I hate reading about studies only in mass media outlets because of the lack of critical evaluation. Unfortunately the real study isn't showing up in the Circulation journal's past three issues and they never did give the complete title.

However, even the LA Times article states "researchers found a statistical relationship between long hours of TV viewing and a shortened life span, but the study did not go so far as to find a direct cause".

This means the results are interesting and the subject should be further evaluated in a controlled study, but are useless for practical purposes. We don't know if it was the sedentary behavior or something else, like the potato chips and soda that often go with watching TV. It may be that their conclusion is correct, but it's impossible to know from this type of study.

Here's a discussion of a study that supports my point nicely.

As I stated, fully half of people with heart disease have normal serum cholesterol levels. No one denies this. So it's a poor indicator - anyone with a good cholesterol test result still needs to be told that there's an equal chance they have heart disease anyway. There are better indicators such as C-reactive protein that are largely ignored. Even the recent, terminally biased JUPITER study gives that method of detection a backhanded compliment.

The data in the Framingham study does agree with me. And so does the data in every other study out there - there is not a single one that demonstrates causality.

The meta-analysis you link to does no better - it shows only a correlation (their word: "association") that is useless to any practitioner. Meta-analyses are always suspect because the input studies and parameters are cherry-picked to have the desired outcome.

Deaths have decreased because emergency treatment at the hospital has improved and people are less likely to ignore the warning signs of a heart attack. Heart disease rates have actually increased.

Please. Researchers ignore data that break their theories all the time.

It may be worst in the medical world. For example, why do you think that cholesterol is targeted as enemy number one for heart health? There is no study that has ever demonstrated causality; 50% of people with heart disease have "normal" cholesterol; nearly all studies on the subject show that all-cause mortality is higher with low cholesterol; much better working theories exist.

So why is that hypothesis still treated as correct? Because reputations and huge amounts of money would be lost. Prominent people and institutions may even be found liable. Good science goes out the window in the face of that.

Regarding the subject at hand, you might want to look at what an ad hoc hypothesis is.