991-01-01 and is currently open access. It has received 1597 citations till now. The article focuses on the topics: Index of dissimilarity & Relative index of inequality.

Summary (4 min read)

Jump to: [I. INTRODUCTION] – [The range] – [Social class] – [The index of dissimilarity] – [The slope and relative indices of inequakty] – [The RII and concentration index compared] – [3. INEQUALITIES IN MORBIDITY] – [Inequalities in morbidity: the Nordic countries compared] – [Trends in inequality in morbidity in Finland] – [Choice of morbidity indicator and the measurement of inequality] – [4. INEQUALITIES IN MORTALITY] – [Trends in inequalities in death: a comparison of Finland, and England and Wales] – [Trena5 in inequalities in potential life lost in England and Wales] – [Choice of mortality indicator and the measurement of inequality] and [5. CONCLUSIONS]

I. INTRODUCTION

  • Much of the recent debate on inequalities of health in Britain and elsewhere dates back to the Black Report [ 11.
  • This led several researchers to investigate trends in inequalities in health using more sophisticated measures of inequality.
  • It is surprising in view of both these radically different findings and the sheer amount of literature on inequalities in health that so little attention has been paid to the question of how health inequality is best measured.
  • The primary objectives of this paper are, therefore, first to provide a critical review of the various measures of inequality that have been employed in the literature on inequalities in health to date and second to identify which measures are best suited to measuring health inequality.
  • *This paper derives from the European Community's COMAC-HSR project on Equity in the Finance and Delivery of Health Care.

The range

  • Its use typically involves comparing the experiences of the top and bottom socioeconomic groups.
  • The authors of the Black Report were, in fact, well aware of this problem, noting, for example, that whilst the position of social class V improved between the 1959-63 and 1970-72 Decennial Supplements, the position of social class IV actually worsened (1, p. 671.
  • This can lead to misleading results when comparisons are performed over time or across countries.
  • Clearly, the same problem arises in the context of cross-country comparisons if the various groups are not all the same size in each of the countries being compared.

Social class

  • Classes I and III enjoy the same mean health status, but members of class II are less healthy.
  • The pseudo-Lorenz curve would also fail to detect a reversal in the class gradient of the type that occurred in mortality from heart disease in the 1950s [12] .
  • In Fig. 4 the class gradient favours the lowest class (as was the case with mortality from coronary heart disease up to the 195Os), whilst in Fig. 5 the gradient favours the highest class (as has been the case with mortality from coronary heart disease since the 1950s).
  • Yet the pseudo-Lorenz curves and pseudo-Gini coefficients for the situations depicted in Figs 4 and 5 are identical.
  • Indeed, they are identical to the Lorenz curve and Gini coefficient for the situation in Fig. 2 .

The index of dissimilarity

  • Then the index of dissimilarity (ID) is where sjh is the jth group's share of the population's health and s,~ is the jth group's population share.
  • The greater the difference between sjh and s,~, the greater the degree of inequality.
  • The ID suffers from the same shortcoming as the pseudo-Lorenz curve discussed above: it is insensitive to the socioeconomic dimension to inequalities in health.
  • What matters in the ID is simply how each socioeconomic group's share of the population's health compares with its population share, not how this disparity compares with the socioeconomic group's socioeconomic status.

The slope and relative indices of inequakty

  • Unlike the Lorenz curve, the pseudo-Lorenz curve and the ID, the slope index of inequality [4,5, 151 and its relative difference counterpart-the relative index of inequality-do reflect the socioeconomic dimension to inequalities in health.
  • The height of each bar in Fig. 6 represents the mean health status of the class in question and the width represents the fraction of the population in the class.
  • It can be interpreted as the absolute effect on health of moving from the lowest socioeconomic group through to the highest ]41.
  • The WLS estimate of the SII can be obtained by using the formula given on p. appropriate weight.
  • It may be verified, for example, that the SII for the data in Figs 2, 4 and 5 are 0.00, -30.00 and +30.00 respectively-a set of results that accords with the notion of inequality as constituting a class gradient.

The RII and concentration index compared

  • This completes their review of the inequality measures employed to date in the inequalities in health literature.
  • It therefore shows the cumulative percent of the population (ranked, as before, by socioeconomic status) graphed against the cumulative amount of health (rather than the share of health).
  • It thus emerges that the RI1 (8/p) is equal to the concentration index divided by twice the variance of the relative rank variable.
  • The latter is arguably more useful as a visual device when performing comparisons across countries or over time.

