Coffee
I'm old enough to remember when every bank and gas station provided free coffee. It was awful, and you had to drink it from styrofoam cups that melted slightly if the beverage was too hot.
We've come a long way. Starbucks offers more than 50 coffee drinks, in cups that don't melt, including a new product called Oleato, which is coffee infused with olive oil. This is a terrible idea, I suspect, but I'm not sure yet, because for now you can only find Oleato at select stores in New York, Chicago, Seattle, and LA. Meanwhile, cafes in Portland and Austin quickly sold out of their Proud Mary Black Jaguar Geisha coffee, even though it costs $150 per cup. (That's not a typo. $150. 8 ounces of coffee. No wonder Mary is proud.)
Roughly two-thirds of Americans drink coffee – more than any other beverage, including tap water – and I drink enough of it myself that when new coffee research makes the news, I perk up. In 2022, for instance, Australian scientists reported that drinking coffee is associated with a longer life span. Since I'm the kind of person who says things like "I'm old enough to remember...", I was very interested in this study.
My focus here will actually be on an amazing, science-fictiony coffee study published in late March 2023. The Australian findings merit a brief look too, because they help illustrate one of the most fundamental challenges to discerning links between diet and health.
Correlational approaches
If you wanted to know exactly how coffee influences human health, how would you proceed?
One approach would be to follow a large group of individuals over time and statistically tease out connections between coffee consumption and health outcomes. That's what the Australian researchers did. They analyzed data from 449,563 participants in the UK Biobank study who'd been tracked for over a decade. One of the key findings was that both caffeinated and decaffeinated coffee consumption were associated with longer life spans, with maximum benefits emerging at around 2 to 3 cups per day. (I don't drink that much coffee, but after seeing the data I was tempted to step up my game.)
One strength of correlational studies like this is their naturalness. People aren't told how much coffee to drink. They're not told to do anything. They just live their lives and occasionally provide data.
Sadly, the very strength of this approach is also one of its key limitations. In a word, correlational research tends to be poorly controlled. The Australian team compared health outcomes among people who drank varying amounts of coffee, but the researchers couldn't control who drank what or how else they behaved. Lack of control makes the findings ambiguous. Maybe people who drink more coffee are more affluent, for example. (2-3 cups of Starbucks per day would cost, at minimum, nearly $4,000 per year.) Since afffluence and life span are correlated, people who drink 2-3 cups per day may live longer owing to the other benefits of affluence, not to anything coffee does for them.
Correlational researchers can address these ambiguities on the front end (e.g., by only sampling people of similar affluence) or on the back end (e.g., by statistically controlling for differences in affluence). The Australian researchers chose the latter approach, and they did a great job. Still, they couldn't control for all possible confounding variables. Here's an important omission: differences in physical activity were not examined. Why is that important? Because other studies show that greater physical activity is linked to greater coffee consumption and, separately, to better health. Maybe drinking coffee doesn't cause you to live longer. Rather, greater activity increases longevity and, independently of that, more active people happen to drink more coffee.
In the end, although the Australian study was well-designed, I didn't increase my coffee intake, because it's unclear whether doing so would promote a longer, healthier life. (Certainly it would be, for me, a more jittery life.)
Experimental approaches
Given the limitations of correlational research, what might you do instead?
Ideally, you'd gather a large number of genetically identical individuals, and ensure that, beginning at conception, their environments were identical too. The only difference between them would be the addition of coffee to the diets of a randomly chosen group. (You might also create subgroups by carefully varying how much coffee they consume.) At the same time, you'd be monitoring health outcomes throughout their lives.
Experimental studies like this have been carried out – with lab rats. These critters are highly similar, genetically speaking, and they can be raised in controlled conditions and carefully monitored. Obviously studies like that can't be conducted with people. Even if it were possible, there would be two key limitations.
First, experiments tend to be artificial. Results obtained under such conditions may not generalize to the real world, which is messy in part because, unlike the rats, we have some control over what we consume and how we live, and we often change our routines.
A second limitation is that in order to create a tightly-controlled experiment, the experimenter must study a narrow range of coffee-consumption behaviors and thus may not capture all the differences among people in how much coffee we drink, what types we drink, and how our consumption fluctuates over time.
In sum, experimental studies are tightly controlled but may be too artificial or narrow to yield generalizable results. Correlational studies are more closely aligned with our lives but may be too weakly controlled to yield credible data. There are statistical band-aids for these limitations but no perfect solutions.
