## The Weight of Evidence

How do you
determine what is the optimum weight for a person? It's probably
different for people of different heights, so you could hold beauty
contests for people in different height groups, and consider the
winners to be of optimum weight. However, I'm sure that you'd find
that the judges would be unable to disregard other differences between
the contestants, so that it would be impossible to get answers that
would be widely acceptable. It would be better to use some measure
of health to arrive at an optimum weight, but it would be hard to
get a generally acceptable measure, since the values of some measures
of health would vary differently than would the values of others.
Probably the best generally acceptable measure would be longevity.
You could reasonably consider the optimum weight to be that weight
which is conducive to the longest life. Of course, you would have
to take height and build into consideration because the optimal
weight seems likely to be different for people of different heights
and build.

That's what
the Metropolitan Life Insurance Company did in 1943 when they published
what became widely-used tables of "desirable", i.e. lowest-mortality-rate,
weights for men and women aged 25 to 59 and of various heights.
The subjects had taken life-insurance exams fully dressed, so the
heights in the tables were considered to include one inch contributed
by shoes and the weights were considered to include three pounds
of clothing for women, five for men. The data were divided into
columns for people with small, medium or large frames, based on
the width of the bones at the elbow (different values used for men
and women). Collection of the data was not done in a rigorously
scientific manner: frame size was sometimes estimated by eye, and
some of the other data were sometimes obtained by questionnaire
rather than measurement.

Inspecting
these tables, Reubin Andres was struck with the fact that people
of different ages were all lumped together to get the data. What
if the ideal weights varied according to age? To find out whether
or not this was true, he examined the same data that had been used
to construct the tables, but made separate analyses for men and
women in different decades of life. What he found and
reported in his book (Principles of Geriatric Medicine, Reubin
Andres et al. McGraw-Hill (1985) pp. 311 ff) was that
the optimal weight (remember, we're talking about mortality, not
winning beauty contests) increased as one got older. He also found
that the optimal weights as he determined them were very similar
for men and women. In the Metropolitan Life charts, considerably
lower optimal weights are given for women.

Having to tabulate
both height and weight is inconvenient, and studies of optimal weight
now avoid this by using a mathematical trick. Instead of recording
heights and weights, a composite index called the body mass index
(BMI) is used. This is a measure of weight divided by height squared,
and would be a good measure of overweight if every body contained
the same percentage of fat. However, women's bodies in general contain
more fat than men's bodies, and there is great individual variation
in body fat content, particularly in women. Since fat has a lower
density than other body constituents, it affects the BMI differently
than does an equivalent volume of other body constituents such as
water, protein and bone. Moreover, the effect of body fat on mortality
seems to be a function of its location in the body, with fat occurring
in the abdominal area or above associated with hypertension and
other undesirable conditions, while fat distributed primarily in
the lower part of the body is relatively benign. Thus, although
a 200-pound person who is six feet, three inches tall, and a 150-pound
person who is 5 feet, six inches tall both have BMIs of about 25,
they don't really represent similar mortality risks.

However, that's
how the data are presented in the medical literature, so, until
we have the resources to undertake original studies, let's see what's
in the literature, but maintain a skeptical view of what's presented.

Another reason
a skeptical view is warranted is that the data in some papers presented
in medical journals are presented in such a way as to suggest that
the authors are operating with a presumption that overweight is
a bad thing and that their task is to demonstrate that this is so.
I'd like to present two pieces of evidence for this. The first is
a paper published in The New England Journal of Medicine
in 2003 (Calle, E. E. et al. "Overweight,
Obesity, and Mortality from Cancer in a Prospectively Studied Cohort
of U.S. Adults." The New England Journal of Medicine
348:17 1625-38 (2003)). The title page lists "Conclusions"
and this reads, in its entirety: "Increased body weight was
associated with increased death rates for all cancers combined and
for cancers at multiple specific sites." However, a glance
at the data showing the relative risk of men of different body mass
indices dying from cancer shows that the risk is lower for
men with a BMI in the range of 25.0 to 29.9 (considered "overweight")
than in men with a BMI which is lower (considered "normal")
or higher (considered "obese") than this. Since the vast
majority of subjects were in the "normal" or "overweight"
groups, increased body weight in men (alas, not in women) was shown
by this study to be associated with decreased death rates
for all cancers combined. This was acknowledged by the Journal
printing my letter in a subsequent issue (see Deutsch, M. E. "Obesity
and Cancer." The New England Journal of Medicine 349(5):
502-4 (2003)).