3. INEQUALITIES IN MORBIDITY

  • The variation of prevalence rates across social classes can be seen from Fig. 10 .
  • The concentration curves, however, tell a quite different story: see Fig. Il .
  • The Swedish curve cuts the British curve from below and is, on average, further from the diagonal than the British curve.
  • This reflects the fact that whilst in Britain the class gradient declines fairly gradually, in Sweden there is relatively little inequality amongst the lower social classes but a substantial gap between the top two classes (which together account for almost 30% of the Swedish population) and the rest of the population.
  • This feature of the Swedish morbidity distribution is, of course, not picked up by the range measure.

Inequalities in morbidity: the Nordic countries compared

  • Figure 12 , based on Table 6 of Ref. [7], shows the variations across income groups in age-standardized chronic sickness rates of persons aged 15-64 in Denmark, Finland and Sweden in 1972.
  • The illness concentration curves corresponding to these data are shown in Fig. 13 .
  • The Norwegian concentration curve, by contrast, crosses the diagonal twice, first from above, reflecting the fact that persons in the fourth income quintile in Norway are apparently more prone to chronic sickness than anybody else except those in the bottom quintile, and then from below, reflecting the low illness rate of the top quintile.
  • This implies that inequalities in health-as measured by chronic sickness-are unambiguously less pronounced in Sweden than in Finland, but more pronounced in Sweden than in Norway.
  • Interestingly, despite the low sickness rate amongst the bottom quintile in Denmark, the overall picture is one of inequality favouring the rich.

Choice of morbidity indicator and the measurement of inequality

  • There is another type of comparison where measures of inequality are useful, namely in comparison across health measures.
  • Finally, their self-assessed health indicator fits in with Blaxter's subjective model, in which ill-health is defined in terms of the individual's perception.
  • This is consistent with Blaxter's [25] results: she found class gradients in most countries in chronic illness and self-perceived health, but detected very little variation across occupational groups in the proportions of persons reporting restricted activity days.
  • In both countries the same picture emerges: the concentration curve lies everywhere above the diagonal for the chronic sickness and health-not-good indicators but crosses the diagonal in the case of the RADs indicator.
  • But they are even more pronounced when one uses self-perceived health as the morbidity indicator.

4. INEQUALITIES IN MORTALITY

  • As with inequalities in morbidity the Black Report [1] did much to stimulate interest in comparative studies of inequalities in mortality.
  • But there has also been a growing interest in crosscountry comparisons [29] .
  • Other work, however, has been based on less reliable measures, such as the range, the pseudo-Lorenx curve and the ID.
  • In this section the authors show how the data in some of these studies can be reworked to obtain the more reliable concentration index measure of inequality.
  • It is interesting to note, however, that the percentage difference between the concentration index values (94.5%) is considerably greater than the percentage difference in the ratio of class IV +.

Trena5 in inequalities in potential life lost in England and Wales

  • That the gap between classes in the risk of dying before the age of 64 is greater in England and Wales than in Sweden does not mean that the gap between classes in the age at death of those dying prematurely is greater in England and Wales than in Sweden.
  • A more attractive measure of mortality that takes into account not only the risk of premature death but also the age at which premature death occurs (if it does) is the number of years of potential lif lost [30].
  • Blane et al. [31] report the results of an analysis of social class inequalities in potential years of life lost in England and Wales.

Choice of mortality indicator and the measurement of inequality

  • The distribution of the population across social classes is taken from fiche CO3 of Ref. [37] .
  • The age-standardized rates of years of potential life lost and age-standardized death rates are taken from Ref. (371 respectively.
  • Figure 23 shows the corresponding concentration curves.
  • Tbis conclusion is, in fact, the same as that reached by Blane et al. [31] on the basis of a comparison of the class V:I ratios of years of potential life lost and SMRs.

5. CONCLUSIONS

  • It is, of course, not always the case that different inequality measures lead to different conclusions.
  • The authors found, for example, that irrespective of whether one uses the range measure of inequality or the concentration index, inequalities in mortality in England and Wales in 198 1 were more pronounced if mortality is measured by years of potential life lost than if measured simply by deaths.

Did you find this useful? Give us your feedback

Figures (7)
Citations
More filters
Journal Article•DOI•
Economics and Information Theory

[...]

