Eating chocolate might be good for people whose metabolisms show up as stressed in blood tests. Though I have to wonder whether attacking the underlying causes of high stress hormones would be more likely to deliver a real benefit.

The "chocolate cure" for emotional stress is getting new support from a clinical trial published online in ACS' Journal of Proteome Research. It found that eating about an ounce and a half of dark chocolate a day for two weeks reduced levels of stress hormones in the bodies of people feeling highly stressed. Everyone's favorite treat also partially corrected other stress-related biochemical imbalances.

One big problem with research on benefits of food on health: research that turns up a positive result is more likely to get published than research that turns up a negative result. So the body of all published research has a bias toward showing benefits.

Another big problem: short term effects do not always translate into long term reduction of illness or death. We end up with lots of promising studies that suggest dietary practices which are unproven or disproved many years later. Long term research takes too long and is so expensive that the number of hypotheses that get tested by long term research ends up being pretty short.

This study reminds me of a third problem: Some studies produce positive results because they happen to use experimental subjects most likely to benefit. Subsets of people who have more stress, a lousier diet to start with, or other problems are probably more likely to benefit from a diet change. Should you eat chocolate? The answer might depend on your levels of stress hormones.

In the study, scientists identified reductions in stress hormones and other stress-related biochemical changes in volunteers who rated themselves as highly stressed and ate dark chocolate for two weeks. "The study provides strong evidence that a daily consumption of 40 grams [1.4 ounces] during a period of 2 weeks is sufficient to modify the metabolism of healthy human volunteers," the scientists say

So does eating chocolate deliver a benefit? I'm still not convinced. But at least with chocolate my taste buds think I ought to lower my standard of evidence.

Update: Big population studies of diet and health will become a lot more useful once it becomes affordable to genetically sequence each person. My guess is that in some of the studies that find a benefit from a dietary practice for some of the people in that study their genomes were well matched to the dietary practice under study. The inability to control for genetic endowment is one of the causes of positive results that fail to generalize to hold up in other studies.

Similarly, if we all had implanted nanosensors reporting our metabolic condition our cell phones could query our nanosensors, report the results to a web site, and then get back recommendations for, say, exercise or chocolate or cruciferous vegetables.