*This transcript has been edited for clarity. *

Most of us need coffee to start the day because unlike my dog, Tobi, we don’t have the luxury of sleeping for 16 hours a day.

Now, some people don’t drink coffee because they are subversive individuals who are trying to undermine democracy. But about 90% of the planet does and they live to tell the tale. Still, there is this pervasive fear that coffee is bad for us, especially from the cardiovascular point of view. And we have been researching the health effects of coffee for centuries.

In what is probably the first randomized trial in history, King Gustav III of Sweden took twins who had been condemned to death and gave them a choice: execution or become the subjects of the world’s first twin study. One twin would drink three pots of coffee per day, and the other would drink tea for the rest of their lives. First twin to die loses.

The tea-drinking twin died first, at 83 years of age, and the coffee-drinking twin sometime later, although his age was not recorded. So clearly, coffee was not the public health treat that Gustav feared.

Now there are a few interesting points in this story. First, Sweden and Switzerland are not in fact the same country (as some of you probably believed). Second, the experiment continued long after Gustav III was assassinated in 1792, which is a testament to bureaucratic steadfastness. True, the nitpickers will say things like:

• Why would the study continue after the king’s assassination, given that his son was then ousted in a coup d’état?
• How likely is it that two twins would both be imprisoned for capital offences at the same time?
• What human being would drink three pots of coffee without becoming sick of it and begging for death after a few months?

While we can never know the true historical veracity of this story, it seems clear that even early on, people’s fears about coffee — the world’s most socially acceptable drug and stimulant — have not really been bolstered by hard data.

Admittedly, most of the data about coffee is observational in nature. Outside of an autocratic monarchy, it is very hard to compel people to drink coffee consistently on a daily basis. I, for example, cannot order someone to drink three cups of coffee per day for the rest of their lives. I don’t have that kind of power. I will one day, but not today.

But there are some data that we can review.

The first is a very interesting case-crossover study. If you are not familiar with a case-crossover study, it goes something like this. Like a case-control study, it works backward. You identify individuals who have had a heart attack, and these are your cases. However, the same people are going to serve as their own controls. So basically, you are a case when you had a heart attack, and you serve as your own control during when you didn’t. Researchers can pick whatever the control period is going to be.

This type of study design works well for outcomes that are quick and repeatable: car accidents, flu virus infections, or heart attacks. If heart attacks happen when your coffee consumption goes up, well, then — you have your correlation.

Now, this type of study design has some key advantages. Because you serve as your own control, you eliminate a lot of confounding. Not all, but a lot. However, it still requires you to measure coffee consumption accurately and the devil is always in the details. But this is a study type that can be very a useful tool if you know what you are doing — like a nail gun. But it would be easy to misuse in inexperienced hands — like a nail gun.

Anyway, this study by Baylin and colleagues found a relative risk of 1.49 for myocardial infarction in the first hour after drinking a cup of coffee. A 50% increase in myocardial infarction is pretty bad, obviously, and if true it would probably mean that we are all going to die imminently. But digging into the data reveals some nuance.

An accompanying editorial by Charles Poole did something interesting. First off, it’s pretty funny and worth reading. But he basically works out when the absolute risk increase is for that 1-hour period of time. There may not be enough hours in the day, but there are 24 of them.

Now, depending on which cardiovascular risk calculator you want to use— and there seems to be a new one every few years or so — we know that average cardiovascular risk for most people is spread out over a long period of time. Let’s arbitrarily say that your risk for myocardial infarction is 10% over 10 years (which I know is harkening back to the old Framingham cutoffs, but whatever, it’s just an explanatory example). In that 10-year period, assuming the risk is spread out evenly over that decade, you have 24 hours per day. Take those 25 hours and multiply it by 365 days per year, and multiply it again by 10 years. That’s 87,600 hours. So, your hourly risk for myocardial infarction is 0.00000114. Increase that by 50% and you get 0.00000171. That’s a risk difference of…look, there’s a lot zeros there, okay? The risk, on an hourly basis, is quite small. To put it in context, the number needed to harm (ie, 1 over that very small number) is 1.7 million cups of coffee.