G. R. Fisher1•
University of Southampton1
01 May 1970

1,935 citations

Journal Article•DOI•
Socioeconomic factors and cardiovascular disease: a review of the literature.

[...]

George A. Kaplan1, J E Keil1•
American Heart Association1
01 Oct 1993-Circulation
TL;DR: There has been a consistent inverse relation between cardiovascular disease, primarily coronary heart disease, and many of the indicators of SES, and evidence for this relation has been derived from prevalence, prospective, and retrospective cohort studies.
Abstract: Despite recent declines in mortality, cardiovascular diseases are the leading cause of death in the United States today. It appears that many of the major risk factors for coronary disease have been identified. Researchers are still learning about different modifiable factors that may influence cardiovascular diseases. Socioeconomic status may provide a new focus. The principal measures of SES have been education, occupation, and income or combinations of these. Education has been the most frequent measure because it does not usually change (as occupation or income might) after young adulthood, information about education can be obtained easily, and it is unlikely that poor health in adulthood influences level of education. However, other measures of SES have merit, and the most informative strategy would incorporate multiple indicators of SES. A variety of psychosocial measures--for example, certain aspects of occupational status--may be important mediators of SES and disease. The hypothesis that high job strain may adversely affect health status has a rational basis and is supported by evidence from a limited number of studies. There is a considerable body of evidence for a relation between socioeconomic factors and all-cause mortality. These findings have been replicated repeatedly for 80 years across measures of socioeconomic level and in geographically diverse populations. During 40 years of study there has been a consistent inverse relation between cardiovascular disease, primarily coronary heart disease, and many of the indicators of SES. Evidence for this relation has been derived from prevalence, prospective, and retrospective cohort studies. Of particular importance to the hypothesis that SES is a risk factor for cardiovascular disease was the finding by several investigators that the patterns of association of SES with coronary disease had changed in men during the past 30 to 40 years and that SES has been associated with the decline of coronary mortality since the mid-1960s. However, the declines in coronary mortality of the last few decades have not affected all segments of society equally. There is some evidence that areas with the poorest socioenvironmental conditions experience later onset in the decline in cardiovascular mortality. A number of studies suggest that poor living conditions in childhood and adolescence contribute to increased risk of arteriosclerosis. Some of these studies have been criticized because of their nature, and others for inadequate control of confounding factors.(ABSTRACT TRUNCATED AT 400 WORDS)

1,829 citations

Journal Article•DOI•
The increasing disparity in mortality between socioeconomic groups in the United States 1960 and 1986.

[...]

Gregory Pappas1, Susan Queen, W C Hadden, Gail Fisher•
National Center for Health Statistics1
08 Jul 1993-The New England Journal of Medicine
TL;DR: Despite an overall decline in death rates in the United States since 1960, poor and poorly educated people still die at higher rates than those with higher incomes or better educations, and this disparity increased between 1960 and 1986.
Abstract: Background There is an inverse relation between socioeconomic status and mortality. Over the past several decades death rates in the United States have declined, but it is unclear whether all socioeconomic groups have benefited equally. Methods Using records from the 1986 National Mortality Followback Survey (n = 13,491) and the 1986 National Health Interview Survey (n = 30,725), we replicated the analysis by Kitagawa and Hauser of differential mortality in 1960. We calculated direct standardized mortality rates and indirect standardized mortality ratios for persons 25 to 64 years of age according to race, sex, income, and family status. Results The inverse relation between mortality and socioeconomic status persisted in 1986 and was stronger than in 1960. The disparity in mortality rates according to income and education increased for men and women, whites and blacks, and family members and unrelated persons. Over the 26-year period, the inequalities according to educational level increased for whites an...

1,517 citations

Journal Article•DOI•
Defining equity in health

[...]

