Geoffrey Rose: Q&A with Socrates (Part 5)

SOCRATES:  Geoffrey, that’s an interesting graph you showed me.  Let’s talk about it some more.

ROSE: I’m glad it piqued your interest, Socrates

Interesting graph

SOCRATES: So, it’s on the basis of a chart such as this one that you have developed your theory, isn’t it?

ROSE: Yes, Socrates.  If you take each setting individually and ask “why do some people have a higher blood pressure than others?” you may reasonably answer that a variability in a large number of predominantly genetic factors among individuals account for this gradual, continuous curve.  Pickering’s position, as you recall, was that hypertension simply identified the individuals at the higher tail end of the distribution.

But if you juxtapose the 2 curves on the same graph, as I did here, and ask the question “why is hypertension common in Britain but absent in Kenya?” you can see that the answer has nothing to do with the characteristics of individuals but everything to do with the determinants of the population mean.  What distinguishes the two groups is a shift in the whole distribution—a mass effect acting on the population as a whole (1).   And although this graph depicts two different populations, the shift is exactly what happens to a given population that has migrated from a low prevalence setting to a high prevalence setting.

SOCRATES:  All this is very interesting, Geoffrey.  Let’s take it one step at a time.  When Kenyan cattle herders move to London their blood pressure distribution moves to the right, even if they don’t become civil servants, is that correct?

ROSE:  You’re being facetious but yes, Socrates, the curve looks more like that of British civil servants.

SOCRATES:  And the 2 curves—the old one and the new—will not have the exact same shape, will they?

ROSE: Assuredly not.  Blood pressure distribution curves in populations where hypertension is present are positively skewed, meaning there is a larger number of patients to the right of the peak.

SOCRATES:  Now when a population of herders moves to London, it is really a collection of individual herders, is it not?

ROSE: Of course!

SOCRATES:  If you take an individual herder whose blood pressure was well to the left of the mean before migrating, and another herder whose blood pressure was well to the right of the mean before migrating, where would their new pressures lie with respect to the new mean after they move to London and become civil servants?

ROSE:  First of all, Socrates—and I know that statistical analysis was not yet established when you lived on Earth—I think you mean the mode rather than the mean, since we are dealing with skewed distributions.

Secondly, as you well know, the first generation of immigrants rarely gets to the point of working in the civil service.  If a Kenyan herder came to London, he would likely take on a menial job and live under very severe conditions.  The distribution curve for such Kenyans is likely to be much worse than that of civil servants.

But to answer your question, it is likely that an immigrant would roughly maintain his position in the curve with respect to the mode, although many of those with pressures  below the mode end up with a pressure above the mode, otherwise there would not be an increase in the positive skew of the curve which you see in hypertension.  But in truth, we do not have data measured in the same persons before and after a migration to be able to answer your question factually.

SOCRATES:  I appreciate your correction, Geoffrey, and I understand and follow what you are saying.  In other words, you think we are dealing primarily with a shift in the individual data points leading to a shift in the curve.  You don’t think there would be too many cases of people who start off above the mode and end up below it after migration?

ROSE:  That’s not inconceivable in rare individual cases, but to imagine this as a general rule would be needlessly complicated.  It would mean that the determinants of the baseline blood pressure distribution have nothing to do with the response to the new environmental conditions and with the final blood pressure distribution, yet both distributions would be smooth and continuous.  Pretty unlikely, if you asked me.

SOCRATES:  In other words, most individuals either shift up or remain stationary.

ROSE:  I would think so.

SOCRATES: I agree with you.  And I take it that you think that the shift to a new distribution is primarily due to new environmental conditions.

ROSE: Exactly.  It’s only by looking at two different populations that the environmental effect can emerge.  One point I have emphasized to illustrate this is that if everyone smoked 20 cigarettes a day, then clinical, case-control, and cohort studies alike would lead us to conclude that lung cancer was a genetic disease (1).  You need a heterogeneity of exposure to be able to distinguish the effect of environmental causes.

SOCRATES:  So tell me, Geoffrey, what then is the cause of hypertension?

