Coffee is the most popular beverage in the United States and many other countries (although tea is more widely consumed globally). But the effects of coffee—and caffeine more specifically—have been notoriously challenging to investigate in a clinical trial. Most dedicated coffee drinkers would balk if asked not to drink their favorite beverage for weeks or months on end. That means most research has been what’s called “observational,” with investigators comparing regular coffee drinkers to non-drinkers rather than assigning participants randomly to either drink coffee or not.
Yet observational studies are fraught with potential bias from any unknown and unmeasured differences between the two groups—such as those related to health and lifestyle—that might impact the results. To get around these issues, Greg Marcus, MD, a cardiologist at UC San Francisco who specializes in arrhythmias, conducted a study with colleagues titled “Acute Effects of Coffee Consumption on Health Among Ambulatory Adults.” The study, which included 100 adult participants, was published last year in the New England Journal of Medicine.
Dr. Marcus, who received his medical degree at the George Washington University School of Medicine & Health Sciences, recently spoke with the Wellness Letter about the study and the implications of the findings.
Wellness Letter: Can you explain the context for the study?
Greg Marcus: Coffee has been difficult to study. Few people who drink coffee will agree to give it up for any kind of extended randomized trial, even for a week or two weeks. There are some randomized studies conducted in laboratory settings that have understandably captured the effects of coffee during only a snapshot in time.
So the great majority of research has been observational because that’s what’s most feasible. But one of the problems with observational studies is they are prone to confounding from factors that we aren’t aware of or don’t think about. The only way to fully mitigate against this kind of confounding is to randomly assign people to one group or the other. But then you run into the obvious hurdle—a lot of people don’t want to participate because they say they couldn’t go without their coffee.
WL: So you structured your study to accommodate that?
GM: Right. In our trial, we randomized people on a day-to-day basis. Participants received a text message each day with instructions for the following day—to either go ahead and drink all the caffeinated coffee they wanted or to avoid all caffeine. We guaranteed that they would never be assigned to go more than two consecutive days without coffee. They also wore sensors that continually monitored and recorded their heart rhythm, step counts, hours of sleep, and glucose levels.
Although this study was relatively small, the design provided us with a lot of data and a lot of possible comparisons. Not only were we comparing the participants with each other, but each person could also serve as their own control to compare the effects on days when they did and did not have coffee. That really helped to cut down on the statistical noise and enhance the power to detect differences.
WL: What were you primarily looking for in terms of arrhythmias?
GM: We were primarily looking at common heart rhythm disturbances. First, we looked at the number of premature atrial contractions, or PACs. People with more PACs are at a higher risk of developing atrial fibrillation, which is a very clinically relevant arrhythmia in the upper chambers that can lead to stroke and other problems. We also looked at premature ventricular contractions, or PVCs, which are early beats coming from the lower chambers. People with more PVCs are at higher risk of developing heart failure or weakening of the heart.
We found no difference in PAC counts when people drank coffee versus when they didn’t, and this fits with data from observational studies, which have really failed to demonstrate coffee as a risk factor for atrial fibrillation. However, we saw more PVCs on the days when people consumed coffee. These PVCs are not in themselves dangerous events, but they are possible precursors of a more serious condition, so you want to assess them and then intervene, if needed. We can do a catheter ablation to get rid of those PVCs or provide medication for the arrhythmia.
WL: What were the other findings?
GM: We also were interested to look at the impact of coffee on physical activity, sleep time, and glucose levels. We found that people took about a thousand more steps when they were randomly assigned to have coffee. That’s not surprising, but I think this is the first time it has been shown in a randomized controlled trial. In large epidemiologic studies, a difference of a thousand steps on average per day is associated with greater longevity. So our findings suggest that drinking coffee could have a meaningful impact on actual mortality outcomes.
At the same time, on days randomly assigned coffee, people slept a half hour less. And it’s important to again emphasize that we objectively recorded their sleep, rather than relying only on what people thought about their prior night’s sleep. As with exercise, sleep similarly is very important to overall health, with poor sleep now known to increase risks for cardiovascular disease, mental health problems, and other health issues. Again, I think this is the first time this has been shown in a randomized trial. But the effect of caffeine on sleep differs between people. How fast someone metabolizes caffeine is to some degree genetically determined. As part of the study, we collected DNA from everyone via a spit kit and divided them into slow metabolizers, intermediate metabolizers, and fast metabolizers. Slow metabolizers exhibited almost an hour less sleep on the days they drank coffee, while the fast metabolizers had no apparent sleep loss when they were exposed to caffeine.
WL: You also investigated the risk of diabetes. What did you find?
GM: In large epidemiologic studies, people who drink more coffee seem to experience a lower risk of diabetes, and there’s some question as to whether this is related to enhanced insulin sensitivity. Our hypothesis was that on days randomly assigned to coffee we’d see lower blood glucose on average. However, we saw no differences. That then raises the question of why prior studies showed a lower risk of diabetes. One explanation is that maybe there’s some more chronic effect of coffee. If people are more physically active, perhaps that’s what is leading to less diabetes. Or maybe there are some anti-inflammatory and antioxidant properties of coffee that aren’t captured over the short term but only over a longer period of time.
WL: What’s next for this research?
GM: I would love to look at the more chronic effects of coffee. What happens after two months of consuming versus avoiding coffee? I suspect that some of our findings were somewhat amplified by the on-off design. For example, it may not just be that coffee makes you more active and that you take a thousand more steps a day, but also that when you’re withdrawing from coffee fairly acutely it might have a greater effect on reducing your steps. So our findings may not fully reflect the actual day-to-day responses. A longer study would tell us more about the true effects for people who drink coffee regularly.
WL: So is there an immediate take-home message?
GM: Yes, a couple. One take-home is that the news on coffee is not all good and not all bad. Coffee has multiple effects. I hope that these results provide information that each individual can use and apply to themselves based on their propensities. If someone struggles with insomnia, this should provide really compelling evidence that, yes, coffee really can meaningfully affect your sleep. On the other hand, if someone is trying to get motivated to exercise, then timing their workout shortly after their coffee might make sense.
Regarding arrhythmias, this is good evidence that coffee may truly increase the frequency of PVCs, so if that’s a concern for someone they should try reducing their intake and see if that makes a difference. But if someone’s worried about atrial fibrillation, and many people are because they have a sibling or a parent who has it or because they have a lot of PACs, these data suggest they don’t need to worry about drinking coffee.
