Correlation versus causation: The media typically gets it wrong

859634_41406292How many times in the last week have you seen headlines such as Coffee as a memory booster, or How Diet Soda Makes You Fat? Well, according to a study conducted by researchers in training at Harvard and the NIH, these observational studies (assessing correlation) are covered more often in the media than randomized controlled trials (RCTs) which are usually far more reliable.

Drs. Senthil Selvaraj, Durga S. Borkar and Vinay Prasad looked at the clinical studies covered by a few of the leading newspapers in the U.S. The Wall Street Journal, USA Today, The New York Times, Los Angeles Times, and the San Jose Mercury Times. They then looked at the trials covered by the top five clinical journals The New England Journal of Medicine, The Lancet, Journal of the American Medical Association, Annals of Internal Medicine and PLoS Medicine. To assess study quality, researchers used the USPSTF (U.S. Preventive Services Task Force) hierarchy of evidence, which evaluates studies based on design and quality.

What they found was [n]ewspapers were more likely to cover observational studies and less likely to cover RCTs than high impact journals. Additionally, when the media does cover observational studies, they select articles of inferior quality. Newspapers preferentially cover medical research with weaker methodology."

The authors point out that this was just one study and did not take into account any press releases put out by the newspapers. They also acknowledge that a large part of what newspapers and other media sources decide to cover is based on appeal to readers.

ACSH s Dr. Elizabeth Whelan acknowledges that point. However, she adds, this desire to appeal to readers does not excuse the media from getting it right when it comes to the difference between observational and randomized controlled trials. No matter the subjects covered by the media, it is imperative that they are able to grasp the difference between these two. Correlation and causation are two very different things.