I think it’s time to start what should be an ongoing series about health reporting in newspapers. I should never run out of material since the papers seem so fond of headlines screaming “X causes Y!!!” or “eat A so you won’t get B!!!!!”

A bit of background — I have a PhD in Epidemiology. I know how studies of health in human populations should be conducted, how they are conducted, and what the biases are in each study type. I think I know the right questions to ask when confronted by reports of causes and cures and preventatives. I also know that the headline is often only vaguely related to the truth for so many reasons.

Shall we begin? Here’s the headline: “Kids in daycare have a lower risk of leukemia risk.”

Let’s start with the last sentence of the article. “Ms. Buffler will present her data today at a childhood leukemia conference in London.” The study wasn’t even presented yet! It hasn’t been published or peer-reviewed (other scientists review articles before they are accepted for publication and most articles submitted to journals are not accepted). So, this article is based on a press release and maybe a press conference.

Next, the study is actually a meta-analysis. What’s that? I’m glad you asked. The researchers compiled the results of 14 studies to give “more confidence that the protective effect is real.” Hmmm. In my experience, meta-analyses are done when the results of the the smaller studies are inconclusive. So … maybe real effect, maybe not.

Lastly, the studies described are case-control studies. These studies compare people with a disease (cases) to people without a disease (controls). This is important because what the study actually found is that kids with ALL were less likely than kids without ALL to have been in daycare.

See the difference?

Kids with ALL have lower history of daycare or playgroups NOT that kids in daycare have lower ALL risk.

That conclusion can only be shown using a cohort study comparing kids in daycare to kids not in daycare. Because the risk of ALL is about 1 in 29,000, a study like this would require a HUGE cohort so it’s not going to happen. The case-control study is the next best substitute but it’s still an estimate.

Oops … I actually have another point. The absolute risk of ALL in children ages 1 – 4 is 7.6 per 100,000 or 0.0076%. That’s a very small risk. If the results of this study are the “truth,” then putting children into daycare reduces a very small risk to a very very small risk.

The headline obviously doesn’t tell the real result of this study. What the study really does is give researchers another clue into what causes ALL. It’s another small piece in a huge puzzle and that’s boring. Boring doesn’t sell newspapers.