Coffee and the Heart: Is caffeine a trigger for AFib?

[Regina] (0:00 - 0:10)
Pope Clement VIII supposedly tried coffee and declared it so delicious that it would be a sin to let only non-Christians drink it.

[Kristin] (0:11 - 0:16)
Oh, all right. If it's Pope-sanctioned, then we have to drink it. It's like our moral obligation.

Great.

[Regina] (0:17 - 0:18)
That's what we're going to tell ourselves.

[Kristin] (0:23 - 0:32)
Welcome to Normal Curves. This is a podcast for anyone who wants to learn about scientific studies and the statistics behind them. I'm Kristin Sainani.

I'm a professor at Stanford University.

[Regina] (0:33 - 0:38)
And I'm Regina Nuzzo. I'm a professor at Gallaudet University and part-time lecturer at Stanford.

[Kristin] (0:38 - 0:43)
We are not medical doctors. We are PhDs. So nothing in this podcast should be construed as medical advice.

[Regina] (0:44 - 0:56)
Also, this podcast is separate from our day jobs at Stanford and Gallaudet University. Kristin, today we're talking about something that people seem to like even more than sex.

[Kristin]
More than sex.

Wow.

[Regina]
Coffee.

[Kristin] (0:57 - 1:15)
Oh, yes, I can see that. Coffee is important. I love my coffee.

And in fact, Regina, I just bought a new coffeemaker, and I think that's the highlight of my day. I did briefly give up coffee after my breast cancer, but I then decided that there was actually no evidence to support that sacrifice. So I have gone back to coffee.

[Regina] (1:16 - 1:19)
You're back with your coffee boyfriend. I approve.

[Kristin] (1:19 - 1:31)
He is a good boyfriend. And in fact, I think boys are just more trouble than coffee. But, Regina, of course, all I want out of this episode now is for you to tell me that coffee is good for me.

Evidence be damned.

[Regina] (1:32 - 2:05)
I've got integrity. Hello. I actually don't handle caffeine all that well, so I'm not a huge coffee drinker.

But Kristin, we are sticking to the evidence whether we like it or not.

[Kristin]
Fine. Okay.

[Regina]
Today, coffee. There is a lot of research out there about coffee, but we are just going to focus on two clinical trials that look at whether coffee is bad for atrial fibrillation, a heart condition, and it's sometimes called AFib.

[Kristin] (2:06 - 2:15)
Ooh, clinical trials and coffee, two of my favorite things. But Regina, AFib, I sort of know what that is, but not really. Can you fill us in?

[Regina] (2:15 - 2:29)
Well, for now, I'll just say AFib is a type of arrhythmia where the heart goes a bit haywire. I actually have a great example from The Pitt, the medical drama, that will illustrate it a little more.

[Kristin] (2:30 - 2:33)
Oh, The Pitt. Everyone is watching that. Yes.

That'll be fun.

[Regina] (2:33 - 3:06)
It is fun. It is. It's a fun show.

But what does AFib have to do with coffee, you might be asking yourself. For years, doctors have told patients to avoid coffee because it can rev up the heart and, they said, potentially trigger AFib. And there were these two recent clinical trials that tried to get at whether that's actually true.

And that's the claim that we'll look at today, that caffeinated coffee is a trigger for AFib.

[Kristin] (3:06 - 3:13)
I can imagine this is important to a lot of people because, I mean, how do you function without your coffee?

[Regina] (3:13 - 3:37)
Exactly. I understand that. So we are going to walk through these two trials because they're both interesting in slightly different ways that you'll see.

And we're also going to talk about survival analysis and Cox proportional hazard regression and hazard ratios and intention to treat versus as-treated analyses.

[Kristin] (3:37 - 3:44)
Hmm. Regina, this is actually exactly what I've been teaching currently in my class at Stanford. So I guess I can send my students to this episode.

[Regina] (3:45 - 4:35)
Ooh. Perfect timing. Great idea.

And actually, speaking of perfect timing, I did not choose this episode about coffee just because we have now put on our site, normalcurves.com, a buy us a coffee widget. It just happens to be a fortunate chance that we managed to figure out how to get this on our website. And people can buy us a coffee.

They can donate a little money and keep us caffeinated. And we actually got our first friend donating coffee to us not long ago. They had a great comment.

They said that they are enjoying the podcast immensely and recommending it to all their students. So anonymous person, whoever it is, thank you. Thank you for the coffee.

[Kristin] (4:35 - 4:35)
Thank you.

[Regina] (4:36 - 4:42)
Okay. So let's now talk about AFib, and then we're going to have some fun with caffeine and coffee.

[Kristin] (4:42 - 4:43)
Ooh, okay.

[Regina] (4:43 - 4:46)
All right. So The Pitt, have you seen it?

[Kristin] (4:46 - 5:05)
Okay, Regina, I watched some of it, but to be very honest, I thought it was kind of slow and boring. I'll admit that my guilty pleasure is medical dramas, but the ones with lots of boys and sex like Grey's Anatomy. And I think this one disappoints on that front.

Not enough cute boys.

[Regina] (5:06 - 5:52)
Yeah, I think The Pitt is going for, let's say, realistic and earnest, very earnest, and not really fantasy with cute guys and scrubs. I'm just going to say that. But there is a scene.

I don't know if you saw it, or maybe you were sleeping through it, in episode three that actually gives a really nice illustration of AFib. And just a background for people who have not seen it, The Pitt is set in the emergency department at the University of Pitttsburgh, and Noah Wiley plays the grizzled attending physician. His name is Dr. Robby, who, by the way, Noah Wiley also played John Carter, the medical student on ER back in the 90s.

[Kristin] (5:53 - 6:08)
Oh, I just made that connection, Regina. Yeah, he looks really, really different now, but you're right. He was really young back then.

And that is quite a throwback. I watched that when I was an undergraduate, which let's just say was a long time ago. George Clooney, right?

[Regina] (6:08 - 6:40)
Those were the days. Okay, we're going to fast forward now, 2020 is at The Pitt. And a patient comes in with heart palpitations that have been going on for a while.

And his heart rate, he says, has been elevated for over an hour. And it's an AFib episode, and he's clearly uncomfortable. And Dr. Robby tells him, we're going to shock your heart back into a normal rhythm. And the patient asks, well, will it hurt? And the doctors just say to him, you'll be sedated.

[Kristin] (6:41 - 6:42)
Well, sedation is good.

[Regina] (6:43 - 7:26)
Yeah, right. And then the guy goes, oh, I see. So it's really going to hurt.

