Intended for healthcare professionals


Importance of monitoring health inequalities

BMJ 2013; 347 doi: (Published 05 November 2013) Cite this as: BMJ 2013;347:f6576

The monitoring of health inequalities has never been sound II

In a 9 November 2013 comment [1] on an editorial in which Marmot and Goldblatt [2] stressed the importance of monitoring health inequalities, I explained that the monitoring of health inequalities in the UK and elsewhere had never been sound as a result of the failure of researchers to recognize patterns, inherent in underlying risk distributions, whereby standard measures of differences between outcome rates tend to be systematically affected by the prevalence of an outcome.

Since then, the European Public Health Association (EPHA) issued a call for papers for the 7th European Public Health Conference, titled “Mind the gap: Reducing inequalities in health and healthcare,” which will be held in Glasgow 19-22 November 2014. While abstracts are not yet posted in the online programme,[3] presentation titles do not suggest that presenters will be addressing the measurement issues described in the earlier comment. Very likely, few presenters will know such issues exist.

The greatest potential for the conference to influence health and healthcare inequalities research in a positive or negative way rests in the opening address, titled “Minding the gap: inequalities in health in Europe,” to be delivered by Professor Johan Mackenbach of Erasmus University in The Netherlands. Professor Mackenbach is a co-author of the 2007 International Journal for Health Equity article by Houweling et al.[4] mentioned in my earlier comment as recognizing the same patterns of correlations between the prevalence of an outcome and relative and absolute differences between outcomes rates described in the comment.

Specifically, responding to reference 5, the Houweling article recognized the pattern described in reference 5 whereby when an outcome becomes less common, relative differences in experiencing it tend to increase while relative differences in avoiding it tend to decrease. That is, for example, as mortality declines relative differences in mortality tend to increase while relative differences in survival tend to decrease; as rates of appropriate healthcare increase relative differences in receipt of such care tend to decrease while relative differences in failure to receive such care tend to increase. Further, the Houweling article independently recognized the same reverse u-shaped pattern by which absolute differences between rates tend to change as an outcome changes in prevalence that I had shown in Figure 4 of reference 6 (at 50) and Figure 4 of reference 7 (at 30). Most important, the Houweling article recognized that one can only usefully employ relative or absolute differences between outcome rates to appraise health inequalities while taking into consideration the prevalence of the outcome.

As suggested in the earlier comment, however, the Houweling authors have failed to reflect that recognition in their subsequent work. I allude further to that failure of the Houweling authors at 343 (page 16 online) of reference 8, a recent updating or references 5 and 6.

When publishing reference 8, however, I was not aware that in a 2012 article in Social Science and Medicine,[9] Professor Mackenbach had attempted to address both references 5 and 6. In doing so, Professor Mackenbach characterized the articles as maintaining that patterns of increasing relative differences in adverse outcomes as those outcomes decline are mathematical artifacts. I am not sure that the points I have made about the need to understand the ways measures tend to be driven by the prevalence of an outcome can be accurately characterized as maintaining that observed patterns are artifactual. More important, however, rather than reading the Houweling article as reaching essentially the same conclusion as references 5 and 6, in the 2012 article Professor Mackenbach reads the Houweling article as demonstrating that the patterns are not inevitable (something already made quite clear in reference 5). Still more important, the Mackenbach article then, by its efforts to draw conclusions and inferences about health inequalities without consideration of those patterns, impliedly maintains that because the patterns will not always be observed, they may be ignored in interpreting data on demographic differences in outcome rates.

The implications of this position may be best appraised with respect to the discussion in reference 8 at 329 (page 2 online) regarding its Table 2 (which shows how decreases in US poverty will tend to increase relative differences between black and white poverty rates while reducing relative differences between black and white rates of avoiding poverty). The discussion poses the question of whether there could be any value in exploring the reasons for changes in either relative difference without consideration of the patterns reflected in the table. While the discussion suggests that the answer ought to be obvious, it returns to that question at 343 (page 16 online). But the approach implied in the 2012 Mackenbach article would support devoting resources to exploring the reasons for the patterns of changes in whichever relative difference the research favored without consideration of the patterns illustrated in Table 2, and to do so even when one observed exactly the patterns in Table 2.

The implications of Professor Mackenbach’s position may also be usefully appraised with regard to the section of reference 8 titled “Illogical Expectations and Unfounded Inferences” at 339-41 (pages 12-14 online), which discusses the way examination of relative differences in favorable outcomes commonly yield opposite interpretations of observed patterns from examination of relative differences in the corresponding adverse outcomes. More generally, I suggest, reference 8 well illustrates the utter futility of, and vast waste of resources involved with, analysing health and healthcare inequalities by means of particular researchers’ preferred measures without consideration of the way each measure tends to be affected by the prevalence of an outcome.

Professor Mackenbach’s opening address at the EPHA conference has great potential either to advance health inequalities research by bringing to the attention of researchers the importance of issues regarding the ways standards measures tend to be affected by the prevalence of an outcome or to undermine that research by leading researchers to believe that no such issues exist or that one can conduct sound research while ignoring them.


1. The monitoring of health inequalities has never been sound. BMJ Nov. 9, 2013. (Accessed Aug. 17, 2014)

2. Marmot M, Goldblatt P. Importance of monitoring health inequalities. BMJ 2013;347:f6576).

3. Dynamic Programme for the 7th European Public Health Conference. (Accessed Aug. 17, 2014)

4. 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. (Accessed 17 Aug 2014)

5. Scanlan JP. Race and mortality. Society 2000;37(2):19-35. (Accessed 17 Aug 2014)

6. Scanlan JP. Can we actually measure health disparities? Chance 2006;19(2):47-51: (Accessed 17 Aug 2014)

7. The Misinterpretation of Health Inequalities in the United Kingdom. British Society for Populations Studies Conference 2006, Southampton, England, 18-20 Sept 2006. (Accessed 17 Aug 2014)

8. Scanlan JP. Race and mortality revisited. Society 2014;51:328-49. (Accessed 17 Aug 2014)

9. Mackenbach JP. The persistence of health inequalities in modern welfare states: The explanation of a paradox. Social Science and Medicine 2012;75:761-769.

Competing interests: No competing interests

17 August 2014
James P Scanlan
James P. Scanlan, Attorney at Law
Washington, DC USA