Statistician breaks down the health risk of vaping

Studies tend to distort the evidence to make e-cigarettes look dangerous. But the data overwhelmingly show that they're much safer than traditional cigarettes.

-----

In February 2022, the World Journal of Oncology published an article by a team of 13 researchers claiming that vapers are about as likely to get cancer as people who smoke traditional cigarettes.

And then the World Journal of Oncology's editors retracted the study because "concerns have been raised regarding the article's methodology, source data processing including statistical analysis, and reliability of conclusions."

The editors of the journal and the paper's peer reviewers failed to notice the study's many flaws prior to publication, but they ultimately concluded that it was bad enough to retract. Even so, it's worth dwelling on its problems because they're typical of what we see with statistical studies on this topic and other public policy issues.

The retracted study claimed a large sample size with data on 154,856 subjects. For assessing the cancer risk of vaping versus traditional smoking, what we should be looking at are vapers who never smoked traditional cigarettes and yet have cancer. There were 180 vapers with cancer in the study. But based on general population percentages, probably fewer than 100 had never smoked traditional cigarettes. That's too small a sample to draw robust conclusions. The median age of vapers in the study was 25, versus 62 for traditional smokers, and they had very different breakdowns of income, race, sex, and medical conditions. Adjusting for all these factors would require a minimum of 1,000 observations.

The authors claimed active vapers had 2.2 times the risk of cancer as the control group. But their logistic regression showed that people who never used cocaine, heroin, or methamphetamines also had a 2.2-times-higher risk of getting cancer. Why didn't the authors run with this finding—that cocaine might be a cancer preventative? Because it's absurd and would have brought ridicule.

An Internal and Emergency Medicine analysis noted 11 flawed studies that linked vaping to various diseases.

The tiny population of vapers who never smoked traditional cigarettes and who started using e-cigarettes before being diagnosed with a health condition is hard to identify, unrepresentative of the general population, and likely too small to draw conclusions from. No amount of tricky statistical work can overcome this basic data issue.

Proving that traditional cigarettes cause cancer, which they do, required two types of data: observational studies and experimental studies. First people noticed that cancer patients were more likely to be smokers than noncancer patients, and then careful experimentation teased out some of the mechanisms by which smoking led to cancer.

Observational studies, even without data issues, can show only an association, not causation. Although most vaping studies claim only an association, journalists, activists, and public officials are quick to assert causation. Experimental studies can show causation but can't measure the practical extent of an issue or possible offsetting factors.

One experimental study of vaping that drew press attention was published in the Journal of Nuclear Medicine.

One problem with this particular paper is that it studied only 15 people: five vapers, five users of traditional cigarettes, and five people who don't smoke at all. However carefully you select groups of five subjects, they can't represent a broad enough cross-section of users to draw any solid conclusions. This experiment also relied on screening volunteers and made no attempt at randomness or sampling the range of the population, meaning that each of the three groups of five subjects differed from each other in important ways.

The paper did not show that vaping causes lung damage—in fact, the researchers didn't check for that. Instead, it looked at "biomarkers," or chemicals thought to be associated with lung damage. By that measure, they found no difference between the five vapers and the control group of five people who had never vaped or smoked.

The study did find that uptake of a chemical thought to react to a biomarker for lung damage was higher in the five vapers than in the control group. But the five cigarette smokers included in the study had a lower uptake of that chemical than the controls, which makes the conclusion suspect since we know traditional cigarettes cause lung damage. Most likely, this correlation can be attributed to random chance.

Editor, Audio Production, and Graphics: John Osterhoudt

Camera: Jim Epstein

Graphics: Adani Samat

Photo: Douglas Graham/Newscom; Peggy Peattie/TNS/Newscom; Stanton Glantz by Noah Berger