AJPH -- eLetters for Icks et al., 0 (2006) 200506333

QUICK SEARCH:

	Author:		Keyword(s):
Year:		Vol:		Page:

Electronic Letters to:

Research and Practice:
Andrea Icks, Burkhard Haastert, Wolfgang Rathmann, Joachim Rosenbauer, and Guido Giani
Trends in Hospitalization and Sociodemographic Factors in Diabetic and Nondiabetic Populations in Germany: National Health Survey, 1990-1992 and 1998
Am J Public Health 2006; 0: AJPH.2005.063339v1 [Abstract] [PDF]

eLetters: Submit a response to this article

Electronic letters published:

Differences in average hospital stay as a measure of inequality: James Scanlan (18 August 2006)

Differences in average hospital stay as a measure of inequality 18 August 2006

James Scanlan,
Attorney
James P. Scanlan, Attorney at Law
Send letter to journal:
Re: Differences in average hospital stay as a measure of inequality

jps{at}jpscanlan.com James Scanlan

While noting that the failure to confirm a widening of the social gradient in hospitalization days per person-year in the diabetic population “may result in part from low statistical power and an overestimation of dispersion,” Icks et al. find that such a “result is surprising, because a widening of social inequality in health is often observed.”1
But Table 2 of their study shows that the ratio of the mean hospital days of the low education group to that of the high education group changed from 1.17 (3.79 over 3.24) in 1990-1992 to 1.33 (3.40 over 2.55) in 1998 or, put another way, the study found a 10 percent reduction for the low education group compared with a 21 percent reduction for the high education group. I assume that one would observe a similar change in direction of the effect size between the two averages, which may be a better approach to measurement. It is true that the change in the difference between the two groups’ average stays was not significant. But when one observes a change in the size of a difference that moves in an expected direction, there is little reason to regard the mere fact that the size of the change in the difference was not of sufficient magnitude to be statistically significant as surprising. That is especially so when the study is deemed to have low power. In such circumstances, it would seem more reasonable to believe that the observed change in difference, being in the expected direction, was more likely to reflect a real change than to have occurred by chance.
The expectation of a widening of the social gradient raises an issue, however. The authors cite two studies for the expectation that socioeconomic inequalities in health are likely to widen.2,3. But each of these authorities merely observes that certain measures of health inequality have been widening. They do not explain why either the measures they employ or other measures should be widening.
There is reason to expect certain measures of socioeconomic inequalities in health to widen. In particular, since mortality is declining, we should expect to observe increasing relative socioeconomic differences in mortality rates. This is so because generally a reduction in adverse outcomes tends to increasingly concentrate them in the most susceptible segments of the overall population, and disadvantaged groups comprise larger proportions of each increasingly more susceptible segment of the overall population. A corollary to this increased concentration is an increase in the relative differences in experiencing the adverse outcome. On the other hand, as mortality declines, we should expect to observe decreasing differences in survival rates, a function of the fact that as an outcome declines, disadvantaged groups will tend to comprise a higher proportion of the population that is now enabled to avoid the outcome than it did of the population previously avoiding the outcome.4-8
Yet the tendency whereby declines in the prevalence of an outcome tend to increase relative differences in experiencing the outcome is only clearly evident as to dichotomous variables. For example, since the average length of hospital stays among diabetics is generally decreasing, we should expect to see an increase in the relative difference between the rates at which higher and lower socioeconomic groups stay beyond some given number of days (though a decrease in the relative difference between rates at which higher and lower socioeconomic stay fewer than that given number of days). But whether an overall decline in a continuous variable like length of hospital stay should similarly lead to an increasing disparity between the average stays of two groups is not as clear. The interpretation of whether changes in measures of difference as to dichotomous variables reflect meaningful changes in the relative health of two groups is problematic due to the tendency noted above and to the ways other measures of size of differences as to such variables change solely as a result of change in prevalence.4,5,8 If measures of differences between averages of continuous variables do not suffer from the same interpretative problem as measures of difference as to dichotomous variables, the former measures may offer a means of identifying meaningful changes in the size of health inequalities over time,8 something that health inequalities research so far seems to be lacking. Hence, the effect of prevalence changes on effects sizes between average of continuous variables is a subject warranting study.
References
1. Icks A, Haastert B, Rathmann W, et al. Trends in hospitalization and sociodemographic factors in diabetic and nondiabetic populations in Germany: National Health Survey, 1990-1992 and 1998. Am J Public Health. 2006;96:xxx-xxx.doi.10.2105/ALPH.2005.063339.
2. Marmot M, Bobak M. International comparators and poverty and health in Europe. BMJ. 2000;321:1125-1128.
3. Mackenbach JP, Bakker MJ, for the European Network on Interventions and Policy to Reduce Inequalities in Health. Tackling socioeconomic inequalities in health: analysis of European experiences. Lancet. 2003;362:1409-1414.
4. Scanlan JP. Can we actually measure health disparities? Chance. 2006;19(2):47-51. In press.
5. Scanlan JP. Measuring health disparities. J Public Health Manag Pract 2006;12(3):294 [Lttr]: http://www.nursingcenter.com/library/JournalArticle.asp?Article_ID=641470.
6. Scanlan JP. Race and mortality. Society. 2000;37(2):19-35: http://www.jpscanlan.com/images/Race_and_Mortality.pdf.
7. Scanlan JP. Divining difference. Chance. 1994;7:38-39,48.
8. Scanlan JP. Measuring health inequalities. Paper presented at: 5th International Conference on Health Economics, Management and Policy, June 5-7, 2006, Athens, Greece: http://www.jpscanlan.com/images/Measuring_Health_Inequalities.pdf.