Paula Braveman1, Sofia Gruskin•
University of California, San Francisco1
01 Apr 2003-Journal of Epidemiology and Community Health
TL;DR: The proposed definition of equity supports operationalisation of the right to the highest attainable standard of health as indicated by the health status of the most socially advantaged group, which is essential to wellbeing and to overcoming other effects of social disadvantage.
Abstract: Study objective: To propose a definition of health equity to guide operationalisation and measurement, and to discuss the practical importance of clarity in defining this concept. Design: Conceptual discussion. Setting, Patients/Participants, and Main results: not applicable. Conclusions: For the purposes of measurement and operationalisation, equity in health is the absence of systematic disparities in health (or in the major social determinants of health) between groups with different levels of underlying social advantage/disadvantage that is, wealth, power, or prestige. Inequities in health systematically put groups of people who are already socially disadvantaged (for example, by virtue of being poor, female, and/or members of a disenfranchised racial, ethnic, or religious group) at further disadvantage with respect to their health; health is essential to wellbeing and to overcoming other effects of social disadvantage. Equity is an ethical principle; it also is consonant with and closely related to human rights principles. The proposed definition of equity supports operationalisation of the right to the highest attainable standard of health as indicated by the health status of the most socially advantaged group. Assessing health equity requires comparing health and its social determinants between more and less advantaged social groups. These comparisons are essential to assess whether national and international policies are leading toward or away from greater social justice in health.

1,423 citations

Journal Article•DOI•
Socioeconomic inequalities in health: Measurement, computation, and statistical inference

[...]

Nanak Kakwani1, Adam Wagstaff2, Eddy van Doorslaer3•
University of New South Wales1, University of Sussex2, Erasmus University Rotterdam3
01 Mar 1997-Journal of Econometrics
TL;DR: In this article, the relationship between two widely used indices of health inequality and explain why these are superior to others indices used in the literature is explained and the role that demographic standardization plays in the analysis of socioeconomic inequalities in health.

1,250 citations

References
More filters
Journal Article•DOI•
Elements of Econometrics.

[...]

Ian Domowitz, Jan Kmenta
01 Mar 1988-Journal of the American Statistical Association
TL;DR: The Elements of Econometrics as mentioned in this paper is a textbook for upper-level undergraduate and master's degree courses and may usefully serve as a supplement for traditional Ph.D. courses in economics.
Abstract: This classic text has proven its worth in university classrooms and as a tool kit in research--selling over 40,000 copies in the United States and abroad in its first edition alone. Users have included undergraduate and graduate students of economics and business, and students and researchers in political science, sociology, and other fields where regression models and their extensions are relevant. The book has also served as a handy reference in the "real world" for people who need a clear and accurate explanation of techniques that are used in empirical research.Throughout the book the emphasis is on simplification whenever possible, assuming the readers know college algebra and basic calculus. Jan Kmenta explains all methods within the simplest framework, and generalizations are presented as logical extensions of simple cases. And while a relatively high degree of rigor is preserved, every conflict between rigor and clarity is resolved in favor of the latter. Apart from its clear exposition, the book's strength lies in emphasizing the basic ideas rather than just presenting formulas to learn and rules to apply.The book consists of two parts, which could be considered jointly or separately. Part one covers the basic elements of the theory of statistics and provides readers with a good understanding of the process of scientific generalization from incomplete information. Part two contains a thorough exposition of all basic econometric methods and includes some of the more recent developments in several areas.As a textbook, "Elements of Econometrics" is intended for upper-level undergraduate and master's degree courses and may usefully serve as a supplement for traditional Ph.D. courses in econometrics. Researchers in the social sciences will find it an invaluable reference tool.A solutions manual is also available for teachers who adopt the text for coursework.Jan Kmenta is Professor Emeritus of Economics and Statistics, University of Michigan.

3,838 citations

Book•
Elements of econometrics

[...]

Jan Kmenta
01 Jan 1971
TL;DR: The emphasis is on simplification whenever possible, assuming the readers know college algebra and basic calculus, and Jan Kmenta explains all methods within the simplest framework, and generalizations are presented as logical extensions of simple cases.
Abstract: This classic text has proven its worth in university classrooms and as a tool kit in research--selling over 40,000 copies in the United States and abroad in its first edition alone. Users have included undergraduate and graduate students of economics and business, and students and researchers in political science, sociology, and other fields where regression models and their extensions are relevant. The book has also served as a handy reference in the "real world" for people who need a clear and accurate explanation of techniques that are used in empirical research.Throughout the book the emphasis is on simplification whenever possible, assuming the readers know college algebra and basic calculus. Jan Kmenta explains all methods within the simplest framework, and generalizations are presented as logical extensions of simple cases. And while a relatively high degree of rigor is preserved, every conflict between rigor and clarity is resolved in favor of the latter. Apart from its clear exposition, the book's strength lies in emphasizing the basic ideas rather than just presenting formulas to learn and rules to apply.The book consists of two parts, which could be considered jointly or separately. Part one covers the basic elements of the theory of statistics and provides readers with a good understanding of the process of scientific generalization from incomplete information. Part two contains a thorough exposition of all basic econometric methods and includes some of the more recent developments in several areas.As a textbook, "Elements of Econometrics" is intended for upper-level undergraduate and master's degree courses and may usefully serve as a supplement for traditional Ph.D. courses in econometrics. Researchers in the social sciences will find it an invaluable reference tool.A solutions manual is also available for teachers who adopt the text for coursework.Jan Kmenta is Professor Emeritus of Economics and Statistics, University of Michigan.

