Science in the Triangle

The National Institutes of Health course on “Medicine in the Media” I am attending is a reminder that scientific studies cannot be taken at face value. Somebody needs to double-check the numbers, put them into perspective and expose hidden conflicts of interest researchers may have.

Why? Here’s one reason: I was surprised to hear from Barry Kramer, an oncologist and the associate director of disease prevention at the NIH, that medical schools do a very poor job teaching budding physicians how to read and interpret medical studies. No wonder, many aren’t up on the literature and drugmakers have so much sway over what drugs get prescribed.

Here’s another reason: Peer-reviewed medical journals and medical conferences are dominated by study results that are positive. Studies with negative results rarely get published.

And another reason: Even respected journals will publish study results without enough data to support outcome claims stated in the summary.

The three main teachers of the course - Lisa Schwartz, Gil Welch and Steve Woloshin, all physicians and medical researchers at Dartmouth Medical School - provided us with tools to read and interpret medical studies. Some highlights:

• Biostatistics is like shopping. Doing the math to determine whether study results are really as important as the authors want to make us believe is like determining how much a sale shaves off prices and how much it saves in dollars.
• Always ask what the harm of the treatment is. In cases where the clinical benefit is small, the harm may outweigh the benefit even if the study results are positive.
• Results from randomized trials tend to be better than results from observational studies. But sometimes, observational study results are the only data available. Then it becomes important to make sure the study results aren’t influenced by something not measured, such as socio-economic factors.