Inequalities in premature mortality in Britain: observational study from 1921 to 2007
BMJ 2010; 341 doi: https://doi.org/10.1136/bmj.c3639 (Published 22 July 2010) Cite this as: BMJ 2010;341:c3639
All rapid responses
This article is less informative than it should be. It would be
interesting to ask elderly people, who lived before the war, to ask how
believable they find these results. What counts to people is death, not
relative death.
The appropriate tables would be absolute age standardised rates, with
SMR and absolute rate differences. The easiest way to lower SMR, is to
increase absolute mortality. If we assume a distribution of mortality in
the population and shift the whole distribution to the left (towards lower
mortality), rate ratios increase. To maintain constant ratios, mortality
differences have to get always smaller and the mortality distribution
always narrower. If we assume that the mortality risks in populations are
diverse, there is a limit to this narrowing. Assuming constant ratios is
actually a strong assumption when mortality rapidly declines.
I applied the SMR to the mortality of England and Wales (55-59 year
old, source www.mortality.org) and calculated the absolute differences
between highest and lowest decile using the SMR of the authors. The
picture is very different now: absolute differences between the worst off
and the best off were high before the war. They declined in the fifties
and sixties, remained stable in the 1970s, 1980s and 1990s and declined
again in 2000-2006. The mortality difference between highest and lowest
decile became less than 5 per 1000 personyears, a record low. The trends
were remarkably different. In the worst off, mortality declined steadily,
speeding up in more recent periods. In the best off, this rate of change
was much more "bumpy". Indeed, the 'best' SMR coincided with conspicuous
lack of progress, in the period 1969-1973. Tobacco was the great social
equalizer, killing then rich and poor alike: hardly an ideal to strive
for. In the 1990s, the speed of mortality decline was very high in the
best off: it is tempting to blame better access of the well off to the
highly effective statins (and comparable interventions). Mortality changes
are nearly exclusively dominated by cardiovascular mortality.
Life expectancies are a highly intuitive aggregate measure of age
specific mortality rates - they show what counts to people. The authors
show these for the extremes only, comparable to the times of the first and
last rider in a stage of the Tour de France. The faster the stage, the
bigger the difference. The trends in both rich and poor suggest rapid
decreases of mortality. The only thing we should learn, is that these
dynamic changes are not homogeneously distributed over all districts, but
that some move quickly and others move more slowly.
The good news is better than the bad: mortality declined rapidly in
all social classes, more in the poor but more rapidly in the rich. This is
likely not a problem, maybe even a cause of celebration. It seems the
natural consequence of great, but unavoidably heterogeneous progress in
diverse populations.
Mortality differences by social class, income and educational attainment
are as ubiquitous as death itself. Relative size of these differences is
uninformative. These differences are unfair, where people die an at
reasonable costs avoidable death because they are poor, lowly schooled or
living at the bottom. We then should keep in mind that the evidence of
effectiveness of targeted health policies is tenuous, compared to the
large numbers of deaths avoided by general progress.
Competing interests:
None declared
Competing interests: No competing interests
In the interests of equality of treatment, I would like to speak up
on behalf of the many BMJ readers between the 8th and 9th income decile,
ie the second richest category. For the nine time intervals from 1990 to
2007 (Table 1), the Group 9 SMR has been consistently 4 or 5 points worse
than the richest group. Furthermore, there has been no change in this gap
since 1921 (Table 2). There is a serious point here, in that if this gap
cannot be explained, then the situation at the bottom end where many
adverse social factors impinge and interact will never be understood.
The above article assumes that the poor health at the bottom end
reflects the outcome of socially patterned exposures, child poverty,
economic depressions, organisation of misery, welfare and health policies,
social inequalities, etc. Let us then propose there is some external,
environmental, ecological, employment or economic variable (Factor E)
responsible for causing differential life expectancy. Factor E would then
have to have the following properties to explain all the data:
1. The relation between SMR and income is roughly linear. Hence
Factor E must be equally effective at both ends of the scale, since the
mortality gap between poverty deciles 1 and 2 is the same as between 8 and
9.
2. Factor E was acting in exactly the same way in 1921 as in 2007.
3. From 1990 to 2007 Factor E has to have consistently increased the
SMR in the three poorest deciles whilst decreasing the SMR in the two
richest deciles (Table 1).
4. Factor E needs to be operative at the earliest age at which social
class health differences can be shown, ie infancy.
This combination of properties is so unlikely that Factor E must be a
product of wishful thinking and cannot exist.