So that shocking thing is called cardioversion. And I saw a cardiologist describe this on social media, that if you did do cardioversion while someone was fully awake, it would feel like getting kicked in the chest by a horse. Ouch.

Yeah. Okay, so they sedate him, they shock him. It works.

And later we hear he's back into normal sinus rhythm. And then the show moves on with its usual ER chaos, which, spoiler alert, involves drug overdoses and a man with rats in his shirt.

[Kristin] (7:26 - 7:33)
Yeah, see, this is where, get rid of the man with rats in his shirt and just give us the cute boys, right? Then it would work for me as a show.

[Regina] (7:34 - 8:06)
That's a whole different axis they've got going on there. Okay, I like this because it apparently showed accurately what is happening with an AFib episode. And you can look at what's actually going on with the heart.

So let me unpack it. AFib is when there's chaotic electrical activity in the atria, and that's the upper chambers of the heart. And instead of contracting normally and moving blood along like it's supposed to, the atria kind of just quiver.

[Kristin] (8:06 - 8:12)
Oh, so like the heart is still beating, but it's not in the normal coordinated way. It's kind of like uncoordinated.

[Regina] (8:13 - 8:33)
Exactly. Perfect. And people feel that.

And it can feel like different things, like a shortness of breath or palpitations or like pounding or racing or fluttering in the chest or the throat. And fun fact, Kristin, about one in three people will have some AFib at some point in their lifetime.

[Kristin] (8:34 - 8:36)
Oh, wow. That is surprisingly common.

[Regina] (8:36 - 8:56)
All right. Now, for some people, it just lasts seconds or minutes, but for others, it can go on for a while. It sends you to the ER, but for others, it can go on for days.

And the whole danger with this is that blood can pool in the atria and form clots, and those clots can lead to things like strokes.

[Kristin] (8:57 - 8:59)
Oh, OK. So that's the part that makes this serious.

[Regina] (8:59 - 9:10)
Right. And that shocking that they're doing, that cardioversion, they put sock paddles on your chest and give a low-level shock to interrupt that chaotic rhythm and reset it.

[Kristin] (9:10 - 9:19)
Are those the same paddles where in the medical dramas they yell, clear, and then everybody backs away and they shock them? Is that same paddles?

[Regina] (9:19 - 9:46)
Yep. Same idea. But in that case, the heart has stopped.

That's different. With AFib, the heart is still beating. It's just disorganized, like you said.

And it's not always this dramatic, though, I want to point out, and we're going to be talking about AFib patients. Some people don't even feel anything. And for some of them, it's picked up only through electrocardiograms, ECGs.

And these days, smartwatches can sometimes pick it up.

[Kristin] (9:46 - 9:50)
Oh, OK. So some people might have AFib and never know it.

[Regina] (9:50 - 10:27)
I know, right? So what puts you at risk for all of this? Well, getting older, of course, like everything else, plus things like high blood pressure, obesity, sleep apnea, diabetes.

But there are also triggers. Sometimes they call it holiday heart, which I found kind of adorable. Like you've had a little too much to drink on vacation or put a little too much salt on your nachos and margarita.

And for years, like I said, doctors have told patients with AFib to avoid caffeinated coffee because they thought that was also one of the triggers.

[Kristin] (10:28 - 10:34)
Well, that makes sense because coffee can make you jittery. So maybe it makes your heart jittery, too. That sounds plausible, actually.

[Regina] (10:34 - 10:44)
Right. It does, doesn't it? OK, but what is the evidence?

We shall see. All right. Let's talk about coffee.

Now, this is the good stuff. Caffeine is kind of amazing, right?

[Kristin] (10:44 - 10:47)
It is amazing. Yes. Hallelujah for caffeine.

[Regina] (10:48 - 11:24)
I don't even love caffeine myself because I'm one of these slow caffeine metabolizers and it just keeps me up forever. But even I do think caffeine is amazing. And we've been drinking caffeine for thousands of years.

I'm going to give you all the fun facts that I learned about caffeine and coffee because it's cool. For example, there's a story about coffee being discovered in Ethiopia about 2,800 years ago when a goat herder noticed that his goats were getting a little more energetic and frisky after eating coffee berries.

[Kristin] (11:24 - 11:27)
Oh, I've never heard that story. That's fascinating, Regina.

[Regina] (11:28 - 11:54)
You can just picture the goats, like, dancing around. So it turned out that lots of plants make caffeine. So in coffee, in tea, in cacao beans, guarana berries, which is what's in energy drinks, and then there's kola nuts, which is the original ingredient in Coca-Cola.

Plants actually make caffeine because it's toxic to insects. It's a natural pesticide.

[Kristin] (11:54 - 11:58)
Okay, so you're telling me that caffeine is a toxin? Great.

[Regina] (11:59 - 12:26)
To insects, not humans, I need to point out. But it is super bitter. And okay, another fun story.

From the 1600s, some European clergy called coffee, quote, the bitter invention of Satan until Pope Clement VIII supposedly tried coffee and declared it so delicious that it would be a sin to let only non-Christians drink it.

[Kristin] (12:26 - 12:32)
Oh, all right. If it's Pope-sanctioned, then we have to drink it. It's like our moral obligation.

Great.

[Regina] (12:32 - 12:59)
That's what we're going to tell ourselves. All right. Quick science here.

What is caffeine actually doing in your brain? Because I didn't know any of this. It is blocking adenosine receptors.

And adenosine builds up the longer you're awake, and that is what makes you feel sleepy and tired and want to take a nap. So caffeine basically blocks that signal so you don't feel tired, even though you really are.

[Kristin] (13:00 - 13:17)
Oh, that is interesting. And, you know, Regina, actually, there's a lot of great data on caffeine in the sports world because it does have an effect on athletic performance. According to meta-analyses, it improves performance by something like two to four percent.

And I think we might need to do an episode on that sometime, Regina.

[Regina] (13:18 - 13:49)
Oh, absolutely. Another fun fact, Kristin, I did not know this, but it makes sense. Starbucks coffee actually has way more caffeine per ounce than many other coffees.

Why? Because of the beans they use and how they brew it. And just as an example, Starbucks Grande 18-ounce blonde roast has about 360 milligrams of caffeine.

But if you get a 16-ounce McDonald's coffee, it's only about 145 milligrams.

[Kristin] (13:49 - 14:20)
This explains a lot about why people are willing to pay so much for Starbucks coffee. 360 milligrams for one coffee, though, Regina, I feel like this episode needs to come with a warning because it is actually possible to have too much caffeine. I know someone who ended up vomiting and in the emergency room with what could have been a lethal dose of caffeine because she accidentally drank concentrated coffee without realizing it was concentrated and that you're supposed to dilute it.