There's something
else we should consider. Since it appears that we all must die from
something, maybe the apparent protective effect of "overweight"
on cancer in men is caused by "overweight" increasing
their chance of dying from something else, such as heart disease,
and the apparent increase in cancer deaths in "overweight"
women is a result of their reduced susceptibility to some other
cause of mortality, such as heart disease. In our quest to determine
optimum weight or BMI with respect to mortality, we must look for
studies which report on total mortality.

Now let us
consider a paper whose results suggest that what is now considered
to be overweight should be considered normal weight for both men
and women. Once again, you wouldn't know it from reading the misleading
title or the paragraph entitled "Conclusions," but here,
at least, you can determine the truth by reading the paragraph entitled
"Results" on the front page.

This paper
(Flegal, K. M. et al. "Excess Deaths Associated With Underweight,
Overweight, and Obesity." JAMA 293: 1861–7 (2005) appeared
last year in the Journal of the American Medical Association
and it described studies of the relative death rates of subjects
who were underweight, so-called "normal" (BMI between
18.5 and 25), overweight (what was referred to as grade 1 overweight
in the previous paper) and obese (what the previous authors called
grade 2 and higher overweight). As promised by the title, the authors
provide estimates of the excess deaths associated with having a
body-mass index not in the "normal" range, but the title
(and the conclusions) are a bit misleading, since the excess deaths
associated with being in the overweight group are a negative
number. That's right, they found that there were 86,094 fewer
deaths in the overweight group than in the same number of subjects
in the "normal" group. If you define normal as the group
with the lowest mortality, then you should award this title to the
group they call overweight. You would not be able to come to this
conclusion reading the title and conclusions, and it is only grudgingly
and indirectly admitted in the results paragraph by the statement
"Overweight was not associated with excess mortality."
That's correct, of course. It's associated with lower mortality.

Okay, I've
presented evidence, provided by investigators reluctant or unwilling
to admit what is shown by their data, showing that the optimal range
of BMI (for maximum longevity) is not the generally accepted normal
range, but the internationally accepted "grade 1 overweight"
range of 25 to 30. Before going on to suggest how one might achieve
an optimal BMI and presumably an optimal weight (that is, one conducive
to maximal longevity) I'd like to point why I am pleased that I've
obtained my data from authors who have arrived at conclusions not
concordant with mine.

The double-blind
collection of data is widely used because many studies have shown
that investigators unconsciously affect the data they collect. They
may "accidentally" record an expected number rather than
the one actually observed. They may even unconsciously reveal to
experimental subjects (even to rats!) what they want them to show,
and the experimental subjects may provide what is wanted, also unconsciously.
The studies I have just referred to, however, would be hard to perform
in a double blind manner, so this wasn't done. However, the fact
that we were able to draw from the data conclusions which were apparently
opposite to what the authors expected or desired suggests that these
unconscious distortions of the data did not take place and that
we should give more weight to the conclusions I have drawn than
to the conclusions of other authors who have found what they wanted
or expected to find.

Now for what
you eat to achieve and maintain an optimal weight?

Sorry, that's
the wrong question.

Like everything
Like all other animals, we have been shaped by evolution to seek
out as food what seems palatable to us and to eat it in amounts
which satisfy our hunger and optimize our life span. Unlike other
animals, we subject foods to artificial processing, which I define
as separating the palatability of food (which evolution has taught
us to seek) from the nutrients of the food (which are the reason
that the sought-after food has led to our thriving and passing on
life-sustaining tastes to our descendants). And just as squirrels
would not thrive on peanut-flavored sawdust, we cannot thrive on
foods which lack the nutrients we need but mimic their palatability.
As for the macronutrients, protein, fat and carbohydrates, which
provide us with the calories we need, I would like to point out
that there are many essential amino acids and essential fats, but
no essential carbohydrates. The Inuit have demonstrated that people
can live on a largely carbohydrate-free diet and Vilhjalmur Stefansson
has demonstrated that this is possible by doing so under carefully
controlled conditions on a hospital ward. But excess carbohydrate
in the diet can lead to the metabolic syndrome or diabetes, while
excess fat and protein are essentially harmless, except for the
trans fats which occur in only tiny amounts in nature. So the next
question is "Has evolution, which has resulted in our having
tastes which direct us to the correct foods for maintenance of optimal
health (if we don't mess up the system by artificial processing
of foods), also directed us to having appetites which direct us
to eating the amounts of these foods which will result in
optimal weight and optimal lifespan?" And the answer is yes,
if a certain condition is met.

