Morita et al.[1] find that, in addition to increasing overall
hepatitis B vaccination rates, a school-entry vaccination requirement led
to a dramatic decrease in racial and ethnic disparities in vaccination
rates.
Among 5th graders, before the requirement, the black rate was 63%
lower than the white rate (3% versus 8%); immediately following
implementation of the requirement, the black rate was only 28% lower than
the black rate (33% versus 46%). Among 9th graders, before the
requirement, the black rate was 30% lower than the white rate (32% versus
46%); following implementation, the black rate was only 6% lower than the
white rate (84% versus 89%).
Among 5th graders, before the requirement, the Hispanic rate was 50%
lower than the white rate (4% versus 8%); following implementation, the
Hispanic rate was only 9% lower than the white rate (42% versus 46%).
Among 9th graders, before the requirement the Hispanic rate was 13% lower
than the white rate (40% versus 46%); following implementation the
Hispanic rate was only 3% lower than the white rate (86% versus 89%).
(Because, due to rounding, a 3.37% relative difference between 86% and 89%
may look like the absolute difference between rates, which below I will
term a “percentage point difference,” for clarity I note that the 3%
figure is indeed the relative difference. The same holds for several
similar situations below.)
But not everyone would agree with the authors’ conclusions about
dramatic decreases in disparities. In particular, the National Center for
Health Statistics (NCHS) would consider the program not to have caused
dramatic decreases in disparities. Rather, NCHS would regard the
disparities to have increased and in some cases dramatically so. For NCHS
would view the matter in the following terms:
Among 5th graders, before the requirement, the black rate of failing
to receive vaccination was only 5% higher than the white rate (97% versus
92%); after the program, the black rate was 24% higher than the white rate
(67% versus 54%). Among 9th graders, before the requirement, the black
rate of failing to receive vaccination was only 26% higher than the black
rate (68% versus 54%); after implementation, the black rate was 45% higher
than the white rate (16% versus 11%).
NCHS would also have regarded the program as leading to increases in
disparities between Hispanics and whites. Among 5th graders, before the
requirement, the Hispanic rate of failing to receive vaccination was only
4% higher than the white rate (96% versus 92%); after implementation, the
Hispanic rate was 7% higher than the white rate (58% versus 54%). Among
9th graders, before the requirement, the Hispanic rate of failing to
receive vaccination was only 11% higher than the white rate (60% versus
54%); after implementation, the Hispanic rate was 27% higher than the
white rate (14% versus 11%).
The figures underlying the above statements and those that follow may
be found in Tables A and B to this comment, which may be accessed by the
following link:
http://www.jpscanlan.com/images/Tables_A_and_B_to_Morita_Comment.pdf
At this point I add a bit of background. In 2000 I published an
article called “Race and Mortality” in the social science magazine
Society.[2] Race and Mortality described the statistical tendency whereby
the rarer an outcome the greater tends to be the relative difference in
experiencing it and the smaller tends to be the relative difference
between rates of failing to experience it – a tendency I had since 1987
described in about a dozen articles on the interpretation of group
differences in the law and the social and medical sciences. See Section A
of this page: http://www.jpscanlan.com/homepage/measuringhlthdisp.html
Race and Mortality principally addressed the fact that during periods
of declining mortality, increasing relative differences in mortality were
being universally regarded as reflecting increases in health disparities
without regard to the extent to which, solely for statistical reasons,
increases in relative differences between mortality rates would typically
accompany declining mortality and without regard to whether relative
differences in survival rates were decreasing. But the article also
pointed out that, solely as a matter of convention, disparities in things
like beneficial healthcare procedures were typically measured in terms of
relative differences in rates of receiving such procedures. Thus, it
noted, since such procedures were becoming more widespread, racial
disparities in those outcomes were perceived to be declining. A more
succinct expression of these tendencies than found Race and Mortality, and
one that addresses as well the implications of changes in overall
prevalence with regard to absolute differences and odds ratios may be
found in reference 3.)
Eventually, in a series of papers,[4-6] NCHS responded to Race and
Mortality’s discussion of the way relative differences in favorable
outcome and relative differences in adverse outcomes tend to give
different impressions of the comparative size of disparities. It did so,
however, not by addressing the implications of the fact that changes in
overall prevalence tend to cause relative differences in adverse outcomes
and relative differences in favorable outcomes to change systematically in
opposite directions. Rather, it simply recommended that, for consistency,
all health and healthcare disparities, including disparities in things
like rates of mammography and immunization, should be measured in terms of
relative differences in adverse outcomes (i.e., the failure to receive
such procedures). As reflected by the Morita and other recent studies,
that approach has not yet gained much following among researchers outside
the government.
