Health News Review

NPR’s Richard Knox has been around the block a few times – a veteran science journalist.  And it shows in the way he covered a study pointing to an association – women who took aspirin had fewer diagnoses of melanoma.

Emphasis on association, not causation.

He allowed one of the author’s enthusiasm to come forth – because that’s what she said:

“We’re really excited aspirin could be used as a potential preventive agent for melanoma,” Dr. Jean Tang of Stanford University Medical School, the study’s senior author, tells Shots. “In terms of cancer prevention, a lower melanoma risk by 20 percent is very large and significant.”

Tang says women who regularly took aspirin for five years or more had a 30 percent lower risk.

“There’s nothing else that I know of that has as large an effect as what we’re seeing with aspirin,” Tang says.

But Knox followed that with the kind of counter clarification that we think is essential in stories about observational studies.  He wrote:

The data came from the 22-year-old, federally financed Women’s Health Initiative. But this study merely observed whether women chose to take aspirin or not, and then correlated that with whether they got melanoma. This kind of “observational” study doesn’t prove anything.

“We would have to do a large clinical trial, randomizing women to receive aspirin versus placebo, following them for 10-plus years,” Tang says.

She acknowledges that type of gold-standard study isn’t likely to happen because “it’s just too expensive to do.”

So this study (like so many others) leaves open the question: What should anybody do?

He ends with this important context:

But skeptics … agree with enthusiasts … on one point: “The worst thing,” Tang says, “would be [for people to think], ‘I can take aspirin, and that justifies me doing indoor tanning.’ That is not the right message.”

Tang says one reason melanoma is rising fast among young women is that so many of them love those tanning booths.


Follow us on Facebook, and on Twitter:


GreggD posted on March 12, 2013 at 9:08 am

After reading this I searched google news “aspirin and melanoma” to find some other sources stories, here’s one from CBS News data March 11, 2013, 12:08pm: here.

    Gary Schwitzer posted on March 12, 2013 at 9:17 am


    Thanks for your note.

    The CBS piece offers a striking contrast to the NPR piece we blogged about. The CBS piece offers no caveats about the limitations of observational studies, none of the explanation about association versus causation. Instead, the journalist chose words and phrases such as “protection from melanoma” and “protect against the deadliest form of skin cancer” when, in fact, such a causal link was not and can not be proven by such a study. Knox/NPR made that clear. CBS did not.

Greg Pawelski posted on March 12, 2013 at 11:46 am

I know you are posting about an example of how to “report” on an observational study. Kirsten Moysich, PhD, professor of oncology at Roswell Park Cancer Institute in Buffalo told Medscape Oncology, “This study provides credible evidence of a role for aspirin in the chemoprevention of melanoma, and adds to the existing body of research that implicates aspirin use in the prevention of tumors, including colon, breast, ovary and lung cancers.” There comes a time when something looks like a duck, walks like a duck, and quacks like a duck…….. While medicine and most of its discoveries have been observational, yet another has been linked to reduced risk by aspirin.

    Gary Schwitzer posted on March 12, 2013 at 12:08 pm


    I agree with you that “there comes a time” when the weight of evidence from observational studies becomes very strong. The association may be so strong – and seen so many times in different studies – that it will lead to recommendations for populations on how to act, or that it would lead individuals to make informed decisions.

    Nonetheless, I suggest that there will never come a time when it is accurate to use causal language in news stories when describing observational studies. The fact remains: association does not equal causation. Language matters. Accuracy matters. That is our message to journalists and to all who communicate about such studies.

Srihari Yamanoor posted on March 16, 2013 at 3:53 pm

This is indeed a good example! Thanks for the write up.

Daniel Pendick posted on March 18, 2013 at 10:02 am

RE: “There comes a time when something looks like a duck, walks like a duck, and quacks like a duck…….. While medicine and most of its discoveries have been observational, yet another has been linked to reduced risk by aspirin.”

Sorry to bring it up again, but that’s what we all thought about the WHI, and proceeded to tell women that hormone replacement offered all kinds of benefits. (I was one of the people relaying that happy news.) But you know what happened when they did the RCT. On the other hand, I feel pretty good telling people that a nutritious diet and regular exercise could prevent disease. If I’m wrong, there is no harm.