About the SA Blog Network

The Scicurious Brain

The Scicurious Brain

The Good, Bad, and Weird in Physiology and Neuroscience
The Scicurious Brain Home

Grab your Coffee, I think this paper may depress you.

The views expressed are those of the author and are not necessarily those of Scientific American.

Email   PrintPrint


A few weeks ago I saw this headline floating around the interwebs. COFFEE: NOW WITH ANTIDEPRESSANT EFFECTS!!! I obviously wanted to look into it. I mean, you’re talking about a girl with caffeine tattooed onto her flesh, here, I have a personal stake in this outcome.

Unfortunately, life and lab (ok, mostly lab) got in the way, and I’m only just now getting to this. But I still have a personal stake! After all, I’m sitting here, writing this to you, and sipping my coffee. It’s 8pm. I don’t have a problem…

So. Coffee. So many of us wanted to think it’s a miracle. And to many of us, it is. It drags us through our first days of parents, through our long work hours, through our longER work hours…For many of us it’s become such a salient drug that it’s only the smell of coffee that hauls our butts out of bed in the morning. It’s the socially acceptable attention and alertness enhancer (nicotine is way out of fashion and prescription drugs still leave people squeamy), and in a society where we value productivity so much, how could we not feel for the great brown bean?

And many of us want to justify ourselves. How else do you rationalize your purchase of beans at $13.99/lb and your own personal grinder? What can you use to justify your cravings for coffee at…8pm? Well, it’d be really nice if coffee was actually GOOD for you. You know, more than just the hallmark of the pale, nervous student pulling an allnighter or the exhausted bedraggled parent. If it was good for you, you would feel all the more justified, right?

That’s really the only reason I can think of as to why so many recent studies have come out on whether caffeine is “good” for you or not. Just keeping you awake and alert apparently isn’t enough of a benefit, now it’s gotta cure your cancer. Or in this case, your depression.

But does it cure your depression? Does it prevent your depression? Well, the authors who wrote up this study seem to think so, but I’m not sure I agree with them. Sci the skeptic strikes again. Depressing? But of course!

Lucas et al. “Coffee, caffeine, and risk of depression among women” Archives of Internal Medicine, 2011.

With all the coffee we drink, you’d think there’d be more studies out there on it. But in fact, there’s only been one other study on coffee and depression, which was done in Finnish men, and found a J shaped relationship between coffee and suicide risk. But no studies have been done in WOMEN. And for depression research in particular, this is…depressing. After all, major depressive disorder afflicts roughly TWICE as many women as men. Up to 20% of women in the US will suffer a depressive episode. That’s a LOT of people. But for all that, relatively few studies on antidepressants are done in women or in female animal models (and I’ll save that for another, longer, more ranty type of post). Anyway, kudos to these researchers for studying women.

So what was the study, and who were the women? In this case, it was a prospective study done on a large group of nurses between 1996 and 2006. It’s a hefty sample, with a lot of nurses and pretty good continuity. This is also a good sample because you can make sure that your participants exhibited NO depression symptoms at baseline, and then look at depression symptoms over time to come up with a measure of “relative risk”.

They asked them for what they typically ate, how much they exercised, marital status, smoking, hormone therapy, heck they even asked if you went to church. And of course, they asked about caffeine consumption, including sodas, coffee, decaf, herbal tea, regular tea, caffeine-free sodas, and chocolate. They estimated each “cup” of coffee the women reported as having an average of 137mg (which I’m a bit leery of, I know my “cups” are about 2x the size of other people’s, and I bet many of these women were the same), and then they tallied the amounts per week. Along with this, they looked at depression scores over the ten years, and looked at the number of depressive episodes. From this, they calculated the relative risk of clinical depression as correlated with coffee consumption.

You can see that there’s a significant trend with decreased risk of depressive episode going with higher consumption of coffee. The authors conclude that coffee may have a protective effect against depression.

I will certainly buy that increased coffee consumption (caffeine in general, but strongest for coffee, probably the cumulative dosing may have something to do with it) correlates with decreased risk for depression. But I was more than a little dismayed to see the headlines trumpeting coffee to prevent or TREAT your depression. Um, correlation with decreased risk? Ok. Preventative? …maaaaaybe. Treatment? I don’t think so.

Here are the issues. First, this is a sample composed entirely of nurses. It’s got it’s benefits, and there are many. But it is not exactly unbiased. These are all women in a high stress profession, many are working shift work, active jobs, etc. Things like shift work and high stress jobs are KNOWN to be associated with increased risk of depression. It’s not a problem if you limit your conclusions to nurses, but you can’t extrapolate to all women from this sample.

But the biggest issue is that they found a whole bunch of correlations here. The one they based their CONCLUSIONS on was symptoms of depression. Here are the others.

1) Smoking. The interaction between depression risk, smoking, and coffee consumption was “marginally” significant (p=0.06), but they dismiss it as being due to chance because it was “unexpected”. Um. Wait. Nicotine is a STIMULANT. It is known to have antidepressant like effects in animal models (though the withdrawal is no fun). This is not unexpected.