Such approach, however, underlies the measurement of progress towards
the health and healthcare disparities reduction goals in Health People
2010.[7] Measurement of disparities by the Agency for Healthcare Research
and Quality (AHRQ), which issues the yearly National Healthcare
Disparities Report, may be slightly different. As of the 2006 report,[8]
AHRQ was measuring disparities in terms of the larger of the relative
difference in the favorable or the adverse outcome. Since the latter
relative difference is almost always larger than the former for the things
that AHRQ examines, its approach is usually consistent with that of NCHS.
But that is not always the case, and, as shown in Tables A and B, at least
for certain periods, it would not be so with respect to some issues
addressed in the Morita study. For example, despite the dramatic decline
in the relative difference between vaccination rates of black and white
5th graders immediately after the requirement was implemented, the
relative difference in vaccination rates remained larger than the relative
difference in rates of failing to be vaccinated. Thus, AHRQ might agree
with the authors of the instant study as to certain periods and agree with
NCHS for other periods. And, with respect to situations where one
relative difference is larger in one period and the other in another
period, AHRQ would probably have some difficulty deciding what to do (as
discussed with regard to the Morita study and some other matters in the
Addendum to reference 9).
There remains the issue of whether the vaccination requirement
affected the comparative situation of blacks and white with respect to
vaccination in some meaningful sense – that is, in ways other than those
that would tend naturally to occur due to the overall increases in
vaccination rates.
Some researchers rely on absolute differences between rates to
measure healthcare disparities. Among 5th graders, the absolute
difference between vaccination rates increased immediately after the
implementation of the requirement (from 5 percentage points to 13
percentage points), while among 9th graders the absolute difference
declined (from 14 percentage points to 5). Thus, some researchers would
consider the disparity to have increased among 5th graders but have
declined among 9th graders. For reasons explained in several places,
however, these changes are simply what would be expected usually to occur
given the range of initial and final rates at issue.[3,9-11] Those and
other works address implications of various patterns of changes in
absolute differences as the overall prevalence of an outcome changes –
roughly, when rare outcomes increase, absolute differences between rates
tend to increase; when common outcomes increase, absolute differences tend
to decline – at sufficient length that there is little value in belaboring
such matter here. (The same holds for differences measured in odds
ratios, which tend to change in the opposite direction of absolute
differences.)
I will note, however, that many researcher who favor absolute
differences as measures of disparities do so both because the absolute
difference is unaffected by whether one examines the favorable or the
adverse outcome and because absolute differences better reflect the burden
on the disadvantaged group of its disadvantaged position relative to the
outcome at issue. NCHS, while principally relying on relative differences
in adverse outcomes to measure disparities, also emphasizes the importance
of absolute differences between rates as indicators of the excess burden
on the disadvantaged group. And in this regard the absolute difference
seems to have some appeal as a measure of disparity even when it changes
solely because of a change in overall prevalence. But we see in Tables A
and B that after implementation of the requirement, among fifth graders,
the absolute difference increased for both blacks and Hispanic,
substantially in the former case. And in fact one will tend to observe
such pattern generally when relatively uncommon outcomes become more
widespread (as discussed in references 9-11). In such circumstances, the
absolute difference as a meaningful measure of disparity loses much of its
appeal.
In references 11 and 12, and a few other places, I have described an
approach to measuring disparities between rates that ought to be
unaffected by changes in overall prevalence. Specifically, based on the
rates at each point in time, the approach derives a difference between
means of hypothetical underlying normal distributions of factors
associated with an outcome. The approach is somewhat speculative given
that we cannot observe the actual nature of the underlying distributions.
But to my mind the approach is superior to anything else so far developed
and is certainly superior to the simple reliance on the standard measures
of differences between rates without regard to the way the measures tend
usually to change solely because of a change in the overall prevalence of
an outcome. In any case, the results of that approach as applied to the
Morita data are included in Tables A and B. And they seem to indicate
that for blacks, among 5th graders the disparity declined immediately
after implementation of the requirement (a decline that occurred, to
return to the point in the preceding paragraph, while absolute differences
between rates increased substantially), and continued to decline in the
following year; among 9th graders, the disparity initially declined, but
increased slightly in the following year.