This brings me to the new study, published on March 23 in the prestigious New England Journal of Medicine. (Coincidentally, this is the same day Oleato was released in the U.S.) What makes this study distinctive is the way it pulls together the strengths of experimental and correlational research, using each to offset the weaknesses of the other.
New study rationale
Dr. Gregory Marcus and colleagues at UCSF, UC Irvine, and UC Berkeley examined the short-term effects of drinking coffee on arrhythmias, activity levels, duration of sleep, and blood glucose. The focus was on caffeinated coffee. (Not to be snarky, but is there really any other kind?)
Arrhythmias are irregular heartbeats (i.e., heartbeats that occur earlier or later than they should). They're perfectly normal – we all experience them on a daily basis – but elevated levels have been linked to heart failure later in life. Prior studies disagree about the role of caffeine consumption. Crudely speaking, some studies suggest that caffeine is generally good for your heart, while others suggest that anything beyond minimal-to-moderate consumption may promote arrhythmias. The researchers sought to address these inconsistencies.
In this study, activity levels were defined in terms of the number of steps taken each day. Prior studies have linked coffee consumption to lower rates of diabetes and longer life spans, but these effects may arise from increased activity. In other words, coffee jazzes you up, you become more active, and your health improves as a result. So, the researchers wanted to better understand whether coffee benefits health directly (e.g., by lowering glucose levels), or indirectly via increased activity.
Finally, duration of sleep was included because few coffee studies have used objective measures to record sleep in naturalistic settings. I'll explain what those objective measures are in a moment.
New study methods
This was not a lab study, yet the researchers did a startlingly rigorous job of collecting data while participants went about their daily business. I had to smile when I read the article, because the methodology is faintly reminiscent of a science-fiction movie. (Nowadays we're anxious about the havoc AI might cause, but when I describe the study methods, you may start worrying about coffee researchers.)
As a participant in this study, you would spend a 14-day period alternating between two days of caffeinated coffee consumption, followed by two days of abstinence from caffeine. Each night and morning, you'd receive texts reminding you of what to do that day. On "coffee days", you'd answer questions about the types and quantities of coffee consumed, while on abstinent days, you'd confirm that you took no caffeine. (The researchers would also obtain data on your health and typical coffee-drinking habits, as well as a DNA sample to determine how rapidly you metabolize caffeine.)
So far so good. Nothing overly intrusive here (although some of us would struggle to go caffeine-free for two days). But there's more. You would've downloaded the Eureka app to your phone so that the researchers could monitor any visits to coffee shops. And, each time you had coffee, you'd push a button.
What button? Well, you'd be wearing an electrocardiogram (ECG) patch on your chest. You'd never remove it, not even while asleep or showering. Each time you drank coffee, you'd push a button on this patch. Doing so would record exactly when you had the coffee, so that questions about coffee consumption could be linked to each specific time you drank some. (Taken together, the survey responses, button-pushing data, and Eureka reports were corroborative and helped ensure an accurate read on coffee intake. Virtually no discrepancies were noted.)
But wait…there's more. For the duration of the study you'd also be wearing an accelerometer on your wrist to measure step counts. This device would include Fitbit INSPIRE technology that records every heartbeat. And, since you'd be wearing this device at night, sleep duration would be calculated. (People don't take steps when they sleep, unless they sleepwalk, but heart rate changes in predictable ways while cycling through sleep stages.)
Last – and far from least – taped to your abdomen for the entire study would be a glucose monitoring device. This device projects a wire into your skin so that blood glucose levels can be continuously recorded. Bluetooth technology ensured that this device, and the ECG patch, fed the researchers real-time data 24/7.
There you go. It's not quite The Matrix, but for two weeks you would be very....monitored. This is clearly a strength of the study. Coffee-drinking and its effects were measured rigorously, in real time, as people went about their daily lives.
Now you can see how experimental and correlational methods were combined in this study. It's experimental in the sense that researchers assigned each participant to a distinct condition for exactly two days at a time. It's correlational in the sense of being carried out in natural settings, with few constraints on participant behavior. Importantly, participants were allowed to choose how much coffee they drank on coffee days, so that natural variability in coffee consumption could be analyzed.
New study findings
1. Sleep.
On average, the 100 participants in this study slept 36 minutes less per night on coffee days than on caffeine-free days. The more coffee people drank on coffee days, the greater the impact on their sleep. DNA testing suggested that sleep was most strongly influenced among those who metabolize caffeine most slowly. In short, (a) coffee consumption reduces sleep a little bit, and (b) people differ in how much their sleep is affected.
