A recent synthetic meta-analysis of a handful of studies from the past four decades produced a flurry of news reports about the unhealthy effects of watching television, particularly type 2 diabetes and heart disease. The relative risks were in the 1.1 to 1.2 range for watching TV for 2 additional hours per day. The inevitable result was, of course, proposals for how to make people watch less television.
Does that glowing screen mess up people's hearts and insulin systems? Of course not. Sitting still, not engaging in much exercise, snacking, gaining weight, and perhaps living a limited social life and other more subtle factors cause the outcomes. Television is what people do when engaged in those activities. Sometimes it causes them and absent television people would be out getting exercise and not snacking, but probably not usually. Many would be at a pub instead, or reading, or on the computer, or something else sedentary and snackable. You could argue that those are more worthy human pastimes in various ways, but that is a different point. As just one example, I usually have the television on when writing this blog (muting or pausing it for a while when I actually write – I am not some superhuman multitasker), and would not be getting much more exercise even if it were not on. (Right now I am not-really-watching the remake of "The Day the Earth Stood Still", redeeming feature of which is that someone figured out that Keanu Reeves's acting skills are perfectly suited for the role of an anthropomorphic alien who learned to speak only that week and does not possess human feelings.)
Did the meta-analysis tell us anything we did not already know? As is usually the case, no. It averaged together a small number of studies in a way that someone could basically do in his head. Though the usual fatal flaw with such syntheses – very different exposures that are treated as if they were the same – is largely absent, since television viewing is pretty much what it is, the populations varied radically due to the timespan if nothing else. Obviously the association with the outcomes is going to vary across populations based on how they are trending toward heavier body weight, for example, or the availability of cholesterol and blood pressure drugs. In other words, it seems rather unlikely that the knowledge value of the new paper was positive.
Why was the unit of exposure 2 hours of TV watching? Undoubtedly because the lower result for 1 hour was insufficiently impressive and presumably not "statistically significant", and 2 was the smallest number for which the results were statistically significant. Standard model fishing by people who mistakenly thing there is some magical importance to statistical significance, in other words.
What can we say about this comment, in the CNN story:
The increased risk of disease tied to TV watching "is similar to what you see with high cholesterol or blood pressure or smoking," says Stephen Kopecky, M.D., a cardiologist and professor of medicine at the Mayo ClinicNonsense. Two of the effects of sitting rather than exercising (whether in front of a television or not) are less healthy cholesterol and blood pressure. So the comparison to those intermediate steps does not make sense. It is like saying that the toll from traffic accidents is similar to what you see from unintentional trauma. As for smoking, not even close – two hours of smoking a day raises heart attack risk by a lot more than 20% (as does smoking in general).
Working on tobacco related issues, I had always thought that Mayo's junk science claims focused on opposing tobacco harm reduction. I guess they are equal opportunity spewers of nanny-state-supporting nonsense, and will understate the risks from smoking if it might help scare people about something else.