James P. Scanlan, Attorney at Law

(Apr. 25, 2010; rev. Feb. 3, 2011)

When observers conclude that the size of a difference between outcome rates is larger in one setting than another, or that a rate appears to be changing more for one group than another, they sometimes offer an explanation for such patterns.  And some who do that will devise names for the explanatory theories.  For example, when it is observed that during periods of improvements in health, advantaged groups experience larger proportionate decreases in adverse outcomes than disadvantaged groups, such pattern is sometimes regarded as a substantiation of the DIFFUSION OF INNOVATION THEORY or the INVERSE EQUITY HYPOTHESIS.  The former reflects the notion that in the course of diffusion throughout the population, innovation goes first to the more advantaged; the latter is essentially the same but with emphasis on the notion that innovations go first to the groups who need them least.  Whatever the plausibility of these theories, however, those citing them in health disparities research are prompted by a perception of the size of differences that are flawed for failure to consider the effects of overall prevalence.  Thus, they fail to consider that as adverse outcomes decrease (i.e., favorable outcomes increase) disadvantaged groups tend to experience a larger proportionate increase in the favorable outcome.  Hence, with regard to the favorable outcome, it is the disadvantaged group that would seem to be first benefiting from the innovation.  See discussion in Subgroup Effects and Illogical Premises sub-pages of the Scanlan’s Rule page of why it is illogical to regard it as somehow normal that two groups with different base rates for some outcome would experience the same proportionate change in the outcome (or in the opposite outcome), given that it is not possible to do both.

See Houweling et al.,[1] which in responding to Race and Mortality,[2] finds patterns of correlations between the size of relative differences in outcome rates and the prevalence of an outcome that are the essentially the same as those described in Race and Mortality.  But Houweling et al. ignore Race and Mortality’s explanations for the patterns and instead cite the diffusion of innovation theory.  In doing so, they overlook that, with respect to the favorable outcome, the disadvantaged would appear to be the first to benefit from the innovation.  See Comment on Eikemo for discussion of Houweling et al.  See also Comment on Victora for discussion of the reliance by Victora et al.[3]  on the inverse equity hypothesis to explain observed patterns of changes in relative differences.

The above pertain to perceptions about relative differences.  Similar issues exist with regard to perceptions about absolute differences.  While no one has yet attributed a name to the perceived pattern, there exists a perception that it is easier to address disparities in health processes than in clinical outcomes.  The perceptions arises from, perhaps among other sources, the findings of Trivedi et al.[4] that during periods of overall improvements in processes and clinical outcomes, absolute differences between process rates consistently decreased while absolute differences between clinical outcome rates increased more often than they decreased.  This theory is probably correct.   But the perceptions of Trivedi et al. concerning the changes in the sizes of the two types of disparities were flawed for failure to recognize that the process outcomes were usually in ranges where further increases tend to reduce absolute differences between rates while the clinical outcome rates were in ranges where further increases tend to increase absolute differences.  See Comment on Trivedi JAMA 2006.  See also the Pay for Performance sub-page of the Measuring Health Disparities page regarding the way reliance on absolute differences between rates as a measure of healthcare disparities has led to opposite perceptions in the United States and the United Kingdom regarding the effects of pay-for-performance on healthcare disparities.

References:

1. 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

2. 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

3. Victora CG, Vaughan JP, Barros FC, et al.  Explaining trends in inequities:  evidence from Brazilian child health studies. Lancet 2000;356:1093-1098.