July 18, 2012

Stanton A. Glantz, PhD

Smoking in PG-13 films causes 18% of new youth smoking

New research from Dartmouth shows that eliminating smoking from youth-rated PG-13 films would cut youth smoking by 18%, a gigantic effect.

Jim Sargent and his colleagues followed a national sample of nearly 6000 students for 2 years to assess the relationship between the amount of smoking they saw onscreen in movies and the likelihood that the youth would start smoking.  Moving beyond their earlier work that simply looked at the total amount of smoking youth were exposed to, the Dartmouth group separated the effects of smoking in G/PG, PG-13 and R rated films.  (There was little exposure to smoking in G/PG films.)

The study found that, while individual R rated films have, on average, more smoking than PG-13 films, youth received about 3 times as much exposure to onscreen smoking in PG-13 films than R rated films.

More important, the study showed that the effect of each exposure to onscreen smoking in PG-13 (teen films) and R films was almost identical.  For every 500 exposures, the odds that a youth started smoking increased by about a factor of 1.4.  This is the strongest evidence to date that it is the onscreen smoking -- not other associated behaviors like drinking or sex -- that stimulates youth to smoke.  (If it was these other factors, the effects of each exposure in PG-13 films would be lower than for R exposures).

Because the Dartmouth researchers separated the effects of PG-13 and R smoking, the could, for the first time, provide a direct estimate of how much youth smoking would drop if smoking were eliminated from PG-13 films:  an 18% drop.

This is a huge effect that could be produced at virtually no cost.

A comment on the 26% overall attributable risk estimate in this study:

The 26% attributable risk for the total effect of onscreen exposure to smoking in movies is lower than the 44 percent value that we have been using based on a pooled estimate (called a meta-analysis) of the four studies available at the time we computed this estimate.  The reason for this differences is that when you do these studies there is associated uncertainty (like in public opinion polling), so one would not expect to get the same answer every time. 

When I have time I will roll Sargent’s new study into the meta-analysis (what the fancy averaging is called) of all the available data and put out an updated pooled estimate.  Since the latest estimate is below 44%, it will bring the average estimate down some, but it will still very big from a public health perspective.   That is the important point, not the specific number.

Add new comment

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.