And it was actually really dangerous. So you can overdose.

[Regina] (14:20 - 14:40)
Oh, yeah, that is scary. I read that you can have serious problems if you ingest 1000 milligrams of caffeine too quickly. That's when problems can happen.

And the FDA recommends no more than 400 milligrams per day. So it's like that's one Starbucks blonde roast and you're done.

[Kristin] (14:41 - 14:43)
And that's not even their largest size. Wow.

[Regina] (14:43 - 14:59)
OK, Kristin, I think now we are ready to talk about the first clinical trial. And this one is a randomized controlled trial called CRAVE, C-R-A-V-E, which stands for Coffee and Real-Time Atrial Ventricular Ectopy.

[Kristin] (15:00 - 15:15)
OK, so they worked really hard to get that acronym. And you know, Regina, I hate acronyms for everything except for study names, because it's actually useful to be able to call the study something. And people often give it memorable names like this.

[Regina] (15:15 - 15:28)
Yeah, it's not bad. I actually like CRAVE. So this study was published in the New England Journal of Medicine in 2023. And it was from a team of cardiology researchers at UCSF.

[Kristin] (15:29 - 15:31)
New England Journal of Medicine, top journal. Great.

[Regina] (15:31 - 15:40)
I know. And it's a fun paper. And Kristin, it has been a while since we touched PICOT and SMART from our holiday guide.

[Kristin] (15:40 - 15:41)
Oh, that was a great episode.

[Regina] (15:42 - 15:53)
Right. So I was thinking, how about you walk us through PICOT to show how it would be applied here. And I'm thinking maybe you could channel your inner grumpy Uncle Joe a bit, maybe.

Right.

[Kristin] (15:53 - 15:58)
So PICOT, the P stands for Population, which means who was in the study?

[Regina] (15:58 - 16:12)
Okay. 100 people in San Francisco average age 39 years old, about half women. And get this, if they had a history of heart troubles, they could not be part of the trial.

This was just healthy people.

[Kristin] (16:12 - 16:25)
Oh, well, that is really important because then the results only apply to healthy people, not to AFib patients. And I assume that who we really care about is actually the AFib patients, not the healthy people. So why did they look at healthy people here?

[Regina] (16:26 - 16:46)
I think the idea was that before they started experimenting on actual people with actual serious heart conditions, maybe they should just look at healthy people first. We want to make sure we don't kill the heart patients, you know, by telling them to go wild at the Starbucks. But don't worry, we are going to see AFib patients in the next trial.

[Kristin] (16:46 - 16:49)
Okay. That makes a lot of sense. That's a safety ethics issue here.

Yeah.

[Regina] (16:51 - 16:53)
That was P, population. What's next?

[Kristin] (16:53 - 17:08)
Next is I for intervention. And after that is C for control. So you said, Regina, this was a randomized trial.

So I assume that people were randomly assigned to either an intervention or a control. And I'm guessing that the intervention perhaps was coffee.

[Regina] (17:08 - 17:23)
It's like your mind reader, Krista. Yes, exactly. They were randomized to drinking caffeinated coffee or abstaining from all caffeine.

And it was, like you said, randomized controlled trial, but it was not blinded.

[Kristin] (17:24 - 17:36)
It would be hard to blind participants because you can probably tell if you've had caffeinated coffee. Also, I think if they were going to use a placebo, you'd have to like go to the lab every day to get your coffee. Maybe that wouldn't be convenient.

[Regina] (17:37 - 17:39)
Right. Right. Exactly.

[Kristin] (17:39 - 17:46)
Okay, so next in PICOT-O, which is for outcome, what was the thing that they were measuring at the end of the day? I assume something about the heart here.

[Regina] (17:46 - 18:13)
The primary outcome was the number of a special kind of arrhythmia called premature atrial contractions. And it was the number of those in each 24 hour period. Now this is not AFib.

This is a type of arrhythmia where there are like extra beats that start in the upper chambers of the heart and it feels like an early beat and then a pause and then a stronger beat. But these arrhythmias are usually harmless.

[Kristin] (18:13 - 18:18)
Wait, so if they're harmless, why was this the outcome of this study? Why not AFib?

[Regina] (18:18 - 18:28)
Mm-hmm. Good question. Because having lots of these premature atrial contractions is often linked with later having AFib.

Oh, I see.

[Kristin] (18:29 - 18:39)
So they're looking at an AFib precursor, which makes sense because these are healthy people. They're probably not having a lot of AFib. So we need something to measure that's the proxy for AFib or a precursor to AFib.

[Regina] (18:39 - 19:02)
Exactly. And they had a bunch of secondary outcomes. Three other kinds of arrhythmias, all of which are rarer and some are more serious than the primary outcome.

We won't go into details. Plus three lifestyle measures, blood sugar levels, step counts, and sleep. And then they also looked at all of these across different caffeine metabolism genotypes.

[Kristin] (19:03 - 19:12)
Wow. So that is six secondary outcomes I'm counting, plus the interactions of those with genotype. That's a lot of statistical tests.

Did they adjust for multiple testing?

[Regina] (19:13 - 19:22)
They did not, but they were upfront about that, Kristin. And it's kind of interesting how they reported things and why. And we'll see that in the results section later.

[Kristin] (19:22 - 19:33)
All right. The good thing is that they had one primary outcome that was clearly spelled out. So we just will consider everything else to be exploratory.

And Regina, how did they measure this premature atrial contraction?

[Regina] (19:34 - 19:53)
They all had an ECG patch, an electrocardiogram patch that they wore that continuously measured their heart rhythms. And for the lifestyle measures, they all also had a Fitbit, you know, for step count and sleep and a continuous glucose monitor for blood sugar.

[Kristin] (19:54 - 19:58)
So they're walking around like a little science experiment for the duration of this trial.

[Regina] (19:58 - 20:13)
They are. Plus, they had to turn on the geolocation feature on their phone. And that's going to be a little teaser, again, for when we get to the results, because it was fun how they used that.

Okay. That was O. So what's next?

[Kristin] (20:13 - 20:21)
T for time. You haven't told us how long this study was that they have to potentially go without coffee and wear a bunch of equipment. Right.

[Regina] (20:21 - 20:24)
Right. Two weeks. It was just a two-week trial.

[Kristin] (20:24 - 20:27)
Okay. That makes sense logistically. Yeah.

Very short.

[Regina] (20:27 - 20:46)
Very short. Mm-hmm. Short and sweet.

Our next trial will be longer. Now, Kristin, for this one, they did something interesting with the randomization. They randomized each of these 100 people to a caffeine group or a no-caffeine group every single day.