Given the unliklihood af a contextual factor, what about a
compositional factor? One does not have to look far for a plausible
candidate, IQ. All cause mortality was 2.74 times higher for the lowest
income group than for the highest, but after adjustment for a 10 minute
paper and pencil IQ test, this fell from 2.74 to 1.89. For two other
socioeconomic indicators excess mortality was completely explained by IQ;
after IQ adjustment, the relative index of inequality for all cause
mortality for current social class fell from 1.70 to 0.96, for years of
education from 1.65 to 0.80. The index for childhood social class was only
reduced from 2.32 to 1,68. so any adverse factors related to social class
must have been operative in early childhood, not adulthood. If there is
then a linear relation between IQ and wealth, then all the recent findings
are simply explained, and there is no need for Factor E and its contorted
logic (or for Group 9 doctors to ask for pay rises!). IQ then a marker or
measure of general biological fitness.
Batty GD, Der G, Macintyre S, Deary IJ. Does IQ explain socioeconomic
inequalities in health? Evidence from a populaion based cohort study in
the west of Scotland. BMJ 2006;332:580-3.
Competing interests:
None declared
Competing interests: No competing interests
The study of patterns of mortality inequalities by Thomas et al.[1],
which relies largely on measures that are functions of relative
differences in mortality rates, is undermined by the failure to recognize
that, for reasons related to the shapes of underlying risk distributions,
as mortality declines relative differences in mortality rates tend to
increase while relative differences in survival rates tend to decrease.[2-
4]
The authors may be correct that mortality inequalities have been
increasing, as illustrated in Table 2 of reference 5. But the subject
must be studied with measures that are unaffected by changes in the
overall prevalence of an outcome, such as that described in reference 5 or
more fully on the Solutions sub-page on the Measuring Health Disparities
page of jpscanlan.com.[6]
Researchers are increasingly recognizing that relative differences
and other standard measures of differences between outcome rates are
problematic for measuring health inequalities because of the way such
measures tend to be affected by the overall prevalence of an outcome.[7-
10] Whether or not the method mentioned in the prior paragraph solves the
problem, health inequalities research cannot be of value without
consideration of the measurement issues.
References:
1. Thomas B, Dorling D, Davey Smith G. Inequalities in premature
mortality in Britain: observational study from 1921 to 2007. BMJ
2010;341:c3639.
2. Scanlan JP. Can we actually measure health disparities? Chance
2006:19(2):47-51:
http://www.jpscanlan.com/images/Can_We_Actually_Measure_Health_Dispariti...
(Accessed July 22, 2010).
3. Scanlan JP. Race and mortality. Society 2000;37(2):19-35
http://www.jpscanlan.com/images/Race_and_Mortality.pdf (Accessed July 22,
2010)
4. Scanlan JP. The Misinterpretation of Health Inequalities in the
United Kingdom, presented at the British Society for Populations Studies
Conference 2006, Southampton, England, Sept. 18-20, 2006:
http://www.jpscanlan.com/images/BSPS_2006_Complete_Paper.pdf (Accessed
July 22, 2010).
5. Scanlan JP. Measuring Health Inequalities by an Approach
Unaffected by the Overall Prevalence of the Outcomes at Issue, presented
at the Royal Statistical Society Conference 2009, Edinburgh, Scotland,
Sept. 7-11, 2009:
http://www.jpscanlan.com/images/Scanlan_RSS_2009_Presentation.ppt
(Accessed July 22, 2010).
6. Solutions sub-page of Measuring Health Disparities page of
jpscanlan.com:
http://www.jpscanlan.com/measuringhealthdisp/solutions.html (Accessed July
22, 2010).
7. Carr-Hill R, Chalmers-Dixon P. The Public Health Observatory
Handbook of Health Inequalities Measurement. Oxford: SEPHO; 2005:
http://www.sepho.org.uk/extras/rch_handbook.aspx (Accessed July 22, 2010).
8. 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 (Accessed July 22, 2010).
9. Bauld L, Day P, Judge K. Off target: A critical review of setting
goals for reducing health inequalities in the United Kingdom. Int J
Health Serv 2008;38(3):439-454.
10. Eikemo TA, Skalicka V, Avendano M. Variations in health
inequalities: are they a mathematical artifact? International Journal for
Equity in Health 2009;8:32: http://www.equityhealthj.com/content/pdf/1475-
9276-8-32.pdf (Accessed July 22, 2010).
Competing interests:
None declared
Competing interests: No competing interests
Re: Study rates instead of SMR
Thankyou Luc, for so clear and gogent an explanation.
I felt something was being tilted very wrongly in this article. The
claim that mortality 'inequality' between rich and poor was now at its
worst was frankly incredible.
As you have neatly explained, this is not so.
Looking out over our estuary, the incoming tide rises, and lifts all
seaworthy boats. The boats nearest the sea will rise first. And over
time, some rise more than others, depending how low their starting
position in estuary. Differences narrow. But differences are not nearly
so important as the hope that everybody's boat is rising.
I wonder what will happen as the economic tide turns, and ebbs ?
Competing interests:
None declared
Competing interests: No competing interests