For Hispanics, among 5th graders, the disparity declined immediately
following the implementation of the requirement and continued to decline
in the following year. Among 9th graders, however, there was no change in
the Hispanic-white disparity immediately following implementation of the
requirement, but a substantial decrease in the following year.
One thing that is clear is that this was a remarkably successful
program. But as with other remarkably successful programs,[13] appraising
the impact of the program on racial and ethnic disparities is much more
complicated than it may seem at first sight.
References:
1. 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.aappublications.org/cgi/reprint/121/3/e547?maxtoshow=&HITS=10&hits=10&RESULTFORMAT=&fulltext=morita&andorexactfulltext=and&searchid=1&FIRSTINDEX=0&sortspec=relevance&resourcetype=HWCIT
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. Scanlan JP. Can we actually measure health disparities? Chance
2006:19(2):47-51:
http://www.jpscanlan.com/images/Can_We_Actually_Measure_Health_Disparities.pdf
4. Keppel KG, Pearcy JN, Klein RJ. Measuring progress in Healthy
People 2010. Healthy People statistical notes. No. 25. Hyattsville, Md.:
National Center for Health Statistics:
http://www.cdc.gov/nchs/data/statnt/statnt25.pdf
5. Keppel KG, Pamuk E, Lynch J, et al. Methodological issues in
measuring health disparities. Vital and health statistics. Series 2. No.
141. Washington, D.C.: Government Printing Office, 2005.(DHHS publication
no. (PHS) 2005-1341.):
http://www.cdc.gov/nchs/data/series/sr_02/sr02_141.pdf.
6. Keppel, KG, Pearcy JN. Measuring health disparities in terms of
adverse outcomes. J Public Health Management Practice. 2005:11(6): 479-83
7. Keppel KG, Bilheimer L, Gurley L. Improving population health
and reducing health disparities. Health Affairs 2007;26(5):1281-1292.
8. Agency for Healthcare Research and Quality, 2006 National
Healthcare Disparities Report: http://www.ahrq.gov/QUAL/nhqr06/nhqr06.htm
9. Scanlan JP. Measurement Problems in the National Healthcare
Disparities Report, presented at American Public Health Association 135th
Annual Meeting & Exposition, Washington, DC, Nov. 3-7, 2007.
PowerPoint Presentation:
http://www.jpscanlan.com/images/APHA_2007_Presentation.ppt
Oral Presentation: http://www.jpscanlan.com/images/ORAL_ANNOTATED.pdf
Addendum (March 11, 2008): http://www.jpscanlan.com/images/Addendum.pdf
10. Scanlan JP. Effects of choice measure on determination of
whether health care disparities are increasing or decreasing. Journal
Review May 1, 2007, responding to Vaccarino V, Rathore SS, Wenger NK, et
al. Sex and racial differences in the management of acute myocardial
infarction, 1994 through 2002. N Engl J Med 2005;353:671-682 (and
several other articles in the same issue):
http://www.journalreview.org/view_pubmed_article.php?pmid=16107620&webenv=00P_2r_lHBKZPkExnEkCR_j5
-u8waNcJ-
87aLnoSJWxvN_ljFKstOR3CAx%402B600907661FF950_0034SID&qkey=1&rescnt=2&retstart=0&q=%22vaccarino+v%22+%22rathore+ss%22
11. Scanlan JP. Can We Actually Measure Health Disparities,
presented at the 7th International Conference on Health Policy Statistics,
Philadelphia, PA, Jan. 17-18, 2008 (invited session).
PowerPoint Presentation:
http://www.jpscanlan.com/images/2008_ICHPS.ppt
Oral Presentation: http://www.jpscanlan.com/images/2008_ICHPS_Oral.pdf
12. Scanlan JP. Comparing the size of inequalities in dichotomous
measures in light of the standard correlations between such measures and
the prevalence of an outcome. Journal Review Jan. 14, 2008, responding to
Boström G, Rosén M. Measuring social inequalities in health – politics or
science? Scan J Public Health 2003;31:211-215:
http://www.journalreview.org/view_pubmed_article.php?pmid=12850975&specialty_id
13. Scanlan JP. Changing social inequalities in SIDS. Am J Public
Health Dec. 11, 2005, responding to Pickett et al. Widening social
inequalities in risk for sudden infant death syndrome. Am J Public Health
2005;95:97-81: http://www.ajph.org/cgi/eletters/95/11/1976
Conflict of Interest:
None declared
eLetters: ![[Read eLetters]](/icons/misc/sectionDOWN.gif){kind=icon}