They randomized them.

[Kristin] (20:47 - 20:51)
Wait. So there was a new randomization done for every person every day? Wow.

[Regina] (20:52 - 21:21)
Every day. Exactly. So how would it work?

You'd get a text every night saying what tomorrow would be for you. So if it was a caffeine on day for you, you were told to drink as many coffee drinks as you like, especially caffeinated coffee, it would say. So it was a fun day.

And if it was a caffeine off day, it told you to avoid all caffeine, including decaf coffee, chocolate, and tea. So that was a horrible day.

[Kristin] (21:22 - 22:02)
So no caffeine at all. Ooh, that's hard. So, Regina, I want to just point out a few technical points here.

This design is what we call a case crossover trial, meaning that each person serves as their own control, and they basically get to cross over from one group to the next in random order. Typically, in a case crossover, though, you only cross over once. But this trial is a little different because it employs something called micro-randomization, which means you keep getting randomized over and over again every day.

And, you know, Regina, I think, isn't this the example that we use to explain micro-randomization in our clinical trials course for Stanford Online? I have a vague memory of this.

[Regina] (22:02 - 22:04)
It absolutely is. Good memory.

[Kristin] (22:04 - 22:19)
So I'm going to point out, if anyone wants to learn more about this study or clinical trials, you can find a link on our website, normalcurves.com, to all of our courses, including our wonderful clinical trials course. But, Regina, why did they do it this way? Why make it complicated like this?

[Regina] (22:20 - 22:33)
Good question. Well, they said in the paper, one, that they didn't want to have cumulative effects of caffeine, and two, they wanted to, quote, enhance enrollment and retention.

[Kristin] (22:34 - 23:24)
Ah, yes, I see the problem. If you tell people to go without coffee for two weeks, that might be too heavy a lift for people. They may not want to sign up.

But if you say you're going to get coffee some days, people might be willing to do it. That actually makes a lot of sense. Mm-hmm.

Regina, I want to hear about the results now, but first, let's take a short break.

Welcome back to Normal Curves. Today we're talking about the effects of caffeinated coffee on the heart, and we were about to get the results of this 2023 CRAVE trial.

[Regina] (23:24 - 23:44)
Kristin, one of my favorite parts in that holiday episode where we introduced Peacock was when you introduced your own brilliant mnemonic, SMART. And I really think this is brilliant. So how about you lead us through how we can use SMART here to look at the results?

[Kristin] (23:44 - 24:03)
I named it SMART because it helps you feel smart. So let's start with S. S stands for signal.

Does the signal exceed the noise? And we usually evaluate this with statistical significance. So Regina, was the primary outcome of that AFib precursor, was there a statistically significant difference on coffee days versus non-coffee days?

[Regina] (24:04 - 24:13)
Nope. Not statistically significant, or as I sometimes like to call it, not statistically discernible. Nope.

P-value of 0.1. Hmm.

[Kristin] (24:13 - 24:25)
Okay. So not significant, but that's just the P-value. And we always want to consider the magnitude, which is the next part of the SMART mnemonic.

M is for magnitude. So how big of an effect was it, even if it wasn't significant?

[Regina] (24:25 - 24:40)
Right. So on their caffeinated days, people had an average of 58 of these premature atrial contractions. And on their non-caffeinated days, it was 53, which is a 9% difference.

Not huge, but not nothing either.

[Kristin] (24:40 - 25:12)
Oh, that's interesting, Regina. So you're saying this is something that happens to all of us all of the time then. 53, 58 times a day is a lot.

Regina, I want to note that there is some difference here, and we have to be really careful when we see that something is not statistically significant. That doesn't mean that we have evidence that coffee is safe. This is a subtle distinction, but all we can say from the results is that we don't have evidence that coffee is harmful, that it causes these arrhythmias.

Right. So did the authors interpret this correctly, or did they try to claim that this proved that coffee is safe for the heart?

[Regina] (25:13 - 25:20)
You will be happy to hear they were actually very careful in their conclusions. They did not try to say that this is proof of anything.

[Kristin] (25:21 - 25:23)
What about those secondary outcomes, Regina?

[Regina] (25:24 - 26:34)
They didn't actually give p-values for some of those secondary outcomes. But they did that for a good reason, not just because they were being strange. The New England Journal of Medicine would not allow them to.

And this is because back in 2019, the journal changed its statistical reporting guidelines to de-emphasize p-values and emphasize confidence intervals more. So here is what they specifically say to authors in their guidelines. If you had more than one secondary outcome, but didn't specify in your statistical plan how you would correct for multiple testing, then you must report only the effect size and the 95% confidence intervals, no p-values.

And they said, you cannot use these intervals to make any definitive conclusions. And you've got to essentially put a little warning label for readers, too. They said, you must tell readers that the confidence intervals were not adjusted for multiple testing and that they can't use the confidence intervals in place of hypothesis testing.

[Kristin] (26:34 - 26:54)
Oh, very interesting. And this is related to the whole debate around p-values, which we actually address in our p-value episode. So I highly recommend people go there to hear that episode.

Whether you like p-values or not, I think one good thing out of this policy is it keeps the focus on the primary outcome so people don't start making stories out of the secondary outcomes.

[Regina] (26:54 - 27:08)
I remember when this came out in 2019, I was working at the American Statistical Association and it was a big deal. We were all very proud. This came out of a lot of the work that the ASA had done in 2016 to come up with a p-value statement.

[Kristin] (27:08 - 27:24)
So this is interesting because it forces researchers to say, hey, if you want to be able to make claims about multiple outcomes, then you have to plan ahead and you have to plan to do multiple testing adjustments. And that's good. It prevents a lot of statistical shenanigans.

So what did they find?

[Regina] (27:25 - 27:48)
Well, I'm just going to focus on the lifestyle outcomes here because they were the most interesting and that was blood sugar, steps, and sleep. So blood sugar, there was not much of a difference on coffee days. It was an average of 95 milligrams per deciliter versus 96 on the caffeine-free days.

And that one milligram becomes like a rounding error, right? You can't make a headline out of that.

[Kristin] (27:49 - 27:52)
Yes, basically the same, yeah. So drinking coffee doesn't seem to spike blood sugar.

[Regina] (27:53 - 28:14)
Now, step count, Kristin, was more interesting. They found that on coffee days, people walked about 1,000 more steps on average. It was about 10,700 steps on coffee days versus 9,700 on caffeine-free days.

And I calculated that's a difference of about half a mile.

[Kristin] (28:15 - 28:34)
I mean, I can certainly believe that people would walk more. I think I would probably walk less if I didn't have my coffee in the morning. Also, Regina, 10,000 steps a day, that's high.

