Another coffee observational study – another round of misplaced emphasis

Here were some of the headlines or lead lines:

  • Coffee May Be Part of the Recipe for a Longer Life (WebMD)
  • “Coffee drinkers who worry about the jolt of java it takes to get them going in the morning might just as well relax and pour another cup.” (MSNBC)
  • Daily Coffee May Help Keep Grim Reaper Away (HealthDay)

In others, cautions were highlighted:

“Indeed, the study has lots of limits.

One is the risk that the effect of other things that change health, like smoking, may not have been completely filtered out. When all of these other influences were left in, coffee drinkers actually tended to have a higher risk of dying than those who abstained.

The participants only supplied information about their coffee habits once, with no updates if they changed over the next 13 or so years. Also, people in poor health might drink less coffee, though the authors did exclude people who started out with cancer or cardiovascular disease.”


“But that finding should be interpreted with caution, researchers said, because coffee habits were only measured at one point in time — and it’s unclear what ingredients in java, exactly, could be tied to a longer life.

“For those who do drink coffee, there’s no reason to stop. Periodically someone will say it’s bad, but I think this strengthens the view that it’s not harmful,” said Dr. Lawrence Krakoff, a cardiologist from the Mount Sinai School of Medicine in New York, who wasn’t involved in the new research.

“Whether it’s beneficial — without knowing the cause, it’s hard to say,” he told Reuters Health. “I wouldn’t encourage people to suddenly drink a lot of coffee with the expectation of benefit.”

And the story finished strong as well:


The study was not without limitations, researchers noted, including that Freedman’s team only knew how much coffee participants were drinking at one point in the mid-1990s — and those patterns could have changed over time.

In past years, some researchers have been quick to jump to conclusions about the effects of coffee. One 1981 study linking the beverage to pancreatic cancer, for example, and also published in the New England Journal of Medicine, was later criticized for its methods.

“There have been false associations with coffee in the past,” Krakoff told Reuters Health. He said it’s important to be cautious about the new findings.

“My thinking is that these associations are very interesting but until you really link it to a causative mechanism, it remains vague.”

Because of the so-called observational design of the new study, it can’t prove that coffee was directly responsible for the lower risk of death in some participants.”

We remind journalists – and the public – that the language used to describe observational study findings is important.  Our primer on the topic should help anyone understand why this is important – and why language suggesting cause-and-effect is misleading.

Interesting addendum on May 18:

Deborah Kotz of The Boston Globe writes:

I wondered why this study was published at all and, no less, in one of the world’s leading medical journals.

“I’m not a fan in general of epidemiological studies of this type because they can be confounded by variables,” Dr. Jeffrey M. Drazen, editor-in-chief of the New England Journal of Medicine, told me in an interview. While the researchers took into account more than a dozen differences (such as smoking rates, body weight, education level) that distinguished heavy coffee drinkers from those that abstained, Drazen said they couldn’t account for everything.

For instance, the survey on which the results were based didn’t ask participants whether they had health insurance or had high blood pressure or high cholesterol levels — all of which could factor into their expected lifespan. It also looked at coffee drinking habits only at one point in time, and people may change their consumption over the years.

Drazen said that there was some disagreement among the journal’s editorial board about whether to publish this study but added that he’s often more inclined to publish a study “if it touches a nerve” because it’s important to stimulate discussion in the community at large. “We went into this fully realizing the strengths and weaknesses of the study,” he added. “I think the authors were very cautious in their interpretation.”

But one of the country’s leading medical researchers, Dr. Steven Nissen, chair of cardiovascular medicine at the Cleveland Clinic, said the study is apt to be misinterpreted by the public at large. Many may assume from the findings — and headlines such as “coffee drinkers live longer” — that all they need to do is drink coffee if they want to erase the effects of bad health habits such as smoking.


You might also like


Please note, comments are no longer published through this website. All previously made comments are still archived and available for viewing through select posts.

Anna Hayward

May 17, 2012 at 9:55 am

What sort of coffee? As any coffee shop customer knows, there are a million different types, including decaf. Here in UK, much of the coffee we drink is weak, freeze-dried instant coffee, but that varies from pathetic floor-sweepings to premium “Freeze-dried ground coffee”. There are also filter coffees, espresso coffee, that disgusting muck that comes out of machines on the railway station, iced coffee in a can… And all of these coffees, being based on a natural product, vary in their caffeine content and other constituents. So what bit of coffee is the problem, if any, and what sort of coffee is worst?

Lyn Carew

May 17, 2012 at 12:49 pm

The study is almost useless. Plain and simple – observational studies do not establish cause and effect. As Dr. Steve Nissen, cardiologist of the Cleveland Clinic was quoted in a USA Today (5/17/12) article by Liz Szabo, “This study is not scientifically sound. The public should ignore these findings.”

Julia Belluz

May 18, 2012 at 10:28 am

Great post! I looked at the Canadian headlines and we were no better at mixing up correlation and causation. A perennial problem in health reporting and one worth drawing attention to.