Wednesday, August 27, 2014

In the CDC-FDA E-Cigarette Study, “Probably Not” Is the New “Yes”

Assume that you conducted a survey in which you posed two multiple-choice questions:
“Do you think you will smoke a cigarette in the next year?”
“If one of your best friends were to offer you a cigarette, would you smoke it?”

Respondents could choose from these answers:

Definitely yes
Probably yes
Probably not
Definitely not

You’d add up the “definitely yes” and probably yes” responses to tally those intending to smoke; and you’d total the negative responses to gauge how many are unlikely to smoke.

This would be a straightforward and uncomplicated task, unless you were a CDC or FDA analyst, milking the National Youth Tobacco Survey (NYTS) for scary numbers.

On August 25, the CDC issued its latest sky-is-falling press release, suggesting that e-cigarettes are driving teenagers to smoke (here).  The release focused on a study coauthored by CDC and FDA researchers (abstract here) whose core finding was: “Among non-smoking youth who had ever used e-cigarettes, 43.9 percent said they intended to smoke conventional cigarettes within the next year, compared with 21.5 percent of those who had never used e-cigarettes.” 

To reach this conclusion, the CDC-FDA re-defined “probably not” to mean “yes, I will.”  Adolescents who answered “probably not to either of the two questions were classified as intending to smoke.

The feds used 2013 data that is not yet public, but using the 2012 NYTS I can show you how much the distorted definition matters.

This table shows the numbers of never and ever users of e-cigarettes intended to smoke, using the CDC-FDA definition (i.e., “probably not” means “yes, I will”).  The percentages in parentheses are weighted to reflect the population of the survey.

Never Users of E-cigarettes
Ever Users of E-cigarettes
No intention to smoke
13,312    (76%)
  70    (41%)
Intention to smoke
  4,360    (24%)
  80    (59%)
17,672  (100%)
150  (100%)

Using conventional definitions, I produced the chart below. Any two yes responses defined intention to smoke, any two no responses were no intention, and mixed responses were just that, mixed.  These are my results:

Never Users of E-cigarettes
Ever Users of E-cigarettes
No intention to smoke
17,103  (97%)
128  (81%)
Mixed intention
     422  (  2%)
  13  (11%)
Intention to smoke
     147  (  1%)
    9  (  8%)
17,672  (100%)
150  (100%)

This paints a completely different picture of the e-cigarette situation.  The appearance that adolescents who have ever used an e-cigarette (even one puff) might be more likely to intend to smoke is based on the responses of just nine survey participants.

Carl Phillips has extensive comments on at the CASAA blog (here) and (here).

This is not the first time that a highly questionable definition has been used to fabricate a highly speculative gateway claim (here).  I assure you that this is probably not the last bogus CDC analysis of youth e-cigarette use.


Oliver Kershaw said...

The Authors have made up their own concept "intention to smoke", justified (in their eyes) by the work of Pierce and colleagues, which they cite.

Yet Pierce and colleagues were looking at the lack of a "firm intention not to smoke" as the central vulnerability, not an "intention to smoke" - this distinction is not subtle, and it makes one wonder how on earth the paper passed peer review.

Lollylulubes said...

I'd hazard the usual guess that they're all in it together but, in the unlikely event someone isn't, well it's the politically correct thing to do. I recently saw something about Glantz from years ago and that's the response the reviewer gave for passing off his junk.