Let us first
consider experiments which were performed on rats. They have shorter
lifespans than people do and they are easier to control and to obtain
in genetically uniform groups. We'll get to people later. First
let us consider an experiment carried out by the late Jean Mayer
and described in his book Overweight,
Causes, Cost and Control
(Prentice-Hall (1968) pp. 73–74).

Jean Mayer
kept rats in cages containing motorized exercise wheels, and allowed
them to eat as much as they wanted, but forced them to exercise
for various lengths of time. He found that rats which exercised
less that one hour a day ate amounts of food which were inversely
proportional to the amount of exercise they performed. And these
rats had weights which were inversely proportional to the amount
of exercise they took However, rats which exercised from one to
six hours a day ate in proportion to the amount of exercise they
got, and their weights were independent of the amount of exercise
they got.

In other words,
the curve relating length of time exercising (up to six hours a
day) to amount eaten was shaped like a check mark, with the minimum
at one hour of exercise a day. The poor rats forced to exercise
more than six hours a day lost weight because they were too tired
to eat.

So here's some
advice for rats. Exercise from one to six hours a day and your appetite
will lead you to eat what evolution has found to be the optimum
amount of food, which presumably leads to optimum longevity. However,
note that if you are a lazy rat (less than one hour a day on the
wheel) and want to lose weight so that you weigh the same as the
exercising rats, you will always be hungry. This is also true if
an experimenter restricts your diet and doesn't provide exercise
facilities for you. This might get your weight and longevity into
the normal range, but provides no useful data for those interested
in having optimal weight and an enjoyable and long life. These data
also suggest why studies have shown that rats kept in cages which
is restricted, and suggests that this might not apply to rats which
exercise for an hour a day or more.

People are
more variable and less controllable than laboratory rats, but Dr.
Mayer found a way to duplicate part of his curves with people. He
found an Indian railway construction site which employed a variety
of workers whose work required the expenditure of different amounts
of calories, and who brought their food with them to the work site,
so that he could estimate their caloric intake.

He
was able to duplicate the beginning of the rat curve, but none of
the Indian workers worked so hard that he failed to eat enough to
maintain his body weight. Sedentary workers and workers in the groups
performing light work ate amounts inversely related to their dietary
needs, and like rats exercising less than an hour a day had body
weights which were inversely related to the amount of work they
did and directly proportional to their caloric intake. However,
for all workers in six groups which expended larger amounts of energy,
caloric intake was proportional to caloric needs and body weight
was a constant value independent of caloric intake (This study was
made before BMI was used for such studies. However, the groups seem
to have been large enough to minimize errors owing to height differences.)
Note that these workers, like the experimental rats, ate as much
as they wanted to eat.

So
mechanic and eat as much as you want to eat and you will maintain
the body weight which evolution has found to be optimum – the body
weight associated with maximum longevity. The only caveat is to
avoid foods which have been artificially processed in the sense
of nutritive value having been divorced from palatability.

But
wait! Suppose you can't get a job on an Indian railway project.
There's other evidence which suggests that the minimum exercise
amount to get you on the upward curving portion of the curve is
twenty minutes of hard-breathing heart-thumping exercise at least
three times a week. This is called aerobic exercise to distinguish
it from anaerobic exercise, which is so intense that it can't be
carried out for more than minutes at a time. Longer periods of exercise
are better, but their effect on appetite will simply be to make
you want to consume more calories to replace the ones you worked
off. I don't think that I have to warn you not to exercise so much
that you're too tired to eat. And the right question is, of course,
"How much should I exercise to maintain an optimal body weight?"

July
27, 2006

Marshall
E. Deutsch, PhD. [send him mail],
an inventor of medical test devices, is a founding member of Thincs,
a group devoted to exposing myths concerning cholesterol.