Minutes from melting in the summer heat, I dumped my stuff at a table and homed in on the hospital café’s soda display for something -cold. The gentleman at the next table glanced my way and said, “Are you in the medical field?”

“Yes sir, I’m a medical student.”

He eyed my drink and asked, “Did you hear that diet soda can increase your risk of diabetes by 70%? Even just a few cans a week.”

“No sir, I hadn’t heard that.”

“Seventy percent. Since then I’ve been drinking regular Coke. Wish I’d known ten years ago.”

Sure enough, on his table was a bottle of regular Coke. And I was drinking diet. I’m not trying to lose weight; for me they’re just an occasional treat. I also far prefer the taste of diet, doubt me as you may. I wondered whether the gentleman had diabetes and was now blaming diet soda for causing it. I wondered where he’d heard that 70% figure, and of course I wondered whether that research was actually any good. Since it wasn’t a clinical setting and he wasn’t my patient, I didn’t ask him any of those questions. But the trim, slightly disheveled gentleman in the hospital café certainly got me thinking.

From the day we medical students get our white coats, we are reminded to “do no harm” to our patients. Commonsense as it may sound, we’re about to earn the privilege to use powerful drugs, sophisticated surgery, and deeply personal questions in the service of our patients. We’re taught every day how medical interventions may backfire and cause harm. But in medical education and in society at large, we speak much less often about the potential for information to cause harm. I wonder if I was witnessing one of those cases.

I dug up the study linking diet soda to diabetes. It showed that women who regularly drink sugar-sweetened beverages or artificially-sweetened beverages are more likely to have Type II diabetes. There were more cases of diabetes among the artificially-sweetened beverage drinkers, but it’s nearly impossible to know why. As any student in statistics 101 can attest, and the authors themselves acknowledge, an association (between sweetened beverages and diabetes risk) is not the same thing as causation. Several scientists responded to the paper critiquing its methods. But no scientist, not even the authors of this study, said it means we should switch from diet soda to regular soda. In fact, the paper confirmed decades of research that showed regular soda is plain awful for anyone trying to avoid diabetes.

So what happened? I can’t say for sure, but I suspect a process something like the playground game “telephone.” As the study was translated into a press release, nuances got lost. As that press release turned into a news story, the cycle repeated. And as every viewer watched the news broadcast, a 30-second segment turned into a one-sentence takeaway: “Diet soda causes diabetes.” I imagine many, like the gentleman I met, stopped drinking diet soda and switched to regular — for the sake of their health, ironically. I’m no crusader for diet soda, and I know it may have its own risks, but regular soda is not preventive medicine.

Science is an imperfect, iterative process. One study comes up with something unusual, hundreds respond with different conclusions, and the truth is usually somewhere in the middle. Trouble, and potential harm, arise when initial anomalies are treated as if they overturn what’s been found for decades before. Something similar happened decades ago when Andrew Wakefield published a study claiming to link the MMR vaccine with autism. The study was retracted, Wakefield was disgraced, and decades of credible research subsequently established that vaccines have not led to a single case of autism. Yet since 2007, 1175 people have died from diseases preventable by vaccination. This is perhaps the best example of information causing harm – though not all cases are as stark.

If this gentleman were my patient, and that study had come up in the clinic, I certainly would’ve tried to put it in context: neither regular nor diet Coke is actually “good” for him, and he should switch to water if he wants to lower his risk of diabetes. But a 15-minute annual visit is unlikely to dispel a year’s worth of confusion. The explosion of health information available to the public needs to be matched by a stronger voice from those who can ensure it does no harm. We in medicine need to acknowledge that the clinical encounter alone is not enough to serve our patients — that the need for a medical presence in the public sphere has never been so great.

Karan Chhabra is a medical student who blogs at Project Millenial. He can be reached on Twitter @KRChhabra. This article originally appeared in the Boston Globe’s Short White Coat.