Watch TV and die

A paper in the journal Circulation last month prompted lots of excitement in the media. Small wonder: it said that watching TV increases your chances of dying. The more you watched, the worse for you.

Or at least that’s how it was reported. The Times ran with: “Television watching can shorten life”, The Guardian with: “Watching television increases risk of death from heart disease”, while The Daily Telegraph said: “Every hour per day watching television "increases risk of heart disease death by a fifth”.
Did the study really say that? And if it did, how persuasive was the evidence?
The research was conducted in Australia by Professor David Dunstan of the Baker IDI Heart and Diabetes Institute in Melbourne and colleagues from Australian universities. A cohort of 8,800 people were asked a lot of questions about their diet, lifestyle, age, sex, and TV viewing habits. They were followed up for an average of 6.6 years, in which time 284 of them had died. The team found a weak but statistically significant correlation between the hours of television-viewing and the likelihood of dying due to all causes or cardiovascular diseases.
Compared with those who watched TV for less than two hours a day, those who watched for more than four hours a day had a 49 per cent greater chance of dying of all causes (risk ratio 1.49, 95 per cent CI 1.06-2.09) and a 85 per cent greater chance of dying from cardiovascular disease (risk ratio 1.85, CI 1.03-3.33). There were very few heart deaths – just 22 in the most TV-addicted group, and 87 in total
The conclusion? “Television viewing time was associated with increased risk of all-cause and CVD mortality.”
But only just. The authors stress repeatedly that the correlation is weak and of borderline significance in most cases, and further note:
(1) The study used TV viewing as a measure of sedentary behaviour. The authors claim that this is a good enough proxy for other sedentary behaviour patterns. But this obviously means that it isn't necessarily television-watching that can increase the risk of death. Hours spent watching TV while doing other activities – ironing, or cooking, for example – weren’t counted.. So, does sitting by the window and staring out of it for over four hours a day also increase the risk of death? If it does, then television is not the specific factor.
(2) The viewing times were self-reported, so respondent bias cannot be ruled out. And TV viewing and exercise were assessed only at the start of the study. This meant that any lifestyle changes in the following years haven't been accounted for.
(3) There is the possibility of confounding. The authors clearly state: “Although we adjusted for several potential confounding variables, it is possible that other unmeasured or unknown confounding factors may have accounted for the associations that we have reported.”
(4) The possibility of reverse-causality, that is, where undiagnosed or other conditions may have limited mobility and caused an increased television-viewing time, have not been ruled out.

(5) The authors acknowledge that those who spent more time watching television in general had worse health. Measures of BMI, waist circumference, blood pressure, total cholesterol, triglycerides and fasting glucose at the start of the study were all higher in those who watched the most hours of TV a week. These factors alone would have accounted for higher heart deaths, and although they were corrected for, it is open to argument whether that correction is adequate.

(6) Finally, they also showed that those with the longest hours in front of the box were less likely to have completed 12 years of education. Given the strong association between education level and disease, this may have biased the results despite tests that showed education levels did not significantly modify the associations between TV-watching and dying.
At best, this study shows that sedentary behaviour, of which hours of TV-watching is a proxy, is associated with modest elevations in death from heart disease and from all causes. There is nothing intrinsic in television that makes people more likely to die. And, given the limitations of the study,
television executives can breathe a sigh of relief. 
To be fair, The Times, Guardian and Telegraph all quoted Professor Dunstan making these points, and can hardly be criticised for reporting the study as they did, given that Circulation had given it the title: Television Viewing Time and Mortality. Reporting in a newspaper a study that shows weak associations without leaving an exaggerated impression of the findings is extremely difficult.