You need to drink a lot of coffee to provoke a heart attack.

Of course, there are lots of caveats here, and we cannot completely eliminate confounding from studies like this. But even at its most extreme, coffee is not a major public health threat.

What about arrythmias? Heart attacks are hard to study because the lead time for what is essentially a chronic disease is so long. But arrhythmias should be easier. And here we have actual randomized controlled trials that definitely happened and aren’t urban legends about long dead Swedish kings.

The CRAVE study was published in 2023 and was a really cool example of how to do a randomized controlled study in food, because remember: Autocracy is bad, and force-feeding people specific foods against their will is not consistent with a free and just society.

To get around their inability to bend research subjects their will, researchers randomized people to coffee vs no coffee in a 1-day-on, 1-day-off type of arrangement. This makes sense, as it’s probably easier to go without coffee for a day rather than for months at a time. To make sure people stuck to their assigned coffee drinking days, the participants got daily text reminders; their coffee intake was collected daily; and they were reimbursed for all the coffee they bought, which encouraged them to keep receipts. Also, their phone’s GPS trackers recorded when they went into coffee shops. The researchers basically did as much as they could to make sure people stuck to the coffee or no-coffee schedule. Plus, the participants were wearing portable cardiac monitors to measure premature atrial contractions (PACs) and premature ventricular contractions (PVCs).

Most people think coffee is a stimulant, and it is. They also think it should increase arrhythmias, but fun fact: It doesn’t.

In CRAVE, they observed no increase in atrial fibrillation. There was no increase in PACs, the atrial ectopy that tends to trigger atrial fibrillation. The only increase was in PVCs, which increased from 102 to 154. And if you’re reading a Holter, that’s a pretty negligible increase. Remember, you have around 80,000 heart beats per day, depending on your heart rate. So, an extra 50 PVCs is not clinically significant.

You probably also heard about the DECAF study, which was just published. (In cardiology, it’s mandatory that all studies have clever acronyms. I wrote an article for Medscape about this a few years back.) DECAF was a bit different. These were patients with atrial fibrillation that were going for cardioversion and who already drank coffee. They were randomized to continuing with daily coffee drinking or abstinence.

There are a few caveats to the study. Only about 69% of people in the abstinence group completely gave up coffee, only about 50% of the people had a continuous monitor, and of the 1965 people screened, 1739 did not meet eligibility criteria — mostly because they didn’t want to change their coffee drinking habits. So, this was a randomized but open-label trial where a very large proportion of patients who were screened were not randomized. Bear that in mind.

The researchers found that there was less atrial fibrillation in the coffee group. Remember, these people were all cardioverted, so in the coffee group, 47% flipped back into atrial fibrillation vs 64% in the abstinence group over 6 months. Fun fact, though: Most of the difference was seen in the group who were getting clinic electrocardiograms. In the continuous monitor group, the rates of atrial fibrillation were the same. Also, hospitalizations were similar in both groups.

So, what does this mean?

1. It could all be random statistical noise.
2. There could have been some subtle selection bias from the screening process.
3. It’s possible the participants in the coffee abstinence group were more aware of palpitations, prompting them to get clinic electrocardiograms, leading to greater detection of atrial fibrillation. Remember, on continuous monitoring, there was no difference.
4. Maybe coffee is protective.
5. Maybe if the study had lasted a year or more, everyone would have reverted to atrial fibrillation and there would have been no difference.

Either way, it’s clear that coffee is not dangerous. Whether it protects against atrial fibrillation is, I think, still an open question. But with CRAVE and DECAF, I think it’s clear that it doesn’t cause it.

So, you can safely drink your morning coffee. Just try to limit yourself to 1.7 million cups.

For Medscape, I’m Dr Christopher Labos…with Tobi.