Research published in the New England Journal of Medicine, “Genetically Determined Height and Coronary Artery Disease,” has journalists scrambling to find ways to explain the results of this intriguing observational study.
But because it’s an observational study, and if you read this blog regularly, you probably know what’s coming next.
Association ? causation.
So news stories should avoid using causal verbs to describe results of observational studies.
It is probably the leading source of public confusion about health care research in the news.
The New York Times hedged its bets with a qualifier in the headline, “Shorter Stature May Pose Higher Risk of Heart Disease.”
But, in the body of the story, it jumped right in with the causal phrase “increases risk.” Excerpt:
To the surprise of researchers who had thought the very notion a joke, an international consortium of investigators reported on Wednesday that shorter stature increases the risk of heart disease.
After gathering genetic data from nearly 200,000 men and women worldwide, the investigators found that each extra 2.5 inches of height brings a 13.5 percent reduction in heart disease risk. The relationship is present throughout the range of adult heights.
…Of course, the increased risk pales in comparison to that caused by smoking, which elevates the odds of getting heart disease by 200 to 300 percent.
I added the emphasis to the phrases of concern. I argue that you can’t establish risk when you haven’t established cause and effect. I also acknowledge that I’m being a semantics purist, and that not everyone might agree. However, I think there are better ways to say this. Some appear in stories below. Some appear in our primer, “Does The Language Fit The Evidence? – Association Versus Causation.”
The Times, did, though, inject a little skepticism, just not to the main point I’m making. Times story excerpt:
“Not everyone is impressed by the new report. Dr. Kari Stefansson, chief executive of deCODE Genetics, said the increase in heart disease risk was real but so small as to be not very meaningful. “It’s a weak effect,” he said.”
But there was an important discussion at the end of the Times piece on the difficulty of establishing more than association:
Studies of huge DNA databases, like the new research, show the power of genetics to resolve questions that had seemed unanswerable.
“The Achilles’ heel of epidemiology has been the inability to distinguish between associations that are correlations and those that are causal,” (Dr. Michael Lauer, director of cardiovascular sciences at the National Heart, Lung and Blood Institute) said. Until now, the only way to do that was through large, expensive and time-consuming randomized controlled clinical trials.
For example, the Women’s Health Initiative studied whether taking hormones after menopause could reduce the risk of heart disease, which seemed well established. Thousands of women were told to take hormones after menopause, or not to. After years of study and millions of dollars, the result was clear: The hormones did not protect against heart disease.
Of course, no randomized trial can test whether height is linked to heart disease risk. But genetic variations are distributed at random in populations, and genes are not affected by lifestyle or environmental factors that have to be accounted for when researchers analyze data they get by simply observing populations.
In a sense, Dr. Lauer said, we are nature’s random experiments.
The Associated Press put a qualifying hedge in its headline just as the NYT did, “Short People’s Genes May Confer Higher Heart Risks.” But the AP story followed with causal language – “risk…rose” – just as the Times did.
“The risk of coronary artery disease – clogged arteries – rose 13.5 percent with each 6.5 centimeter (about 2.5 inches) decrease in height.”
Scientists have long considered there to be a link between height and heart health, but the latest research found that genes controlling height were directly linked to heart disease risks.
To be sure, height is only one of many factors that affect the level of risk.
“In the context of major risk factors this is small – smoking increases the risk by 200-300% – but it is not trivial,” Nilesh Samani, a professor of cardiology at the University of Leicester and lead author of the study, told the BBC News website. “If you’re 6ft 1in, you still need to stop smoking.”
I can’t imagine how badly readers are spinning after that explanation, which allows the researcher to leap from an observational finding to comparing “stature risk” – as it were – to smoking – even in the context of calling it a “small” risk.
Lest you think I’m splitting hairs and holding out for an ideal that is unattainable, see these two examples:
“Shorter people are more likely than taller folks to have clogged heart arteries, and a new study says part of the reason lies in the genes.”
“However, while the current study was able to show an association between genetics, height and a higher risk of heart disease, it wasn’t able to prove a cause-and-effect relationship.”
Follow us on Twitter: