At last, a randomized trial on coffee. Here's what happened.

Hi, everyone! Most of today’s Inside Medicine is free and in front of the paywall. But the bottom section is a Q&A with Dr. Gregory Marcus, the lead author of the coffee study I’m covering today—and that’s for paid subscribers. So, for my non-paying subscribers, I hope you’ll consider upgrading or using a free trial to read it if you’re interested (and get access to other premium essays that I put out a few times per month). Thanks for reading! -JF

---

Journalists love to write stories about the nexus of health and food. Readers gobble up headlines describing studies purporting to show that some food or drink they like is good for them. And they can’t resist stories that purport to show the opposite. Either way, these stories prove to be irresistible clickbait.

Most studies on diet, nutrition, and health outcomes are uninterpretable.

Media stories on food and health tend to have two major problems. First, correlation is not causation, and most of these studies can only establish the former. A classic example is red wine. It’s supposed to be good for you. Maybe it is. But drinking red wine is just as likely to just be something that healthy (and wealthy) people do. Said another way, people are not healthy because they drink red wine. They drink red wine because they’re wealthy—and wealth and health tend to travel together. Second, even if some food or drink is behind some improvement in health, the effect is likely to be very small, noticeable only on a population level. The contorting of statistics to turn small effects into eye-popping numbers is an entire industry.

Once source of the problem is that studies describing the health effects of various consumables are rarely randomized trials with a control group.

Drumroll, please…

Researchers did a randomized trial on coffee consumption!

Coffee is perhaps the perfect subject for a randomized trial on the health benefits of a consumable. The effects are likely to be noticeable in a short period and with modern technology, they’re measurable.

Volunteers signed up for a 14-day experiment where they would alternate between coffee consumption and avoidance. Each day, patients were told whether they should drink coffee or avoid it (but volunteers never had to go more than 2 days in a row without coffee). They wore Zio patches (heart monitors), FitBits (step and sleep monitors), and Dexcom glucose monitors. Aside: 25 out of the 100 volunteers in this study said they drink 2-3 or 4-5 cups of coffee per day. The fact that these people were willing to participate in a study where they had to abstain from coffee for 1-2-day stretches randomly assigned over a 2-week-period is honestly amazing. My hat is off to these weirdos. (I’m in the 2 cup/day zone).

Data: New England Journal of Medicine. Image: Inside Medicine.

Cardiac outcomes: Neither group had any dangerous heart rhythms. But in both groups, there were brief and harmless heart rhythms. Of these, there were more “premature ventricular contractions” (PVCs; the ventricles are the larger heart chambers) on coffee-drinking days, in particular among people who had more than 1 cup.

What this study does not tell us is whether people with high coffee consumption have more frequent harmless heart rhythms which could predispose them to going into a more dangerous rhythm which in turn leads to heart failure down the road. So there’s a chain of risk here. It would be interesting to see how this would play out among coffee “super users” compared to occasional coffee drinkers or abstainers.

Steps: Step counts were higher on coffee days, to the tune of >1,000 more (10,600 vs 9,600). That’s about 1/2 a mile difference per day.

Sleep: Sleep was around 36 minutes shorter on coffee days.

Glucose levels: No difference.

Genetics: The researchers also did genetic testing on the volunteers, looking to see if people who had genes that alter their coffee metabolism had different outcomes. The short answer is that appeared to matter some. Those with genes implying faster caffeine metabolism had more PVCs. Those with genes associated with slower caffeine metabolism got even less sleep on coffee days than those without those genes. Note: This part of the study was really impressive to me. So many studies have described genes like this, but it’s rare to get a readout on any “real life” implications like we got here.

What I’m doing with this information.

Let’s be honest: People like me would not have signed up for this study. After all, volunteers had to be willing to not drink coffee for 2 days in a row, and as the kids say, ain’t nobody got time for that. Still, what impressed me here was that the effects were noticeable on a short time frame. One could abstain from coffee and expect more sleep in the next day or two. One could drink coffee and expect it to be easier to be a bit more active.

That said, the odds of me stopping drinking coffee every day are currently forecasted at 0.0%.

So, what I’m going to do is this: Because the effects of caffeine consumption do seem quite brisk (i.e., on the order of hours, not days), I’m going to move my last caffeine consumption to earlier in the day, or at least move to half-decaf on days where I need to get to sleep earlier and for longer.

Ending caffeine consumption early is hardly new advice, and there’s debate as to when you should do this each day. And I suppose based on this new study, the best answer depends on your genetics (which is unknowable for most, including me). But somehow, knowing that this actually matters on a same-day basis has me inspired to try to make a little change here.

Meet the scientist.

The lead author of the New England Journal of Medicine study is Dr. Gregory Marcus, a cardiologist at the University of California, San Francisco. I emailed Dr. Marcus with a few questions about the results and the possibility of future insights.

Here’s what he said—including a kind offer to answer further questions that the Inside Medicine community may have—so please share yours! (This “interview” was assembled/collated from a few emails, and was edited for readability and flow.)