This is a pretty select population then. They're clearly not unhealthy, sedentary people. All right.

And the last one you said it was sleep. What did they find on sleep then?

[Regina] (28:34 - 28:49)
Okay. On coffee days, they slept about 36 fewer minutes per night on average. It was about six hours and 37 minutes on coffee nights versus seven hours and 12 minutes on caffeine-free nights.

[Kristin] (28:50 - 29:16)
Oh, wow. So they walked more, but they slept less. And I don't want to hear this.

I already don't get enough sleep, so don't tell me that my new coffee maker is the problem. 36 minutes is not trivial. Although I'm going to refer people back to our last episode on exercise, sleep, and aging, where actually the optimal sleep, according to their crazy analysis, was 6.5 hours a night. So I guess we're all fine, even with our coffee.

[Regina] (29:16 - 29:28)
Oh, this is true. But getting back to your coffee maker, Kristin, I feel like your coffee maker boyfriend is doing what all good boyfriends do, which is keep you up at night. That's one way to look at it, right?

[Kristin] (29:28 - 29:36)
That is one way to look at it, yes. I feel like the boyfriend keeping up at night might be more fun than the coffee keeping up at night. Don't you think?

[Regina] (29:37 - 29:41)
Yeah, but again, in the absence of a boyfriend, we take whatever we can get.

[Kristin] (29:41 - 29:43)
Me tossing and turning, not as fun as sex.

[Regina] (29:46 - 29:54)
I guess. Well, when you put it that way. Okay, that was M, moving along.

Now, what was A, Kristin?

[Kristin] (29:54 - 30:07)
Okay, A is for alternative explanations for the results. And this is a really important one that people forget, right? You have to think, could the results be explained by something else, like biases or confounding?

And do the authors think about that here, Regina?

[Regina] (30:07 - 30:27)
Right, well, biases and confounders, we get to worry about this a little less because this is a randomized trial. But this does get us to adherence. And this is what I hinted at before.

Yes, people were assigned some days to avoid all caffeine, but did they sneak some anyway?

[Kristin] (30:27 - 30:47)
Ah, yes. So I can imagine that people did sneak some and they might have trouble giving up their caffeine. So maybe that's why we're not seeing a significant difference on coffee days for that primary outcome, because maybe people cheated and everybody was just having coffee all the time.

I mean, it was only 14 days, but still giving up coffee for even one day can be hard.

[Regina] (30:47 - 30:57)
Right, exactly. So it is kind of cool how the researchers devised clever ways to gauge how much coffee sneaking was going on.

[Kristin] (30:59 - 31:09)
I love in clinical trials, Regina, when they come up with ways to figure out if people are complying, because compliance is so important and there are often clever ways to get at it.

[Regina] (31:09 - 31:33)
And I love this. It is like psychology and medicine and statistics all rolled into one. And this is a perfect example of what they did.

Okay, well, first of all, they had participants click their little ECG patch every time they had coffee. So that meant they had a timestamp in their data of the day and the time of each cup.

[Kristin] (31:33 - 31:47)
Oh, I love that. So they were measuring compliance, but also maybe that helped people to comply more, right? Because if they know they have to click the little button, they're going to feel guilty if they drink the coffee and don't click it.

So it might keep them from drinking the coffee, maybe?

[Regina] (31:47 - 32:02)
You have a better outlook on humanity than I do, Kristin, surprisingly, because I just assumed they would cheat. They just wouldn't click it. I think people would click it.

I do. Oh, I love that about you. I love that you say you would be honest.

[Kristin] (32:02 - 32:14)
Well, if you're part of a study and if you're like, I absolutely have to have my coffee, but you would still feel, I think, obligated so that the data weren't contaminated, that you would let people know, I think.

[Regina] (32:14 - 32:38)
Again, I love it. But the researchers maybe weren't quite as optimistic as you because they came up with a backup plan, which is even cooler. What they did is reimbursed people for every single cup of coffee they drank, as long as they had a time-stamped receipt, regardless of whether it was supposed to be a coffee day for them or not.

[Kristin] (32:39 - 33:05)
Oh, that is so clever. And, you know, Regina, that's probably how they got people to join this study at all. So let's say you had seven coffee days in the study, two cups of coffee per day.

I mean, Starbucks is like, what, seven bucks a coffee? That adds up. It's probably almost like a hundred bucks.

And this is also great because people then had a financial incentive to tell the truth. That seven bucks might get you to say, yeah, oops, I did have a coffee that day.

[Regina] (33:06 - 33:32)
Right. This is incentive. And it worked.

It looks like they caught about 20 cheaters that way. They also had another sneaky approach. At the start of the study, the researchers asked participants where they normally got their coffee fix every day.

And then during the study, they used their phone's geolocation to flag whenever the participant went near their favorite spots.

[Kristin] (33:33 - 33:36)
Oh, that is clever. Wow. Big brother.

[Regina] (33:37 - 33:38)
They were spying on them, right?

[Kristin] (33:38 - 33:39)
Oh, wow.

[Regina] (33:40 - 34:05)
They got about nine cheaters that way. Yeah. So overall, it looks like at least half of the people admitted to cheating at least one day out of the two weeks.

So it actually looks like you were right. People were generally honest about being dishonest. OK, so what did they do with all of this compliant data?

They ran what's known as an as-treated analysis.

[Kristin] (34:06 - 34:45)
OK, right. So Regina, let's back up. Because the earlier analysis that you talked about was what we call an intention-to-treat analysis.

And this is typically how we analyze data in clinical trials. It means that we analyze everyone according to how they were randomized. Once randomized, always analyzed.

And we ignore what they actually did in real life. And this is really important because it preserves all the benefits of randomization, that the groups are similar to one another. Sometimes, though, researchers will do a secondary analysis where they look at what people actually did.

And that is called an as-treated analysis. And here, that would mean we would count people as being in a coffee day on the days that they actually drank coffee, even if they weren't supposed to.

[Regina] (34:46 - 35:05)
Excellent description. I love that. So here's the nice thing.

When they did that, even when they analyzed people in this as-treated analysis, it looks like there was still not more of these premature atrial contractions on coffee days. Nothing really changed. All the secondary outcomes were similar, too.

[Kristin] (35:05 - 35:13)
Well, that is very reassuring. We always like it when our results line up. No matter exactly how you analyze it, that suggests that they're robust.

Exactly.

[Regina] (35:14 - 35:16)
OK, that was A. Now, R.

[Kristin] (35:17 - 35:38)
Right, R was for reality check. This is our gut check. Do the results pass the smell test?

