Wednesday, January 3, 2018

Tobacco Gateway Report Omits Important Information

Here we go again.  Another federally funded study from the University of California, San Francisco, claims that “Nonsmoking adolescents who use e-cigarettes, smokeless tobacco or tobacco water pipes are more likely to start smoking conventional cigarettes within a year.” (UCSF press release here)  Researchers analyzed data on some 10,400 teens enrolled in the Population Assessment of Tobacco and Health (PATH) Survey in 2013-2014, then followed up with the subjects one year later.

“We found that teens who experimented with tobacco in any form were at greater risk of future smoking,” said senior author Dr. Benjamin W. Chaffee; his study appears in JAMA Pediatrics.

The researchers report that teens who had ever used e-cigarettes, hookah, other combustible products or smokeless tobacco at the baseline survey were two or three times as likely to be smoking cigarettes (in the past-30 days) one year later than those who had not tried any tobacco product.  Those who had tried two or more products were 3.8 times more likely to be smoking.

The authors used a sophisticated analysis to arrive at these results, but news stories describing this as evidence of a gateway to teen smoking (examples here and here) are inaccurate.

Chaffee and his colleagues, including anti-tobacco crusader Stanton Glantz, omitted information that is critical to putting their findings in perspective.  Although teens trying other tobacco products were more likely to smoke, the majority of new smokers after one year came from the group that had not tried tobacco at baseline.  I offer the following calculations based on obscure information in the published article.

Odds Ratios (ORs) And Numbers of Teens Smoking Cigarettes After One Year, According to Ever Tobacco Status at Baseline
Ever Tobacco Status- Baseline (n)OR (95% Confidence Interval)Number Smoking At One Year (%)

Never tobacco use (9,058)Referent175 (79.9)
E-cigarettes (255)2.1211 (5.0)
Hookah (189)2.158 (3.7)
Other combustible (114)3.087 (3.2)
Smokeless tobacco (93)1.533 (1.4)
Two or more products (200)3.8115 (6.8)

All (9,909)
219 (100)

After one year, 219 teens had smoked a cigarette in the past 30 days, and 175 of those (80%) had never used any tobacco product at baseline.  Even though the odds of smoking were higher among youth who had tried other products, the number of smokers contributed by each of these groups was minuscule.  (While actual survey numbers may vary slightly, the relative contributions of the groups will not change.)

The Chaffee article emphasizes odds ratios but omits or obscures important contextual information.  While teens who try one tobacco product are more likely to try another, the dominant gateway in the PATH survey was from no previous tobacco use to cigarettes.

No underage tobacco initiation is acceptable; neither is misdirection by researchers.


Bill Godshall said...

Thank you for exposing this research misconduct.

And if the UCSF researchers had NOT excluded all "ever cigarette smokers" in their study at baseline, they would have almost certainly found more than 219 baseline "ever cigarette smokers" who were cigarette smokers a year later.

Unfortunately for scientific integrity and government accountability, many DHHS officials and grant administrators are fully aware of this research misconduct by UCSF, which is why they gave them the money.

Roberto Sussman Livovsky said...

Focusing on odds ratios and ignoring (or hiding or playing down) the actual demographics is common in many "gateway" studies. I found out the same problem in a study on Mexican teenagers [P Lozano et al, Drug and Alcohol Dependence 180 (2017) 427-430]. They report that vaping but not smoking kids at baseline in 2015 had a 40% risk of becoming "conventional" smokers (at least a cigarette in the last 30 days) in 2017 in comparison with never smokers at baseline. When you break down the demographic data, it turns out that of the 247 conventional smokers in 2017, 223 were never smokers in 2015 and only 24 (about 10%) had tried vaping but not smoking in 2015. Also, the correlation between vaping trial and conventional smoking was not statistically significant (only the relation between trial of e-cigs and trial of cigarettes was significant). All this, the demographics and the lack of statistical significance, is buried in the paper, so you have to dig it up.