Just by reading all the research about diet, exercise and so on and by trying to apply it to your own life. For instance, bicycling is good for you, right? It's great for people who need a low-impact form of aerobic exercise. But then it turns out that you get osteoporosis from too much cycling:
Many factors contribute to osteopenia or osteoporosis (very low bone mineral density) in cyclists, but one of the culprits is the nature of the exercise itself. Cycling is a low-impact sport that puts little mechanical load on the bones. That's great if you have joint problems, but it's the weight-bearing nature of exercise that signals bones to create more mass. Without such stress, bones don't get stronger, making them more prone to injury.
Avid cyclists, both amateur and professional, seem to be especially at risk of bone injuries if they don't do any type of cross-training. (Swimmers may also be in danger, since that sport requires little mechanical loading as well.) The lower spine is a particularly susceptible area, since it gets almost no loading. The hips may get some from the action of pedaling.
There ya go. Or not, as the case might be. At least the article points out that the problems don't apply just to women, so that's an improvement over much of the oh-dear literature on elite athletes.
Then there's red wine. I'm sure you've read how it's wonderful for your heart if taken in moderation. Now it turns out not to be so wonderful for your mouth and colon (if the research is properly done which I don't have the energy to check):
The most comprehensive review of literature on the subject, carried out by the World Cancer Research Fund concluded that alcohol increases the risk of cancers of the mouth, voice box, throat, colon cancer, liver and breast cancer.
The report said there is a dose response, with the more alcohol drunk the greater the risk, and there is no level which has found not to increase the risk of some cancers.
Scientists had thought that drinking small amounts of alcohol was good for heart health but this is now being disputed.
Of course even a 160% increase in the risk of a rare cancer is actually a very small increase, even though those numbers look so frightening.
What's worth keeping in mind when reading all these studies is that they are almost always observational studies, not laboratory studies, and that establishing single causes for various illnesses from such studies is very hard indeed.
Then there's the problem of correlated variables. For example, if the rich eat more salads than the poor, on average, and if a study shows salad-eating to be linked to better health it could just be that what it is measuring is the impact of having more money on health. To rule out that possibility necessitates holding all the income and wealth related measures constant in the study, and often the data for that hasn't been collected.
The history of popularized medical research should remind us to be careful about accepting such findings at face value too quickly. Remember how eggs were the root of all evil once? Moderation in all things is still a good plan.