Policy Forum

Priorities for Research on Equity and Health: Towards an Equity-Focused Health Research Agenda

  • Piroska Östlin mail,

    Affiliation: World Health Organization Regional Office for Europe, Copenhagen, Denmark

  • Ted Schrecker,

    Affiliation: Department of Epidemiology and Community Medicine and Institute of Population Health, University of Ottawa, Ottawa, Canada

  • Ritu Sadana,

    Affiliation: World Health Organization, Geneva, Switzerland

  • Josiane Bonnefoy,

    Affiliation: School of Public Health, University of Chile, Santiago, Chile

  • Lucy Gilson,

    Affiliation: University of Cape Town, Cape Town, South Africa, and London School of Hygiene and Tropical Medicine, London, United Kingdom

  • Clyde Hertzman,

    Affiliation: Human Early Learning Partnership (HELP), University of British Columbia, Vancouver, Canada

  • Michael P. Kelly,

    Affiliation: Centre for Public Health Excellence, National Institute for Health and Clinical Excellence, London, United Kingdom

  • Tord Kjellstrom,

    Affiliation: National Centre for Epidemiology and Population Health, Australian National University, Canberra, Australia

  • Ronald Labonté,

    Affiliation: Department of Epidemiology and Community Medicine and Institute of Population Health, University of Ottawa, Ottawa, Canada

  • Olle Lundberg,

    Affiliation: Centre for Health Equity Studies, Stockholm, and Department of Health Sciences, Mid Sweden University, Östersund, Sweden

  • Carles Muntaner,

    Affiliation: Social Equity and Health Section, Centre for Addiction and Mental Health and Bloomberg Faculty of Nursing and Dalla Lana School of Public Health, University of Toronto, Toronto, Canada

  • Jennie Popay,

    Affiliation: Division of Health Research, Lancaster University, Lancaster, United Kingdom

  • Gita Sen,

    Affiliation: Indian Institute of Management, Centre for Public Policy, Bangalore, India

  • Ziba Vaghri

    Affiliation: Human Early Learning Partnership (HELP), University of British Columbia, Vancouver, Canada

  • Published: November 01, 2011
  • DOI: 10.1371/journal.pmed.1001115
  • Published in PLOS Medicine

Reader Comments (1)

Post a new comment on this article

The crucial priority for research on equity and health is the development of a sound method of measurement.

Posted by jscanlan on 03 Dec 2011 at 14:56 GMT

The article by Östlin et al.[1] calls for substantial expansion of research into the social determinants of health along with the monitoring of health inequalities on multiple dimensions and the encouraging of governments to inform health policy choices by perceived effects on health inequalities. But in calling for a paradigm shift in health inequalities research the authors fail to recognize that virtually all efforts to appraise the size of health inequalities to date have been suspect because of a failure to consider the ways that, for reasons inherent in normal risk distributions, standard measures of those inequalities tend to be affected by the overall prevalence of an outcome.

As adverse outcomes like morbidity and mortality decrease in overall prevalence, relative differences in experiencing them tend to increase while relative differences in failing to experience them tend to decrease. Similarly, as appropriate healthcare rates increase relative differences in receipt of such care tend to decrease while relative differences in failing to receive such care tend to increase. Absolute differences between rates and differences measured by odds ratio tend also to be affected by the overall prevalence of an outcome, though in a more complicated way. Roughly, as uncommon outcomes becomes more common absolute differences tend to increase; as common outcomes become even more common absolute differences tend to decrease. Differences measured by odds ratios tend to change in opposite directions of absolute difference as the overall prevalence of an outcome changes. Other measures that are in some manner functions of dichotomies (e.g., relative indexes of inequality, Gini coefficients, longevity differences, concentration indexes) tend also to be affected by the overall prevalence of an outcome.[2-6]

The problem is not simply that different measures may commonly yield different conclusions as to whether an inequality has decreased or increased (though that is a quite important matter, particularly if policy decisions are to be based on perceived effects on health inequalities). Rather, it is that none of the standard measures can effectively identify patterns that are other than consequences of differences in the prevalence of an outcome in different settings or at different points in time.

