Media coverage of the relationship between coffee consumption and health is pervasive. Unfortunately, it’s also pervasively skewed in the positive direction.
This makes sense. Everyone loves their cup of joe in the morning (and afternoon, possibly evening, if you’re anything like me). The fact that we enjoy coffee so much and consume so much of it lends itself to the fact that we will be receptive to any research, observational or otherwise, that it’s good for us.
(A similar and more detrimental receptiveness to alcohol’s health benefits in the literature is another example. I wrote about that here.)
A quick google search gives you headlines from major news outlets concerning recent links between drinking coffee and health (all from 2018) such as…
Coffee is good for you, more science shows (NBC News)
Coffee drinkers everywhere can rejoice upon reading these articles (or just the headline) and use peer-reviewed literature as a justification for a habit that they can continue to enjoy, and feel good about. Coffee tastes great AND it’s good for us. Win win scenario.
Unfortunately, almost all of the data in which a link between coffee consumption and health outcomes (CVD, all-cause mortality, etc.) comes from large-scale, epidemiological studies. Two of these (below) provide some of the most-recent data on coffee consumption and health.
Association of Coffee Drinking With Mortality by Genetic Variation in Caffeine MetabolismFindings From the UK Biobank (JAMA Internal Medicine)
Findings: Large prospective cohort study of a half million people. Inverse associations for coffee drinking with all-cause mortality for coffee intake from 1 up to 8 or more cups per day.
Conclusion: Coffee drinking can be part of a healthy diet, offering reassurance to coffee drinkers (and heavy ones, at that).
Findings: umbrella analysis of 2018 meta-analyses. Coffee consumption was more often associated with benefit than harm in a variety of health outcomes. Largest risk reduction at 3–4 cups per day vs. none for outcomes such as all-cause mortality, CV mortality, CVD.
Conclusion: Coffee consumption seems generally safe within “usual” levels of intake, and 3–4 cups per day appears to be the intake associated with largest risk reduction
These and other studies have a commonality; they all link coffee consumption with potential health benefits and/or a reduction in disease. All of them are also observational studies.
In other words,they imply association (or correlation), not causation.
A recent perspective piece by Jack James in Journal of Psychopharmacology addresses this point, asking the question (also the title of the article)
“Are coffee’s alleged health protective effects real or artifact?”
Indeed, James states that while meta-analyses and “umbrella” reviews of coffee consumption for the most part all claim “multiple health benefits” associated with moderate coffee consumption — caution must be taken. Authors of a recent large high-quality review state that the observational research provides “only low or very low quality evidence…capable of suggestion association but ‘unable to make causative claims.’”
Furthermore, James states that much of the experimental evidence is in discordance with the epidemiological evidence. What caffeine does to mice and men in the lab doesn’t mirror the proposed mechanisms of action by which coffee consumption might be responsible for reduced blood pressure or type 2 diabetes risk.
Why does this mismatch exist?
James provides a plausible answer, stating that “contradictions concerning the implications of coffee/caffeine consumption for health between observational and experimental research are attributable mostly to poor control over potential confounders in observational studies.”
Health benefits associated with coffee might have nothing to do with coffee at all.
- Experimental studies show that ‘dietary doses’ of caffeine (3–5 mg/kg) acutely increase blood pressure; systolic by 5–15 mmHg and diastolic by 5–10 mmHg.
- Observational reports suggest that coffee/caffeine consumption is associated with lower rate of cardiovascular mortality
While caffeine raises blood pressure and thus could contribute to hypertension, there is some plausible evidence that the chlorogenic acids in coffee (cited as being antioxidant in nature) may protect against CVD. This is likely unwarranted, says James.
Type 2 diabetes
- Early experiments have shown that blood glucose levels are higher during 4-weeks of coffee consumption vs. a similar period of abstinence.
- Caffeine was also shown to disrupt glucose metabolism, or otherwise be “unprotective” against dysregulated glucose metabolism
- Observational studies report that coffee “protects” against type 2 diabetes
Neurodegeneration (Alzheimers and Parkinson’s disease)
- Observational studies suggest that caffeine may possess ‘neuroprotective’ properties an an ability to prevent or delay conditions related to neurodegeneration
- One particular study (Lothian Birth Cohort) found that individuals with higher IQ in childhood had higher IQ in adulthood, and consumed more coffee in adulthood than lower IQ children
The above observation was explained by the fact that “coffee-related superior cognitive ability in adulthood was not due to any protective effect of coffee, but was entirely the result of lifelong cognitive advantage stemming from superior cognitive ability in childhood”
A uniting theme. Confounding variables
As in all epidemiological literature and in particular that cited by James in his article, confounding variables have the ability to create associations where no-such relationships exist — a “spurrious” correlation. Confounders are simply factors (lifestyle, personal, environmental) that influence your depenedent variable (health outcome) but are unrelated to your independent variable (in this case, coffee consumption).
Confounders have a “hidden” effect and if not controlled for, can lead to erroneous relationships (as we have seen). Causal links appear where they simply don’t exist.
What might be the confounder(s) in studies of coffee consumption and health-related outcomes?
The most salient, I believe, is not one factor in particular, but lifestyle in general. People who drink moderate-higher amounts of coffee (amounts shown to be beneficial) live differently than those who don’t. Presumably, high-functioning individuals who are motivated, active, and involved; those may be the individuals who report the highest coffee consumption.
It is safe to assume that high-functioning individuals, also now aware of the health-benefits of coffee (perhaps they read the news often, or come across an article) will likely participate in other habits conducive to health — regular exercise, social relationships, higher-quality diet.
This is all speculation, but a lot of these habits go hand-in-hand. So, while coffee intake probably has a modest physiological contribution to overall health (be it positive or negative), all of the other habits (confounders) play a much larger role in influencing health — independent of how much coffee anyone is drinking.
And so James suggest caution. Not caution in the possibility that modest caffeine consumption might be harmful — as it probably isn’t. However, the evidence available does largely point to acute (short lasting) negative effects of caffeine on things like blood pressure and blood glucose.
Experimental evidence of causation should be expected to take precedence over observational evidence of association. The latter has tended to overshadow the former where research into coffee and health is concerned. This has contributed to the disproportionate confidence that coffee/caffeine is health protective.
As stated in the introduction, the worldwide “addiction” to coffee present a perfect scenario for a self-serving bias where the public is overly sensitive to positive “news” about health benefits of this beverage. While we shouldn’t marginalize coffee for the small but significant physiological effects seen in experimental studies, we should be wary of drawing conclusions about health benefits that just aren’t there.
James concludes, “if the health effects of coffee consumption could be “determined by popular vote, the outcome would be decisive. Alas, popular vote is no guarantee of the true nature of things”
Question everything and for now, keep drinking coffee.