STATS ARTICLES 2012
Your Life on Coffee
Rebecca Goldin, Ph.D., Cindy S. Merrick, May 21, 2012
Clearing up the coffee confusion
Last week, The Atlantic reported that “Coffee Really Does Make You Live Longer;” and yet, “people who drink coffee tend to die sooner than non-coffee drinkers.” Apparently realizing that this message could be difficult to process, the writer sums it up with “we don't know for certain coffee is having a cause and effect.”
The cause of this acute attack of “on the one handism” was a study conducted by Neal Freedman, et al, at the National Institutes of Health (NIH). Data were gathered in the NIH-AARP Diet and Health study among AARP members aged 50 to 71, and statistics were generated regarding various mortality causes vis a vis coffee consumption. The health data were all self-reported. Over 200,000 men and about 173,000 women participated in the coffee consumption study. Those already diagnosed with cancer or heart disease, or who had a history of stroke, were excluded.
After adjusting for confounders, including smoking, coffee had a slight inverse relationship with mortality – in other words, drinking more coffee was correlated with less death among the participants. Increased coffee consumption was also associated with lower rates of diabetes, heart disease, respiratory disease, and stroke. On the other hand, coffee-drinkers were more likely to be smokers. They typically had a lower education level, and were less likely to engage in vigorous activity. They also ate fewer fruits and vegetables.
All data were self-reported and subject to recall error and other individual biases. For the moment let us assume that the associations in this large sample reflect real effects. Two significant considerations reported by the authors (but not by The Atlantic) affect the conclusions we can draw. For one, there exist plausible explanations for the opposite causal relationship. Those with poor health (or chronic diseases not excluded in the study) may stay away from coffee. Indeed, people with heart disease may reduce their coffee intake as recommended by plenty of doctors, resulting in data suggesting people with lower rates of heart disease drink more coffee. If the people with heart disease then die of a heart attack, they up the death rate for the non-coffee drinkers.
Freedman, et al, also found that coffee is strongly associated with unhealthy behaviors – not just smoking, but also high levels of alcohol and red meat consumption. Only when these behaviors are controlled for is the mortality risk for coffee-drinkers lower than that for non-coffee drinkers. The problem is that we don’t know if coffee consumption is somehow closely related to activities such as drinking too much. While it seems implausible, increased coffee consumption may be causally related to unhealthy behavior – we all know the coffee at Dunkin’ Donuts is just an excuse for donuts.
It may have been this substantial call for caution in the study’s discussion of results that prompted The Atlantic’s concluding (and confusing) disclaimer. In fact, the last line of the study itself says it best: “Whether [there] was a causal or associational finding cannot be determined from our data.” So far, the NIH’s findings are a head-scratcher, and although they are in line with previous studies suggesting benefits to coffee consumption, the only certainty is that there needs to be more rigorous study.
Rebecca Goldin is the Director of Research of STATS at George Mason University.