2) Drinking: heavy coffee drinkers drink more. But note that they don’t say that drinking coffee puts you at risk for drinking alcohol.

3) Obesity: heavy coffee drinkers are, on average, thinner, but not more physically active. They do not conclude that coffee drinking prevents obesity.

4) Church going: heavy coffee drinkers are less likely to go to church. Less likely to go to church, less likely to develop depression…heck, forget depression, maybe coffee prevents religion now! Now THAT would be a heck of a finding.

Here’s the thing. I do believe that high coffee consumption correlates with decreased risk of depression. But a lot of other things do as well. I am not convinced that the high coffee consumption wasn’t part of a lifestyle that correlated with decreased risk of depression, maybe they have stronger support networks or less incidence of depression in the family. It could be many other things. People who consume a lot of coffee may well have something going on neurobiologically which protects against depression (in which case taking up coffee if you don’t like it might not help you). Heck, maybe just staying away from church and having a martini is lifting their spirits! I’m just not convinced that coffee is the real preventative here. And I’m DEFINITELY not convinced that coffee has antidepressant effects.

So, oh most Skeptical Sci, you might say, what WILL prove it to you? Here you go.

1) If you want to prove to me that coffee is having antidepressant effects, I want to see effects in traditional animal models of antidepressant activity. You should be able to do acute and chronic administration either via injection or via the drinking fluid, mice and rats LOVE them some sweetened milk and they’d slurp up milky coffee in a snap. I want to see changes in tail suspension test, forced swim test. I’d like to see chronic effects on neurogenesis if you can get it. I combed through the lit but not a lot has been done on this. And then, I want to see a cohort of depressed people, half given coffee every day, half given decaf, blinded (you can certainly do this, though the IRBs would be awful). I’d want to see depression scores go down. The humans would be the ones I’d really want to see. Considering the common usage of caffeine, you might be able to go straight to humans here.

2) If you want to prove to me that coffee is a PREVENTATIVE, I want to see something similar. This time, I want to see mice/rats given coffee or decaf. I want to see them exposed to stress or other paradigms that create a depression-like model (things like chronic mild stress). I want to see the coffee rats show resistance to the depression-like effects. In humans, I want to see a large study of people over time, controlled for regular coffee vs decaf, and blind to which one they get. I’d like them to be controlled for stressful life events if at all possible (say, a big class of undergrads all taking the same class, though that would have bias of its own). I’d need to see more than nurses. Even then, with the humans, I wouldn’t really be sure.

Yeah, I’m a picky one, aren’t I. But while I’m convinced of the correlation in this study, I’m just not convinced of the causation. The paper didn’t really suggest there WAS causation…but the media certainly did, and for those who don’t really know what “risk” means in scientific studies (and most people don’t), it looks like coffee’s going to prevent your depression, and that’s not what this study said. That said, they certainly didn’t see any correlations with harm! So in the meantime, I’ll keep on sipping.

Lucas M, Mirzaei F, Pan A, Okereke OI, Willett WC, O’Reilly EJ, Koenen K, & Ascherio A (2011). Coffee, caffeine, and risk of depression among women. Archives of internal medicine, 171 (17), 1571-8 PMID: 21949167

Scicurious About the Author: Scicurious is a PhD in Physiology, and is currently a postdoc in biomedical research. She loves the brain. And so should you. Follow on Twitter @Scicurious.

The views expressed are those of the author and are not necessarily those of Scientific American.

Rights & Permissions

Comments 10 Comments

Add Comment
  1. 1. missteresita 12:35 pm 10/25/2011

    I enjoy your writing style and agree with your logic. However, I must be annoying and point out one problem – not all nurses are women just as not all biomedical postdocs are men. Next time you might want to clarify that the study was done on a group of female nurses. Yeah, I’m a picky one, aren’t I? I should probably have another cup of coffee so I’m not so moody. Still, very nice article. I look forward to reading more of your work.

    Link to this
  2. 2. scicurious 12:40 pm 10/25/2011

    Thanks for the correction! I thought since I was focusing on studies of depression in women that it was clear, but I guess not. I’ll change it.

    Link to this
  3. 3. nauvillain 1:51 pm 10/25/2011

    You can never prove causation, strictly speaking. I can drop a ball 100 times and it will fall 100 times, and if I hold it, it does not fall, that does not prove that if I drop it one more time, it will fall again (for all I know, it might have been an alien hidden in a flying saucer, using their telekinetic powers just to toy with me).
    I agree that the media should be more careful and emphasize the correlation and not hint at a causation. But this is the media – they need a good story they can sell, and people like simple things, with a clear conclusion. Correlation hold no conclusion for them.

    Link to this
  4. 4. LanceJZ 4:04 pm 10/25/2011

    I have an issue, decaf is more then just coffee without caffeine. A better way would be to have coffee and pure caffeine. That way the only difference is the coffee.