3,096 citations

Book•
On Economic Inequality

[...]

Amartya Sen, James E. Foster
01 Jan 1973
TL;DR: In this paper, Amartya Sen relates the theory of welfare economics to the study of economic inequality and presents a systematic treatment of the conceptual framework as well as the practical problems of measurement of inequality.
Abstract: In this classic text, first published in 1973, Amartya Sen relates the theory of welfare economics to the study of economic inequality. He presents a systematic treatment of the conceptual framework as well as the practical problems of measurement of inequality. In his masterful analysis, Sen assesses various approaches to measuring inequality and delineates the causes and effects of economic disparities. Containing the four lectures from the original edition as well as a new introduction, this timeless study is essential reading for economists, philosophers, and social scientists. In a new introduction, Amartya Sen, jointly with James Foster, critically surveys the literature that followed the publication of this book, and also evaluates the main analytical issues in the appraisal of economic inequality and poverty.

2,826 citations

Book•
Economics and information theory

[...]

Henri Theil
01 Jan 1967

2,245 citations

Journal Article•DOI•
Economics and Information Theory

[...]

G. R. Fisher1•
University of Southampton1
01 May 1970

1,935 citations

Related Papers (5)
Socioeconomic inequalities in health: Measurement, computation, and statistical inference

[...]

01 Mar 1997-Journal of Econometrics
Nanak Kakwani, Adam Wagstaff, Eddy van Doorslaer
On decomposing the causes of health sector inequalities with an application to malnutrition inequalities in Vietnam

[...]

01 Jan 2003-Journal of Econometrics
Adam Wagstaff, Adam Wagstaff, Eddy van Doorslaer, Naoko Watanabe 
Measuring the magnitude of socio-economic inequalities in health: an overview of available measures illustrated with two examples from Europe.

[...]

01 Mar 1997-Social Science & Medicine
Johan P. Mackenbach, Anton E. Kunst
Correcting the Concentration Index

[...]

01 Mar 2009-Journal of Health Economics
Guido Erreygers
Income-related inequalities in health: some international comparisons.

[...]

01 Feb 1997-Journal of Health Economics
E. van Doorslaer, Adam Wagstaff, Han Bleichrodt, Samuel Calonge, Ulf-G Gerdtham, Michael Gerfin, J Geurts, L Gross, Unto Häkkinen, Robert E. Leu, Owen O'Donnell, Carol Propper, Frank Puffer, Marisol Rodríguez, Gun Sundberg, Olaf Winkelhake 
Frequently Asked Questions (6)
Q1. What contributions have the authors mentioned in the paper "On the measurement of inequalities in health*" ?

This paper offers a critical appraisal of the various methods employed to date to measure inequalities in health. The paper also presents several empirical examples lo illustrate of the dangers of using other measures such as the range, the Lorenz curve and the index of dissimilarity. It suggests that only two of these -- the slope index of inequality and the concentration index-are likely to present an accurate picture of socioeconomic inequalities in health. 

Of the indices reviewed only the SII, RI1 and concentration index meet what would appear to be the minimal requirements of an inequality measure in this context: (i) that it reflect the socioeconomic dimension to inequalities in health; (ii) that it reflect the experiences of the entire population (rather than just, say, social classes The authorand V); and (iii) that it be sensitive to changes in the distribution of the population across socioeconomic groups. 

In this section the authors show how the data in some of these studies can be reworked to obtain the more reliable concentration index measure of inequality. 

despite the low sickness rate amongst the bottom quintile in Denmark, the overall picture is one of inequality favouring the rich. 

the concentration curve corresponding to Fig. 2 crosses the diagonal at the 50% point, resulting in a concentration index value of 0.000. 

Some idea of the extent of inequality in these countries can be gleaned from the concentration index values: Denmark: - 0.048; Finland: -0.115; Norway: -0.024; Sweden: -0.072.