When you go on a fishing expedition, you hope to come back with some fish to fry. But even if you don’t catch anything, you can still come home with a “ fish story” to tell. Perhaps about the “one that got away.” Or even the one that was never there—which is the sort of “whopper” statisticians sometimes like to tell.
I’ve warned you about these kinds of statistical fishing expeditions before, in my report “The Secret to Spotting the Truth Behind the Headlines.” And here in the Dispatch, I’ve pointed out several specific examples of bad science and false alarms, as well as the reasons why modern medical science is so broken in general. Recently, there’s been more “news” about dietary supplements credulously reported by both the medical and the mainstream media.
According to one recent study in the Archives of Internal Medicine, calcium supplements reduce mortality in older women. However, other nutrient supplements increase it.
Most of the media outlets that reported on this study essentially stopped there. But wait, there’s more…
Because when you dig a little deeper into the details of this so-called “scientific” investigation, you find that the women in this study who took dietary supplements didn’t really have higher mortality rates. In fact, the supplement-takers were healthier overall!
And because the supplement-takers were actually healthier, on average, with better diets, more physical activity, etc. the statisticians had to “adjust” their analysis to show their supposed results that dietary supplements led to increased mortality. They could just as well have “adjusted” results to show that it was poor diet, lack of exercise and heavy cigarette smoking that had lower mortality!
The fact is, iron was the only nutrient that showed a dose-response effect on mortality. (The higher the dose of iron, the greater the mortality.) Demonstrating a dose-response effect, even in a statistical study, gets you closer to real “forensic medicine” in proving that a substance caused a disease or a death. All the other nutrients in this study, supposedly associated with slight increases in mortality, did not demonstrate a dose-response effect. So biologically, these results aren’t at all meaningful.
I’ve warned about the dangers of iron supplementation before. Unless you have a frank, diagnosed, clinical iron deficiency, the vast majority of people do not need to supplement with iron. Some pregnant women, and some women of child-bearing age are virtually the only exceptions—and even then the “need” for supplemental iron is rare.
But aside from that one meaningful insight (which is hardly “news”), nothing of value came out of this study.
To make matters worse, the authors themselves admitted that the study was not even designed to test dietary supplements, and that their results were just “incidental findings.” There was no scientific hypothesis about supplements proposed for the study in the first place. And the “results” were reported without any biological plausibility. This kind of “evidence” would never pass muster in a court of law. And it shouldn’t pass the common sense court of public opinion either.
But now these statistical fishermen can (and did) conclude that we need “more research”—thus helping to secure more funding (and their own careers). Score one more for throwing good research money after bad.
“Dietary supplements and mortality rate in older women: the Iowa Women’s Health Study.” Arch Intern Med. 2011; 171: 1,625-1,633