Because each participant spent time in both the experimental and control conditions, the researchers were able to control for differences in caffeine metabolism and other key variables. However, interpretive issues still arise. Although it appears that caffeinated coffee slightly reduces sleep, here's another way to look at the data: Participants may have slept a bit longer on caffeine-free days because their bodies were recovering from the two days they'd been drinking coffee freely. If that's the case, the study doesn't yield reliable sleep data, because the control condition was contaminated by the after-effects of the experimental condition.
2. Blood glucose.
Coffee drinking showed no connections to glucose levels. The researchers suggested that although prior studies link coffee consumption to a lowered risk of diabetes, this benefit may not be observable in the short run (and, again, it may be attributable to increased activity rather than coffee per se).
3. Steps.
Participants took 10% to 15% more steps per day on coffee days, which suggests that coffee increased activity levels. This is a good thing. Light to moderate coffee or caffeine consumption has been linked to more positive mood, greater productivity, better cardiovascular health, reductions in body fat, and a longer life. We might add the positive impact on activity levels to this list.
One could quibble a little here and say, well, maybe those extra steps were merely extra trips to the bathroom. (Coffee is a diuretic, so you pee more when you drink it. And, it stimulates colonic and intestinal contractions, so that would inspire even more bathroom visits.)
One could also speculate that on days that a person sleeps less, they take more steps simply because they're awake longer and have more opportunities to walk.
The researchers didn't present their data with enough granularity to evaluate these speculations. To the extent that they hold water, coffee might still be said to increase activity levels, just not directly.
The good news is that the actual increase in number of steps per day was roughly 1,000 on average. Separate, large-scale epidemiological studies have linked that many more steps to significant increases in life span. So, even if the extra steps merely lead to and from the bathroom, your health may still benefit.
4. Arrhythmias.
Drinking one cup of coffee per day had no impact on cardiovascular rhythms, while drinking two or more cups per day was associated with slightly more arrhythmias of one type but not others. Specifically, drinking two or more cups per day led to more premature ventrical contractions. PVCs are extra beats (i.e., contractions) in a chamber of your heart called the ventricle. Everyone experiences them, but they're often not perceived. When you do notice them, you feel what's sometimes described as a fluttering, or even the sense that your heart has skipped a beat.
In this study, PVCs did not increase to anywhere near worrisome levels, but the change was statistically significant and thus consistent with recent evidence that too much caffeine can increase PVCs to the point of slightly elevating the risk of heart failure. In a moment I'll discuss what "too much" could mean.
Conclusions
I like this study. The researchers made a heroic attempt to gather data in natural settings. And, in the end, they didn't try to force the results into a simple "coffee is good for you" or "coffee is bad for you" narrative. Their data suggest that in some respects coffee is beneficial (more daily activity), in some respects it's neutral (no effects on blood sugar and, assuming one cup per day, no effects on cardiovascular rhythms), and in some respects it has the potential to undermine your health (too much may diminish sleep and increase PVCs.)
Earlier I described several interpretive issues, so I would treat the findings cautiously. And, there's an elephant in the room. Does this study tell us something about coffee or about caffeine more broadly? There's no way of knowing, but it's a meaningful distinction. Whereas just over 60% of us are regular coffee drinkers, nearly 90% of Americans take some form of caffeine most days (via tea, energy drinks, etc.).
Whether caffeine or caffeinated coffee drove the results, the take-home message seems more positive than negative, because the "good" effects seem very good, while the "bad" effects weren't very bad per se. Instead, they were merely consistent with other evidence that drinking too much coffee slightly increases the risk of health problems.
How much coffee is too much? That depends on your size, your metabolism, your caffeine intake from other sources, your general health, and so on. The FDA recommends no more than 400 mg of caffeine per day for healthy adults, which corresponds to about four or five 8-ounce cups of medium-strength brew. Reputable sources like the Mayo Clinic agree with the FDA's recommendation. However, since coffees differ quite a lot in caffeine content (a strong brew might contain four times as much caffeine as the same amount of instant), and people differ at lot in size, metabolism, etc., that four-or-five cup advice should be adjusted based on how you feel. If you’re anxious, jittery, and/or not sleeping well, you might try cutting back on the caffeine. (Caffeine can stay in your system for up to roughly 10 hours, so if sleep is an issue, do the math and avoid caffeinating yourself too late in the day.)
Bottom line
Here's where I set aside the nuance.
1. Light to moderate coffee consumption is probably good for you.
2. Excessive coffee consumption is bad for you.
3. Don't put olive oil in your coffee.
4. Don't pay $150 for a cup of coffee.
Thanks for reading!