Figures that Lie -- and Liars that Figure
(June 13, 2017)
#Not all that long ago, I boasted in these pages that I had stopped getting my news from the internet and that this bit of discipline had immeasurably improved my life. This mostly remains true. Mostly. I do sneak back briefly to my favorite website a couple of times a day now, prepared to quickly depart if nothing substantial presents itself. Will I weaken further? Stay tuned.
During one of these adventures, a headline froze me in place. This one -- “just one alcoholic drink a day increases risk of breast cancer, study says". !!!!! We've known for a long time that two-plus daily alcoholic drinks does this. But only one drink? And a “small” one, at that? That's very scary. Who would want to drink wine at all any more after hearing that?
Well, I would. In fact, it makes me want to drink MORE. Because I don’t believe them. Not because they’re lying – they aren’t. But they are doing something worse. They are misleading their trusting readers with truth that is so limited that it should be classified as a lie.
This particularly nasty little bit of mathematical legerdemain -- the "increases risk" thing -- is so common that now any time I see it, all my merde detectors go off. I automatically assume that I am looking at the work of some hack who was too lazy to do any real research. And usually, sad to say, I'm right.
Let me give you an example. If I never leave my basement, I have decreased my chances of being killed by a meteorite by 85%. Is that a life-changing statistic for me? Not unless I am an idiot. My chances of being hit by a meteorite are so small in the first place that increasing (or decreasing) my risk of it by 85% is completely meaningless.
This is the subject of relative versus absolute risk, and in health and disease it is a particularly important one. It is easy to frighten people by insinuating that the chances of their acquiring some nasty disease are "greatly increased" when it’s something that for all practical purposes is not likely to happen at all.
Back to that article. Passing by the alarmist headline, the first paragraph actually ups the ante and tells us that alcohol “significantly” raises the risk of breast cancer. One “small” glass does this. Wow, alcohol must be really lethal if even a small amount can kill you! I'm more terrified by the moment.
Reading on (which is something which, at that point, only 60% of the people who started the article are still doing -- another mistake the wary reader should never make) you will see that this “significant” lifetime increase in the risk of acquiring breast cancer is – five percent for pre-menopausal women and nine percent for post-menopausal, women. Frankly, that's not a lot. And it doesn’t sound significant to me.
But it seems that I am wrong! Because if you read on to the third paragraph, one of the authors of the original study is quoted as saying, "This suggests there is no level of alcohol use that is completely safe in terms of breast cancer”. And we all want to be COMPLETELY safe, don’t we?
Let’s crunch some actual numbers here. Let's take that five percent increased risk for the pre-menopausal women thing. What is being said here is that if one of these ladies is drinking one smallish alcoholic drink every day, her chances of getting breast cancer will be about 5% greater than that of the average woman in her demographic. That's her RELATIVE risk. But what is her ABSOLUTE risk? Meaning, how likely is this actually to happen?
You may have heard (correctly) that one out of eight American women gets breast cancer at some time during her lives. But listen -- that does NOT mean that every woman out there has a 13% risk of getting breast cancer. The risk varies enormously from person to person. How to figure that risk? Well, you can get a pretty good idea with an online calculator. Like this one:
Here is a typical example. A white 40-year-old woman with one child that she had after age 30, with no genetic abnormalities, no close relatives who have had breast cancer, and no previous history of cancer or treatment for same, has -- over her next five years of life -- a ZERO POINT EIGHT PERCENT (0.08%) chance of developing breast cancer. Way less than one percent. In other words, 20 out of every 25,000 women who fit that description will get breast cancer some time in the next five years. But if they all start downing one small glass of wine daily, that number goes up to -- 21. Big whoopee.
So, ladies, how do you feel about that glass of wine now? Wanna “risk” it? Maybe you don't. But now your decision is a whole lot better informed than it was before.
This use of relative risk is almost always a scare tactic, an unethical and infuriating little bit of accounting that the “health” media works continually and shamelessly. Don't let them get away with it.
--dr. diane holmes
Copyright © 2017