Well, here's another stunner. After all this time, we still don't know what amount or what kind of exercise is best. With all the money, marketeering, and other scientific interest involved in understanding what makes us sick, what we know is that some exercise is good for you...but not too much. And older people don't like vigorous exercise.
How could it be that after billions of dollars in research (and promotion) over many decades, we don't know? After all, you can find routine comments about the virtue of exercise in, say, Victorian novels (and, though we haven't looked, probably in Hippocrates 2500 years ago).
This reflects the serious if not fundamental problem with epidemiology, a problem at least as serious as making causal inference in genetics, where at least DNA segments can be objectively identified. In our posts last week on reductionism, we discussed the way that science 'tunnels' through complex reality by reducing studies to isolated variables to identify their causal effects on some measured outcome. We noted that in genetics this has provided a phenomenally successful way to discover many things that genes do. But it leaves the genetic contribution to most interesting traits, known by a catchword as 'complex', poorly accounted for in any specific way.
In environmental and lifestyle epidemiology, that deals with the common causes of disease and death in the industrial world, even tunneling (with all its limitations) is exceedingly problematic. That's at least in part due to the fact that exposure variables, unlike genetic sequences, are poorly defined and measured. 'Exercise' or 'diet' are examples, but so are many other aspects of modern life. This is especially true when the fact as well as the amount and duration of exposure are needed. Who can remember what they did, or ate, decades ago--all the way, perhaps, back to early childhood? At least, we're exposed to our genes from conception, and infectious diseases often are clear-cut point-causes where a single exposure can do the trick: one sneeze from someone with the flu is enough.
Besides the fact that it's hard to isolate individual variables, such as diet or exercise, in a cogent experiment, the variables of interest are heavily confounded, that is, diet is correlated with many other aspects of our lives, measured (e.g., education, income) and unmeasured. The same is true in genetics (e.g., what's called 'linkage disequilibrium' or 'epistasis', for those who know genetics), but easier to at least think about if not measure (though by no means trivial). But if eating chocolate rather than, say Supersize sodas, is correlated with exercise, and because these are both correlated with social status, jobs, education, neighborhood, race, and so on, then it is nearly impossible to isolate one of those variables. Confounding is a very serious, if not lethal, problem for application of 'tunneling' approaches in epidemiology, even if they would work. But because of complex causation, it probably wouldn't.
We've previously posted about the problem that preventive screening as in mammography to detect breast cancer may over-diagnose and over-treat lesions that would go away on their own or turn out not to be cancer after all. Studies showed that those not screened had fewer deaths from cancer than those who were screened. This has both monetary and psychological costs. It challenges various vested interests, too, so it's not surprising that we need yet another study (with its costs, too). So the latest U-turn in this saga is getting headlines, threatening policy changes (or no-changes), because it defends screening. And a BBC report interviewed a Scandinavian expert who bluntly ridiculed the current study as patently wrong. We can't judge without attempting to read the latest studies, and their skeptics, which we haven't done. The statistics are subtle and one can probably find in them what one wants to find. But the point here is the problem of uncertainty, even in heavily studied issues, is what concerns us here.
There's no easy answer, but the fact that there's a problem is clear. Candid epidemiologists who recognize this (privately, of course, because doing so publicly will jeopardize their funding changes, as we have been told many times) often turn in frustration to genetics. They believe that we have The Truth because genes, being molecules and the blueprint for life, have to be more tractable. Well, sorry, but it's not so simple on the genetic side of the fence either.
Dealing with complexity is a challenge. Somehow, the latest finding still makes it into the news, even though we really know that next week or next year the reverse will be announced. When and how we'll start making better progress is anyone's guess.
Meanwhile, do your pull ups, but don't pull out of your health club membership.