Are there any red flags or errors that undermine the study? A perfect example of not passing the smell test is the last episode we did where the authors claimed that all of a sudden, everybody in America started sleeping more in 2015. That did not pass the reality check.

How about this study?

[Regina] (35:39 - 35:59)
I love these little elements because they really force you to think through these things. It's so easy to skip over. So reality check, I did not find any red flags in the data.

All the numbers seem plausible. The authors did a nice job. So I am going to say that it passes the smell test.

[Kristin] (35:59 - 36:01)
The yummy coffee smell test.

[Regina] (36:02 - 36:05)
Yep. OK, so that leads us to T.

[Kristin] (36:05 - 36:18)
Right. T is for trustworthiness or transparency. And this is asking, were they honest and transparent?

Did the authors give us everything we'd want to see or did they seem to be secretive? When we have transparency, it gives us more trust in their work.

[Regina] (36:19 - 36:35)
Exactly. So here, they registered everything on clinicaltrials.gov, and they were transparent about all the changes they made to the protocol, even little ones. And I also like here for trust, they did not try to oversell the results.

[Kristin] (36:35 - 37:01)
Yeah, I like that. We love to see that. The study generally seems well done.

They seem to be conservative in their conclusions and everything was pre-registered. So all gives us trust in the data. All right, just to summarize, though, it sounds like they didn't find anything here, although we can't rule out an effect.

Regina, that was in healthy people, though. And you mentioned that there's another study about people who actually were suffering from AFib. Can we hear about that study now?

[Regina] (37:01 - 37:10)
Of course. On to the second trial, also a randomized, controlled, non-blinded study. And it was by the same group of researchers, actually.

[Kristin] (37:11 - 37:13)
Oh, that's good because this seems like a pretty good research group.

[Regina] (37:13 - 37:24)
Yeah, yeah. Okay, this one was published in JAMA in 2026, very recently. And here, they actually looked at heart patients.

And this study is called DECAF.

[Kristin] (37:24 - 37:27)
I can't wait to hear how they made DECAF fit.

[Regina] (37:28 - 37:36)
Yeah, I think this one actually works. They didn't have to contort themselves too much. It stands for, Does Eliminating Coffee Avoid Fibrillation?

[Kristin] (37:36 - 38:24)
Oh, yeah, they didn't have to contort too much. That actually works. And that's a great name.

So that's two great names that this research group has come up with. All right, before we get to talking in more detail about the DECAF study, let's take a short break first.

Welcome back to Normal Curves.

Today, we're talking about coffee in the heart. And we were about to hear about the DECAF trial. So who were the participants in this trial, Regina?

You mentioned that they were actually heart patients, not healthy people.

[Regina] (38:24 - 39:01)
Yep, heart patients this time. There were 200 adults. Average age now is 69.

And about 70% of them were men. And they were all regular coffee drinkers or had been recently. And this is the part you asked about, Kristin.

They all had a history of persistent AFib or a related arrhythmia that's called atrial flutter. That's very similar. And in fact, to qualify for this trial, their AFib had to be so persistently bad that they were scheduled to get that cardioversion shocking thing that Dr. Ravi did.

[Kristin] (39:02 - 39:11)
Oh, wow. Okay, so this is a really different set of participants than the other trial. And that cardioversion thing, that is where they shock your heart to get it back to that normal sinus rhythm?

[Regina] (39:11 - 39:23)
Right, exactly. And everyone, people like this, was randomized after that cardioversion just to make sure everyone started the trial with a good baseline, a normal heart rhythm.

[Kristin] (39:23 - 39:31)
So everybody starts with a heart that's doing well. And then did they do the same micro-randomization where people got coffee some days and not others?

[Regina] (39:32 - 39:44)
No, they changed it up this time, actually. The whole study was six months and they randomized them only once at the beginning of the study to be in the caffeinated coffee group or the caffeine abstaining group.

[Kristin] (39:44 - 39:53)
Wow, so they needed to stay in their randomization group for the entire six months. So some people had to give up coffee, tea, and chocolate for six months?

[Regina] (39:53 - 40:18)
No, right? In fact, this caused problems for the researchers because the study, Kristin, was originally supposed to be for a full year, but they could not recruit enough people. So they quickly changed it to just six months, but that still was not enough.

So they had to branch out from just the U.S. and add a site in Australia and then another one in Canada.

[Kristin] (40:19 - 40:46)
Oh, wow. Yeah, I mean, I can believe that this would be hard to recruit for, although weren't these people's doctors telling them to give up coffee anyway? Isn't there some incentive for your health to join the trial?

But still, yeah, if you're already a coffee drinker, really hard sell. So how did they check adherence? So now I'm worried about cheating a lot more in this one, right?

Because how do we know that people actually stayed off the caffeine for that long? Did they reimburse people for coffee again?

[Regina] (40:47 - 40:49)
They would have gone broke doing that.

[Kristin] (40:49 - 40:56)
Oh, yeah. Well, it would have been a good recruitment tool to get people in the study, but then if you got randomized to the control and you didn't get your free coffee, you'd probably be pretty mad.

[Regina] (40:57 - 41:14)
Missed out on that benefit, yes. No, this time they just checked in with the participants after one month and after three months. And at those check-ins, they reminded their participants of which group they were in and asked them how many cups of coffee they had been drinking recently.

[Kristin] (41:14 - 41:29)
Oh, OK. So they just asked them about compliance. None of these clever tricks.

And this decaf trial, then it's really looking at more the long-term effects of coffee, right? Because maybe giving up coffee for a day wouldn't be enough to, like, help your heart or hurt your heart. All right.

What outcomes did they look at here?

[Regina] (41:30 - 41:53)
This time they looked at whether or not participants had a new episode of AFib or this atrial flutter that lasted more than 30 seconds. And they used medical records to confirm this. And if participants did have a recurrence, then the researchers noted when they did have it and that was it.

That was the end of their participation in the study.

[Kristin] (41:53 - 42:14)
Oh, so this sounds like a time-to-event outcome. And Regina, we've talked about time-to-event outcomes before in this podcast. That means we're following people over time until they have the event of interest here in AFib.

And that also means that they're going to be analyzing their data with survival analysis. And I bet we're going to get some Cox proportional hazards regression models coming up, right?

[Regina] (42:14 - 42:34)
Exactly. You are, again, a mind reader. So that was the primary outcome.

And the secondary outcomes were just three this time. They looked at AFib episodes independently, atrial flutter episodes independently, and then adverse events, like did people have a heart attack or have a stroke or die?

[Kristin] (42:35 - 42:39)
I like how they had one primary outcome and then just a few secondary outcomes. That's very disciplined.

