The media constantly bombards us with studies purporting to show new evidence about medicines we should avoid, or foods we should eat, or behaviors that either protect us or harm us. Many of these studies, because of their design, actually tell us absolutely nothing. Nevertheless, the media is not in the business of downplaying the news, so the studies are inevitably announced in the mainstream press with much fanfare and little scientific scrutiny.

So that leaves us, the consumers of the news stories, with the responsibility of figuring out whether or not a study deserves our attention. A great example was a study last week in the Journal of the American Medical Association that linked use of proton pump inhibitors (PPIs, antacids such as Prevacid, Prilosec, Aciphex and others) to the risk of hip fractures. The most important thing to understand about this research is that it was what scientists call an observational study, and not a randomized trial. That means that the researchers followed many people who were taking a PPI and many who weren’t. They then counted how frequently hip fractures occurred in the two groups. They did not assign subjects to either take a PPI or not. They simply observed. A randomized trial, in contrast, would have been one in which each patient is randomly assigned to take either a PPI or a placebo (preferably without the patient or his doctors knowing which he is taking), and then the patients are followed and the number of hip fractures are counted.

Observational studies have very frequently given us misleading conclusions. For a generation we believed that estrogen protected women from heart attacks and strokes until finally a randomized study showed us that the opposite was true. Observational studies also suggested that lowering homocysteine might prevent heart attacks, and that dietary fiber might decrease cancer risk. None of these conclusions withstood randomized trials. Why do these trials give us flawed information? Because individual people are different from each other in myriad ways that can never be fully accounted for by the designers of the study. A generation ago, the women who chose to take estrogen were clearly healthier than those who didn’t, but it wasn’t because of the estrogen. Similarly, the patients on PPIs may be frailer or sicker than those not on them. Only a randomized trial will tell us for sure.

I think observational studies are very useful to help scientists generate ideas for large randomized trials. But patients and doctors are better off ignoring them.

Tangential Miscellany:

The response to the Pedometer Project has been very encouraging. I expect to receive the pedometers in about two weeks, so it’s not too late to sign up or to persuade a loved one to do so. Email me if you want to join. (Click on the link if you didn’t read about the Pedometer Project last week.)