Beware of false knowledge; it is more
dangerous than ignorance. ~George Bernard Shaw
The media regularly provides tantalizing
information from health studies that allege people who eat dark chocolate, for example, have lower blood
pressure and can be protected from cancer. Eating dark chocolate, many readers
conclude, will thus make you healthier and live longer.
Other published studies relate better health
to drinking red wine and eating fruit and veges – come on George, eat your broccoli! The august New York Times published a blog in July that reported a study avowing
that people who ran at least 5 minutes a day lived 3 more years than folks who
didn't run. Chocolate-, wine- and sneaker-makers hop on the bandwagon promoting
sales that supposedly provide greater health and longer life.
Don't count on it.
The vast multitude of these studies only calculate
a correlation between some outcome measure
of good health (say, lower blood pressure) and differences in a single
treatment variable (say, eating dark chocolate or running). Such studies measured
blood pressure levels in a (usually small) group of people that show a
relationship between blood pressure and eating tasty dark chocolate.
These studies dominate health care and public
health assessments and rely on statistical correlation between the treatment
variable and the outcome variable to make pronouncements. They are termed
"observational" in that they may draw inferences about the possible
effect of a treatment on the participants, but the assignment of participants
into a treated group versus a control group (the group that does not receive
the treatment – like chocolate) is outside the power of the investigator(s). In
other words, the investigators do not first pick certain people to be the
chocolate-eaters and others to not eat the dark stuff as part of the study's a
priori design; instead, they separate the chocolate-eaters and non-eaters after they have the data.
So who'd of guessed that just eating chocolate
or drinking red wine could really provide better health? Not someone who
actually understands the important difference between correlation and
causation. Correlation is a statistical
measure that indicates the extent to which two or more variables fluctuate
together. A correlation between variables, nevertheless, does not necessarily
mean that the change in one variable is the cause of the change in the values
of the other variable. Causation, however, indicates that the value of one variable is the result of
the other variable(s); i.e. there is a causal relationship between the two variables.
Folks who rush to eat Scharffen Burger
chocolate after learning about observational studies' results have perpetrated the
all too prevalent post-hoc fallacy. This
fallacy involves believing the mistaken notion that just because one thing
happens (lower blood pressure) after another (eating dark chocolate), the first
event is a cause of the second event. Post hoc reasoning is the basis for many
erroneous, but firmly-held beliefs that are espoused and acted upon all too
often by all too many people.
Observational health studies – like those
mentioned above – at best offer correlations between variables, not a confirmed
causal relationship. Such studies cannot control for other important variables
that also probably influence one's overall health outcome. Eating chocolate is
but one quite narrow, singular slice of a person's daily life. Other slices
include, your sex, how old you are, all of what you drink and eat (beyond
chocolate), how much you sleep, how much you exercise, where you live and work,
what your family's health history is, what medicines you take and whether you
smoke. Observational studies rarely account for these other factors. Thus, you
should take the results of any observational health study with several grains
of salt (good thing you're on that low-salt diet, right?). Correlation is not
causality.
A recent Slate
article illustrates – in spades – the blunder
of following the post-hoc fallacy. The authors offer a table that provides specific
calculations for how many daily minutes of your life are subtracted by eating a
single hamburger, smoking 15-24
cigarettes per day, being a couch-potato and/or residing in Sweden vs. Russia
(huh?) among other actions. They call these time units "microlives per
day" (that I first interpreted as a type of small olive, silly me). This
table also distinguishes these alleged forever-lost daily microlives' effects
between whether you're a male or a female over 35 years old. FYI, being male
rather than female costs you 4 microlives/day according to the authors. Ready
for that trans surgery now?
To be complete, the table also cites possible
factors that increase your microlives per day – like exercising, taking
statins, and/or eating fruits and veges. Eating 5 servings of fruits and veges
each day apparently can gain you 5 microlives. I don't believe it.
This table of misplaced over-quantifications
is almost-certainly based on observational studies.[1]
You shouldn't assume any factor or number in this table is causally-connected
with your health or lifespan. The authors cannot say with any validity, as they
do, that a "single dose of ionizing radiation" (from taking one
transatlantic flight) will lessen your life expectancy by 30 minutes (that's 1 microlife
per day). They believe their work offers a means of "calculating the exact
impact of daily choices on every precious minute of your life." I beg to
differ since each and every one of their listed microlives' effects is
discredited by their subscription to the post-hoc fallacy.
Such observational studies are not randomized,
controlled trials which can provide more statistically-precise information
about causality if properly designed but are more costly to conduct. Randomized,
controlled trails account for about 80% of biomedical experiments (often
called "clinical trials" like those undertaken through the FDA)
involving drug or medical device appraisals.
So by all means continue jogging, eating dark
chocolate and fruit and vegetables with a glass of Zinfandel, but don't think
such activity, by itself, will cause your health to improve or make you live even
1 microlife longer. Right now we have no idea about those specific causal
linkages, despite what people say.
[1] The table in the Slate article contains no references
identifying where any of the numbers (e.g., microlives) come from. Apparently you
have to buy the authors' book to get that source information.
No comments:
Post a Comment