[Regina] (42:40 - 43:21)
I think it helps them get clean results too. So let's talk about those results now. Like you said, they used survival analysis.

And Kristin, back in our exercise and cancer episode, you introduced the idea of survival analysis and Cox regression with a great game show analogy, but we didn't spend much time explaining the hazard ratio, which is a measure that comes out of the Cox regression. So I thought now we could do a statistical detour on hazard ratios. And I came up with a game show analogy.

I was inspired by yours that you might hate actually.

[Kristin] (43:22 - 43:33)
Oh no, I'm never going to hate a game show analogy, Regina. And yes, the hazard ratio is what we get out of these Cox regression models, but we've never really broken down before on this podcast exactly what that is.

[Regina] (43:33 - 43:58)
Right. Okay. This is how I'm picturing survival analysis and Cox regression and hazard ratios.

It's like we've got an American Ninja Warrior style game show competition. We've got two teams where everyone runs along their own balance beam and below them is a water tank full of hungry sharks. And once they fall in, they're immediately eaten.

[Kristin] (43:59 - 44:04)
Oh, this is a very exciting game show actually. It's like a relay race. Isn't it?

[Regina] (44:05 - 44:40)
They're all starting at the same time and they're just, so we have lots of balance beams and they're falling in and it's very exciting because now there's shark bait and there's lots of thrashing about and we want to see which team is better at surviving, which one has fewer people falling off and getting eaten. And so for the two teams, here's what I pictured. One team all in high heels and the other, they're wearing those giant cushioned clown running shoes.

You know the ones I'm talking about?

[Kristin] (44:40 - 44:47)
Like the hokas? Yeah. I don't see how anyone can run in those.

Definitely not on a balance beam and I think I'd prefer the high heels team for this one, Regina.

[Regina] (44:48 - 45:35)
You would be excellent in high heels. I've seen you in high heels. So I'm going to be at home rooting for you out there on team high heels.

So the key to remember here is that researchers are like TV commentators. They don't just fast forward to the end of the show and then count who's left standing. That would be boring.

No, it's like they are looking at the unfolding drama frame by frame and they're talking about which team is more likely to lose someone to the shark tank next. And that gets us, Kristin, to one of the most confusing things about survival analysis, the hazard ratio. And the hazard ratio is kind of like a moment by moment comparison for the teams.

[Kristin] (45:35 - 45:41)
Oh, I like that analogy, Regina. The hazard ratio is kind of this, we're comparing the teams but moment by moment.

[Regina] (45:41 - 46:17)
Exactly. So I'm imagining we're doing a freeze frame at every point, right, in the show. And we're asking, OK, between a random uneaten person on team high heels and a random uneaten person on team clown shoes, who's going into the shark tank next?

And let's say that at any given moment, it's less likely to be team high heels and more likely to be team clown shoes. And let's say, just to use some numbers, it's one to three odds.

[Kristin] (46:18 - 46:22)
So the hazard ratio here then would be one to three or one third, Regina?

[Regina] (46:23 - 46:55)
Right, that is the hazard ratio. It's basically the hazard that a random person from the high heels team is facing falling into the shark tank at any moment in the study versus a random person from clown shoes team. And if the hazard ratio is below one, like it is here, it means the high heels team is winning.

They have lower hazard and therefore better balance beam skills. And if the hazard ratio is above one, it means they're losing, they have greater hazard and they are being eaten up more quickly.

[Kristin] (46:55 - 47:05)
That's an excellent analogy and a really good visual picture of the hazard ratio. All right, so Regina, what did they get in this study for their hazard ratio, comparing the coffee and non-coffee groups?

[Regina] (47:06 - 47:21)
The primary outcome here was statistically significant, statistically discernible, but, Kristin, not in the direction you might think. The coffee group had fewer of these AFib episodes than the no caffeine group.

[Kristin] (47:21 - 47:28)
Wow, wait a minute. So the coffee group is like the high heels group? Coffee was protective and you're saying it's good to have coffee, yay!

[Regina] (47:29 - 47:32)
That, team coffee is winning the day, right?

[Kristin] (47:32 - 47:34)
And what were the numbers, Regina?

[Regina] (47:34 - 47:44)
Right, this is where our hazard ratio comes in. So the hazard ratio was about 0.6 for the coffee group compared to the non-caffeinated group.

[Kristin] (47:44 - 47:52)
P-value of 0.01. Okay, so Regina, that's about a 40% reduced hazard for the coffee group, which is actually a pretty decent effect size.

[Regina] (47:52 - 48:24)
Right, not bad. Now, Kristin, just like we talked about before, we can bring that back and translate it for humans in a human unit here. So that hazard ratio of 0.6, that means the odds are 6 to 10, which is 3 to 5. So it just means out of every eight times, three times, it's going to be the random coffee person going into the shark tank first. And five out of eight times, it's going to be the non-caffeinated person going in first.

[Kristin] (48:26 - 48:34)
That's a very nice visual, Regina. And that's great. That's looking good for coffee.

So I'm getting eaten by sharks less if I drink my coffee, yay!

[Regina] (48:34 - 48:44)
Exactly, exactly. And the secondary outcomes also supported this. It was significant for AFib and there were no big adverse events for the coffee group.

[Kristin] (48:45 - 48:53)
OK, so but what about like alternative explanations and hearings? Did they do an as-treated analysis in this one? Because I'm thinking people might have cheated.

[Regina] (48:53 - 49:38)
They might have cheated. That is exactly what they did. So the non-caffeinated group surprised me, I've got to say.

So at baseline, before they started the study, people in that group were typically drinking one cup a day. And some, though, were getting as much as four or five cups a day. Then they were randomized to this non-caffeinated group.

One month later at check-in, most had complied. About 80% said they had managed to get down to zero. That surprised me.

Except, Kristin, you've got to love the one person who promised not to drink caffeine for six months, but then after one month, they came back and they were telling the researchers that they were averaging three cups a day.

[Kristin] (49:38 - 49:42)
Well, at least they were honest. I mean, yeah, I didn't expect everybody to comply in this trial, Regina.

[Regina] (49:43 - 49:49)
I love the way you spin it, you know, positively, though, that you like they were honest about being dishonest.

[Kristin] (49:49 - 49:53)
Yeah, honesty is a high-value commodity in my world, Regina, actually. Yep, it is.

[Regina] (49:54 - 49:56)
You have good reason to value it highly.

[Kristin] (49:56 - 49:57)
I have good reasons, yeah.

[Regina] (49:58 - 50:22)
So the researchers did run that as-treated analysis, just like before. So they were analyzing people in the group. They were actually participating in.