    Link to this
  5. 5. phenylphenol 5:46 pm 10/25/2011

    Before you snarkily deride the use of the term “marginally significant” you should educate yourself as to its meaning. You also use the term “significant trend” in a way inconsistent with statistical significance, although perhaps that wasn’t your intent. In any case, discovering an unexpected high-order interaction term in the model that shows marginal significance is not unheard of; interpreting it without setting out to test it explicitly in the first place would be dangerous — these authors wisely pass on the temptation to theorize after the fact.

    This appears to have been a longitudinal study, which is a far better approach to conducting this sort of research than cross-sectional studies — it has the force of actually following people through time. For you to suggest that a mouse model is the only way to “prove” something in your eyes suggests that you haven’t thought too heavily about the importance of converging evidence in establishing a proposition, scientific or otherwise.

    If this was meant as a rant against misinterpretations by mainstream media, then that’s fine and dandy, but you shouldn’t commit the same sort of breathless errors of interpretation that they do.

    Link to this
  6. 6. scicurious 7:13 pm 10/25/2011

    I did not intend to deride the idea of marginal significance, which I am aware in human studies represents a larger distance from the mean than the public conception of the term implies. But I think in light of the known effects of nicotine on depressive like tests and symptomatology, it should not have been mentioned without some comment as to the potential interaction. I don’t expect them to conclude on it specifically, but I personally find this to be worth commenting on at more length.

    With regard to the usage of “significant trend”, I believe I am using the phrase correctly, as a measure of a trend difference from the null hypothesis, as currently the null hypothesis would be ‘coffee has no effect on depression risk’. There is a significant effect of coffee amount on risk, expressed as a significant trend varying by amount of coffee consumed.

    I think the study itself is quite good and I’m sorry you interpreted my comments otherwise, but I wanted to emphasize that correlation in these cases cannot prove the causation, and in this case specifically cannot prove the headlines stating that coffee exhibits antidepressant effects. I admit that my estimation of what “proof” is composed of in this case involves a physiological mechanism which mediated the effects of caffeine on depressive symptomatology, which this paper cannot perforce provide. It does not mean that it is a bad paper by any means, it does, however, mean that there is not yet proof of a mechanism by which coffee prevents depressive illness. I by no means wished to be “breathless” in my interpretation, I wished to point out the issues associated with reporting on correlations in the media.

    Link to this
  7. 7. davetrindle 9:04 pm 10/25/2011

    This leap from simple association to “cause and effect” is apparently a hard-wired defect of the human brain. It permeates our so-called science, even in so-called respected, peer-reviewed journals. None of us, not even the scientists can resist taking that one extra–and totally unsupported–step from association to cause. In particular, our medical science is a mess. As a result, what used to be bad for us becomes good for us and vice-versa. For decades women are told to use hormones, now they are a killer. For decades we were told soy was a miracle food, now it is poison. For decades we were told saturated fat and cholesterol were killers, then, “surprise!” they discovered “good” cholesterol and bad. Then they found good-bad cholesterol versus “bad-bad cholesterol (the very low size or density), and now there is a groundswell movement that all that cholesterol “science” was pure bunk. It’s all nonsense. There should be an extra year of medical school where the students simply stand up at the board and write “association is not causation” over and over thousands of time until it finally sinks in. Enough with all the bogus discoveries already. Do some real science.

    Link to this
  8. 8. DeathBreath 6:52 am 10/26/2011

    Yes, the measurement of depression is a bit tricky isn’t it? Over the years of reviewing psychological abstracts, I am still mystified about how affective symptoms are measured. Of course, we have the Beck Depression Scale. Oh, but that is a Depression screening instrument devoid of validity indices common with many standardized objective personality inventories. Then, we have the BPRS (Brief Psychiatric Rating Scale). Did I mention that neither of these so-called instruments have central tendency scores? But, the the BPRS seems to be the gold standard used by many drug companies to measure efficacy. I just shake my head in disgust. I know. If the readers are concerned about gender, perhaps they should consult the brilliant studies of Dr. John Money regarding sex role re-assignment. They are pretty juicy. As far as there being a functional relationship between Major Depressive Disorder and caffeine consumption, I remain skeptical, at best. But, then again, I have not had my morning IV drip of coffee.

    Link to this
  9. 9. figleaf 12:33 pm 10/26/2011

    Based on past experience with customers in a coffee shop (albeit long ago enough that even Starbucks had only one location) if I was constructing a test I’d want to make sure I very carefully screened for people who a) can tolerate caffeine but b) don’t consume it anyway and only use them as the control. There already have to be some pretty significant differences between people who don’t get the jitters, etc., and those who do.


    Link to this
  10. 10. DrKrishnaKumariChalla 11:38 pm 10/26/2011

    Interesting!But I never ever drank coffee or tea in my life! I don’t even know the tastes of these two beverages! I am not kidding!
    But I never get depressed or feel low!It is all in your mind!If you can control it, you don’t need external help!

    Link to this

Add a Comment
You must sign in or register as a member to submit a comment.

More from Scientific American

Email this Article