Jump to comment:

• Interpreting changes in mortality rate differences
  James P. Scanlan
  Published on: 27 April 2016

• Published on: 27 April 2016

  Interpreting changes in mortality rate differences

  • James P. Scanlan, Attorney

  Dear Editor,

  Shaw et al. find that during a period of substantial decline in child mortality, relative differences among socioeconomic classes increased, while absolute differences remained stable (actually declining very slightly). Such findings are entirely to be expected.

  Socioeconomic mortality differences, measured in relative terms, tend almost invariably to increase during times of declining mortality...

  Show More

  Dear Editor,

  Shaw et al. find that during a period of substantial decline in child mortality, relative differences among socioeconomic classes increased, while absolute differences remained stable (actually declining very slightly). Such findings are entirely to be expected.

  Socioeconomic mortality differences, measured in relative terms, tend almost invariably to increase during times of declining mortality. The tendency is a consequence of the fact that disadvantaged groups comprise larger proportions of the part of the population that is very susceptible to some adverse outcome than they comprise of the part of the population that is only somewhat susceptible to the outcome. Thus, as effective measures to reduce mortality increasingly limit avoidable mortality to only the most susceptible segments of the population, disadvantaged groups comprise a larger proportion of those continuing to experience avoidable mortality. Correspondingly, the ratio of the mortality rate of a disadvantaged group to that of an advantaged group will increase.

  This pattern can be observed in almost every data set that allows one to examine the demographic makeup of the population falling below various points in a continuum. Published U.S. income data provide a ready illustration. In 2004, the black poverty rate (24.7 percent) was 2.3 times the white poverty rate (10.8 percent); but the black rate of falling below 50 percent of the poverty line (11.7 percent) was 2.7 times the white rate (4.4 percent).[1] Suppose, then, that a program to reduce poverty enabled everyone with incomes between the poverty line and 50 percent of the poverty line to escape poverty. The ratio of the black poverty rate to the white poverty rate would increase from 2.3 to 2.7.

  Anyone inclined to regard the increase in the black-white poverty ratio as reflecting a true worsening of the status of U.S. blacks compared with whites should examine the opposite outcome -- the avoidance of poverty. For that same across-the-board decline in poverty would reduce the difference between black and whites rates of avoiding poverty -- changing the black-white ratio of avoiding poverty from .84 (75.3/89.2) to .92 (88.3/95.6). But the pattern is not unique to income distributions. Rather, with virtually every set of dichotomous outcomes disparately affecting two groups, the decline in the prevalence of one outcome tends both to increase the relative difference in experiencing the outcome and to reduce the relative difference in avoiding the outcome. And, if we accept changes in relative differences as reflecting meaningful changes in quality, whether we regard inequality to be increasing or decreasing will turn on whether we examine the unfavourable or the favourable outcome.[2,3]

  On the other hand, absolute differences are the same regardless of whether one examines the favourable or the unfavourable outcome. Nevertheless, absolute differences fail to provide a satisfactory means of evaluating changes in the inequality of mortality rates over time because, like relative differences, absolute differences also change when there occurs an across-the-board change in the prevalence of an outcome. When the rates at which two groups experience the unfavourable outcome are close to 100 percent, the absolute difference will be close to zero, as it also will be when the rates of experiencing the favourable outcome are close to 100 percent. But as the prevalence of an outcome declines from 100 percent, the absolute difference will rise for a time and then decrease on its way back towards zero. At the points on the distribution where we usually examine mortality, the absolute difference will tend to decline as the overall prevalence of the unfavourable outcome declines. This is what we observe in the case of the hypothetical across-the-board reduction in poverty described above, where the absolute difference between black and white poverty rates would fall from 15.5 to 3.5.

  Thus both the increasing mortality disparity measured in relative terms and declining disparity measured in absolute terms are to be expected in times of declining mortality. The patterns may not occur in every case, particularly when rates approach an irreducible minimum.[3] But the tendency is a strong one, and without recognizing it, one cannot draw meaningful conclusions about whether inequalities are increasing or decreasing over time.

  References

  [1] U.S. Census Bureau, Current Population Survey, 2004 Annual Social and Economic Supplement, Table POV01: "Age and Sex of All People, Family Members and Unrelated Individuals Iterated by Income-to-Poverty Ratio Below 100% of Poverty -- White Alone or in Combination (A.O.I.C.)," May 19, 2005. http://pubdb3.census.gov/macro/032005/pov/new01_000.htm

  [2] Scanlan JP. Divining difference. Chance 1994;4:38-9,48.

  [3] Scanlan JP. Race and mortality, Society 2000;2:29-35.

  Show Less

  Conflict of Interest:
  None declared.

  • Back to top