For those who actually bothered to read the study's full text, which appeared in the European Respiratory Journal days after the global feeding frenzy had ended, they would have learned the following. Among the 79 lung cancer subjects who participated in the trial, 70 of them smoked tobacco. These individuals, not surprisingly, experienced a seven-times greater risk of being diagnosed with lung cancer compared to tobacco-free controls. As for the subjects in the study who reported having used cannabis, they -- on average -- experienced no statistically significant increased cancer risk compared to non-using controls.
So how'd the press get the story so wrong? There are several reasons. First, beat writers based their stories on a press release rather than the study itself. Unfortunately, this is a common practice used by the mainstream media when writing about cannabis-related science. More often than not, media outlets strive to publish their reports prior to a study's publication -- a desire that all but forces reporters to write about data they have never seen. (Likewise, as a marijuana law reform advocate I'm also frequently asked by the press to comment on studies that are not yet public, though I typically choose not to.)
Second, the media chose to selectively highlight data implicating cannabis's dangers while ignoring data implicating its relative safety. In this case, the study's authors (and, by default, the worldwide press) chose only to emphasize one small subgroup of marijuana smokers (those who reported smoking at least one joint per day for more than ten years). These subjects did in fact, experience an elevated risk of lung cancer compared to non-using controls. (Although contrary to what the press reported, even the study's heaviest pot smokers never experienced an elevated comparable to those subjects who reported having "ever used" tobacco.) By contrast, cannabis consumers in the study who reported light or moderate pot use actually experienced a decreased cancer risk compared to non-using controls. (Bottom line, the sample size in all three subgroups is far too small to draw any sound conclusions.)