Currently, there is no consensus on how to measure health inequalities even at the most basic level – e.g., determining whether inequalities in immunization have increased or decreased. Reference 7 involves a situation where the authors, relying on relative differences in immunization rates as a measure of inequality, found that a program that dramatically increased overall immunization rates dramatically decreased immunization inequalities. Those relying on relative differences in failure to be immunized would have found dramatic increases in inequalities. Cancer journals commonly discuss relative inequalities in cancer survival and cancer mortality interchangeably without recognizing that improvements in cancer care tend to reduce the former while increasing the latter or that less survivable cancers tend to show larger relative differences in survival but smaller relative differences in mortality than more survivable cancers.[8]

In the United States a study finding that increases in cardio bypass graft surgery (CABG) tended to increase the absolute difference between white and black CABG rates led many to believe that pay-for-performance programs were likely to increase healthcare inequalities. It went unrecognized that the CABG rates were in ranges where, solely for reasons related to the shapes of the risk distributions, general increases in an outcome will tend to increase absolute differences between rates.[9] One state responded to the concern about the impact of pay-for-performance on healthcare inequalities by including inequalities measures as part of its pay-for-performance program. The program relied on a measure that was a function of absolute differences between rates and tended to correlate with higher rates in the same way absolute differences between rates tend to do. Since the rates for the types of care examined in the program were in ranges where higher overall rates would tend to be associated with smaller absolute differences, better general performance tended to correlate with lower perceived inequalities, but for reasons unrelated to a rational measure of within-hospital inequalities. And since better-performing hospitals tended to serve fewer minorities, the end result of rewarding perceived lower inequalities in care may well be to increase inequalities by directing resources away from the institutions that serve large numbers of minorities.[10]

To date there has been only limited recognition that different measures commonly yield different conclusions as to the directions of changes over time, even less recognition that different measures may lead systematically to different conclusions, and almost no recognition that observed patterns are functions of the shapes of the underlying risk distributions.[11]

Lack of understanding of these issues creates particular problems studying the intersection of various potentially health-related factors, an area that the Östlin article suggests be an important focus of future research. In comparatively advantaged populations relative differences in adverse outcomes tend to be large while relative differences in favorable outcomes tend to small.[2-4] Or, from a difference perspective, a factor that increases or decreases adverse outcomes will tend to show larger relative effects on those outcomes in an advantaged population than in a disadvantaged population, while showing a larger relative effect on the opposite, favorable outcome in the disadvantaged population.[12,13] Determining whether a meaningful interaction exists requires other than the traditional methods and traditional assumptions.

Unless health inequalities research is devoted first to ensuring that health inequalities are being effectively measured (either by the method discussed in references 14 and 15 or by some more effective method), more ambitious research programs will offer little guidance on reducing those inequalities. In theory, of course, the authors’ recommendation impliedly calls for resolving all basic methodological issues along the way of implementing that recommendation. But unless the problematic nature of standard measures of health inequalities is specifically recognized in any call for action, health inequalities research is likely to move forward examining ever more complex issues, but with the same tools that are problematic for examining even the simplest issues.


1. Östlin P, Schrecker T, Sadana R, Bonnefoy J, Gilson L, et al. (2011) Priorities for Research on Equity and Health: Towards an Equity-Focused Health Research Agenda. PLoS Med 8(11); e1001115. Doi:10

2. Scanlan JP. Can We Actually Measure Health Disparities? (2006) Chance 19(2):47-51:

3. Scanlan JP. Race and Mortality. (2000) Society 37(2):19-35

4. 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:

5. Scanlan JP. Measuring Health Disparities page of

6. Scanlan’s Rule page of

7. Study illustrates ways in which the direction of a change in disparity turns on the measure chosen. Pediatrics Mar. 27, 2008 (responding to Morita JY, Ramirez E, Trick WE. Effect of school-entry vaccination requirements on racial and ethnic disparities in Hepatitis B immunization coverage among public high school students. Pediatrics 2008;121:e547-e552): http://pediatrics.aappubl...

8. Morality and Survival page of
9. Pay for Performance sub-page of Measuring Health Disparities page of

10. Between Group Variance sub-page of Measuring Health Disparities page of

11. Section E.7 of Measuring Health Disparities page of

12. Scanlan JP. The perils of provocative statistics. The Public Interest 1991;102:3 14:

13. Subgroup Effects sub-page of Scanlan’s Rule page of

14. Scanlan JP. Measuring Health Inequalities by an Approach Unaffected by the Overall Prevalence of the Outcomes at Issue, presented at the Royal Statistical Society Conference 2009, Edinburgh, Scotland, Sept. 7-11, 2009:

15. Solutions sub-page of Measuring Health Disparities page of

No competing interests declared.