ROSE:  Well, I don’t know that there is any one cause.  It is a complicated condition with multiple factors likely playing a role.  As you may know,  I was co-principal investigator in the Intersalt study along with Jeremiah Stamler, and we observed a positive relationship between salt excretion and the slope of blood pressure with age in this study of more than 10,000 subjects in 52 centers worldwide.  Within centers, blood pressure itself was also positively related to salt excretion in a way that was partly independent of body mass index and alcohol intake, but the relationship was not present across centers.

SOCRATES: Geoffrey, I can hardly follow your jargon here, but would it be fair to say that salt intake is not an evidently strong cause of hypertension, that it may have an effect in conjunction with other factors, and that the findings were not as strong and clear-cut as anticipated?  A debate ensued, the topic is still under dispute more than 25 years after the publication of your report, and there are some who even suggest that salt restriction may be harmful.

ROSE:  I’ll grant you that.  There is undoubtedly a multiplicity of factors that muddy the waters.  We did find a positive relationship between body mass index, alcohol intake, and high blood pressure, for example.

SOCRATES:  But this inconsistent with the graph you are so proud of, the one about shifting curves.

ROSE:  What do you mean?

SOCRATES:  I mean that the graph you produced is a striking one, where as you said, the entire blood pressure distribution curve is shifted to the right.  This suggests the presence of a single or a few strong factors that work on the entire population to cause the shift.  Don’t you find it odd that so far, nothing of the sort has been identified?

ROSE:  I disagree with you Socrates.  Granted, we have not identified simple factors that account for hypertension the way we hoped, but we have repeatedly demonstrated that when groups of individuals move from society where the prevalence of hypertension is low to societies where it is high—as in industrialized countries—the curve shifts.  Similarly, when pre-industrial countries transition into more urban and industrial societies, the curves for blood pressure and other coronary risk factors also shift.  That has to point to environmental factors associated with industrialization.

SOCRATES:  The “ills of affluence (2)” as you put it.

ROSE:  Exactly.

SOCRATES:  I’m not convinced.

ROSE:  What’s the problem?

SOCRATES:  You’re missing an important piece of information.

ROSE:  What would that be?

SOCRATES:  What happens to British civil servants who become cattle herders in East Africa?

ROSE:  What?!  You’re joking again.

SOCRATES:  I’m serious, or at least, the topic raised by the question is serious.  According to your theory, the blood pressure curve should shift to the left.  Is there any evidence of that?

ROSE:  As you can imagine, Socrates, there may be occasional idealists who give up the comforts of living in Western societies and move into backwards surroundings to adopt a primitive lifestyle, but not nearly enough of them to allow us to do epidemiological studies.  So the answer is no, I do not have evidence one way or another about this.

SOCRATES:  But that’s a very important point, and it’s not obvious to me that if indeed there were enough civil servants who traded their desk, their suburban home, and their predictable future for a cow, a vast expanse of wilderness, and exposure to drought, injuries, or malaria, that their blood pressure would necessarily improve.

ROSE:  I don’t know either, but to me, this is all contrived and unhelpful.  The reality remains that millions of people are moving the other direction and the prevalence of hypertension, obesity, and diabetes is skyrocketing.

SOCRATES:  Which leads me to my second objection.

ROSE: And what is that, Socrates?

SOCRATES:  Don’t be so sullen, Geoffrey, and don’t take this personally.  I consider you a philosopher, not a sophist, willing to follow the truth, wherever it takes you.  But let’s take a break now and pursue this inquiry on another day.


Bold References:

(1) Rose G. Sick Individuals and Sick Populations.  Int J Epidemiol (1985) 14:32-28

(2) Rose G. Rose’s Strategy of Preventive Medicine.  2nd ed. Oxford University Press Inc., New York.  2008. P 36


UPDATE:  This series is now complete and consists of a total of 9 posts.  Beginning with Post #3 I am aided in this exploration by a legendary cross-examiner who interviews Geoffrey Rose’s directly and in depth in an informal Q&A format.  Quick links to the part are here: 1, 2, 3, 4, 5, 6, 7, 8, 9

2 thoughts on “Geoffrey Rose: Q&A with Socrates (Part 5)”

  1. Michel,
    Greatly enjoying this series of articles, and looking forward to the next installment. You are doing a fabulous job teasing out the shortcomings of Rose’s theories, and in a highly entertaining way.


Leave a Reply

Your email address will not be published. Required fields are marked *


You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>