So all of those cheaters in the non-caffeinated group got put in the coffee group. And when they did that, now the hazard ratio went down to 0.53. So the effect looks even stronger and we're still seeing the coffee being protective.

[Kristin] (50:22 - 50:24)
So that gives us some confidence in the results.

[Regina] (50:25 - 50:28)
Team coffee is definitely winning here.

[Kristin] (50:28 - 50:37)
I've got to ask, though, Regina, like biologically, people expected, you know, it's going to make your heart jittery. And so why would coffee be protective? Do they have any biological theories?

[Regina] (50:38 - 51:15)
They do, actually. So the author said, yes, it is true that caffeinated coffee revved up the heart, which seems bad, yes. But at the same time, it might also protect the heart in even bigger ways.

And they gave a few different ways this might happen. They said, for example, it blocks adenosine. And adenosine can mess with the heart's electrical signals and lead to AFib.

The caffeine also calms down inflammation. That's a risk factor for AFib. And they pointed out caffeine might help people get in a little bit more physical activity.

[Kristin] (51:16 - 51:20)
Oh, they might take more steps. Like the healthy people took more steps. That makes sense.

[Regina] (51:20 - 51:27)
Exactly. And lastly, I found this interesting. Caffeinated coffee can make you pee more and reduce your blood pressure.

[Kristin] (51:27 - 51:32)
Oh, well, it does make you pee a lot. That's true. So I didn't know that was good for you, though.

I thought that was just annoying.

[Regina] (51:33 - 52:05)
I'm not sure about that one. But the whole thing is plausible, which is why it's not all as crazy as it first sounds. There is some biological plausibility.

Now, there is a mixed report card, Kristin, on some of the other details. On the one hand, the researchers did a great job with their pre-registration on clinicaltrials.gov. Detailed all their changes to the protocol, like having to change the planned study length, for example.

[Kristin] (52:05 - 52:13)
That's good. It shows, you know, clinical trials sometimes don't go the way you planned, but at least they were transparent about it. What was the mixed part on their report card, though?

[Regina] (52:13 - 52:33)
One table in the supplementary information didn't match up with a figure in the published paper. The supplementary table said that the heaviest coffee drinkers were getting six or more daily cups at baseline, which is a lot. But then the figure showed that the maximum was only about 4.4 cups a day.

[Kristin] (52:35 - 52:39)
Okay, so we're getting an inconsistency, which always makes us a little bit worried about data integrity.

[Regina] (52:40 - 52:48)
Yeah. I even got our boyfriend, graph2table, to give me the exact numbers from the figure just to make sure I wasn't eyeballing it wrong.

[Kristin] (52:48 - 53:15)
Yeah, that's not great. But Regina, I will say, I'm going to forgive them a little bit because you said this was in a supplementary table. And I will admit, sometimes researchers just dump everything into supplements, and we like that transparency.

And they might not always be quite as careful as when they're preparing the tables and figures for the main paper. So it's a little bit easier to understand why errors might sneak in. Maybe people didn't double check.

So I would say this is probably not a fatal flaw.

[Regina] (53:16 - 53:29)
I agree. And all of the other results looked reasonable. This was the only glaring error that I saw.

And they were so transparent about their other analyses. So I still have good trust in the research team in general.

[Kristin] (53:29 - 53:57)
Okay, so just to sum up, this decaf trial found that not only did drinking caffeinated coffee not contribute to more AFib episodes, it actually seemed to protect against them for people with a history of AFib. And this was on top of the CRAVE trial, which found that coffee did not significantly increase those AFib precursors in healthy people. So maybe looking pretty good for coffee.

So Regina, I think now we are ready to rate the strength of evidence for the claim today. And what was the claim today?

[Regina] (53:57 - 54:01)
Caffeinated coffee is a trigger for atrial fibrillation.

[Kristin] (54:01 - 54:16)
And how do we rate the strength of evidence on this podcast? It's with our trademarked, highly scientific one-to-five smooch rating scale, where one smooch means little to no evidence for the claim, and five smooches means strong evidence for the claim. So, Regina, kiss it or diss it.

[Regina] (54:16 - 54:52)
Yeah, so notice I framed my claim to be what is the conventional wisdom around coffee and AFib, because that's what we always hear. That's what the doctors are saying. So we are rating that claim, not the papers.

It's opposite what the paper found. So I'm going to give the conventional wisdom claim only one smooch. And then I'm just going to reallocate the four other smooches that I could have used.

And I'm going to give that to the studies and the authors, except I'm going to turn them into coffee. Nice, steaming, hot cups of delicious coffee.

[Kristin] (54:52 - 55:14)
Yeah, we definitely liked these studies, but in terms of the claim, which was the conventional wisdom, I'm also just going to go one smooch, Regina. I'm not saying that we can completely rule out that coffee is a trigger for AFib, but if we're looking just at these two pretty well-done randomized trials, neither of them provides evidence for that. So we're lacking evidence.

And that's a one smooch for me.

[Regina] (55:15 - 55:18)
Yep, okay. What about methodological morals? Do you have something good?

[Kristin] (55:18 - 55:30)
I think this is a great example of when conventional wisdom isn't always right. So my methodologic moral is never trust conventional wisdom until you see the randomized controlled trial.

[Regina] (55:30 - 55:41)
Oh, I love that one. You know, this applies to so much, not just about coffee and AFib. We need more randomized controlled trials out there.

[Kristin] (55:41 - 55:42)
Absolutely. How about you, Regina?

[Regina] (55:43 - 55:53)
I'm going to go with the honesty thing. So here's mine. Trust your participants, but design the study so that they can be honest about their dishonesty.

[Kristin] (55:53 - 56:07)
I love that recognition of human nature. It's very important to factor in human nature when you're designing a study. Yes.

Well, Regina, this has been fascinating and not just because I love coffee so much and I could talk about coffee all day, but I learned a lot in this episode. So thank you.

[Regina] (56:07 - 56:20)
Yes. And you know, it really brought me around to maybe some of the benefits of coffee. And I actually had a cup of real caffeinated Starbucks coffee this morning just for you.

[Kristin] (56:21 - 56:33)
Just so you could last through the taping of this podcast, which went well into the evening for East Coast time, your time. Regina, you know, I'm going to take away that it's fine for me to keep having my coffee. And next time you're in town, I'm going to make you a coffee with my new coffee maker.

[Regina] (56:34 - 56:35)
All right.

[Kristin] (56:35 - 56:36)
So thank you, Regina.

[Regina] (56:37 - 56:44)
Thanks, Kristin. And thanks everyone for listening. Bye.