Study Claiming that Vaping Causes Lung Cancer Fails to Document that a Single Case Occurred After the Person Started Vaping

An article published recently in the journal Frontiers in Oncology concluded that vaping causes lung cancer among young adults (age <50) based on the finding that lung cancer risk among dual users of cigarettes and e-cigarettes was higher than the risk among exclusive smokers. This was a case-control study of young adults in which the investigators used electronic medical records to identify patients with lung cancer and a matching set of patients without cancer and then compared the odds of having lung cancer among the dual users to the odds of having lung cancer among the smokers. 

The study found that the odds of having lung cancer among the dual users was 2.8 times higher than among the exclusive smokers. Based on this finding, the authors concluded that "vaping and smoking together accelerate lung cancer risk among young people, particularly in the development of pulmonary adenocarcinoma."

The Rest of the Story

To evaluate the study's conclusion, we need to review some basic epidemiology. In a case-control study, the odds ratio measures the odds of disease (in this case, lung cancer) among the exposed (in this case, dual users) to the odds of disease among the unexposed (in this case, smokers). Dual use is the exposure and it is being compared to exclusive smoking. 

OK, now a key thing to understand about the case-control design is that the ratio of the odds of disease among the exposed to the odds of disease among the unexposed is the same as the ratio of the odds of exposure among the diseased to the odds of exposure among the controls.  

To repeat that in terms of this study, the relevant odds ratio comparing the lung cancer risk of dual users to that of smokers is equivalent to:

The odds of dual use among smokers with lung cancer divided by the odds of dual use among smokers without lung cancer. We'll come back to this in a second.

Now, critically, in a case-control study, one has to go back in time in order to measure the exposure because if exposure and disease are measured concurrently, there is no assurance that the exposure preceded the disease, which is a key criterion for drawing a causal conclusion. If a case-control study measures exposure at the current time (concurrently) rather than retrospectively, it risks severe reverse causality, where the outcome influences the exposure rather than vice versa. This design is unsuitable for establishing causation, as it cannot confirm if the exposure occurred before the disease.

Importantly, in this study, it appears that smoking and vaping status was measured concurrently with the disease. The paper fails to state that the investigators went back in time to determine the historical trajectory of an individual's smoking and vaping status. It appears that they just drew the information from the record of the particular visit in the record database rather than searching historical records to determine smoking and vaping status in the past. For this reason, it cannot be documented that dual use preceded the lung cancer. It is very possible that dual use came after the lung cancer. And in fact, that is the most plausible explanation for the study results.

Let's go back to this point: The relevant odds ratio comparing the lung cancer risk of dual users to that of smokers is equivalent to the odds of dual use among smokers with lung cancer divided by the odds of dual use among smokers without lung cancer. The study found that the odds of dual use among lung cancer patients who smoked is 2.8 times higher than the odds of dual use among controls who smoked. So essentially, the authors are concluding that vaping causes lung cancer because when smokers are admitted to the hospital with lung cancer, they are more likely to be vaping than smokers who are admitted to the hospital but don't have any type of cancer.

Clearly, the most likely explanation for this finding is that getting lung cancer serves as a highly motivating event to trigger an attempt to quit smoking. And since e-cigarettes are a commonly used smoking cessation strategy, many of those diagnosed patients likely used e-cigarettes in an attempt to quit smoking.

Thus, it is not at all a surprise to find that the odds of dual use among smokers diagnosed with lung cancer is about 3 times higher than the odds of dual use among smokers who are not diagnosed with cancer. All that this suggests is that smokers who are diagnosed with lung cancer are more likely to be motivated to try e-cigarettes in attempt to quit than smokers who are not diagnosed with cancer.

I don't understand why the authors drew a causal conclusion from a case-control study that measured exposure concurrently with the outcome. This is a situation where reverse causation is the most plausible explanation for the study findings. 

Moreover, the authors do not even mention this possibility in the skimpy limitations section. In fact, the limitations "section" is just a single sentence: "One study limitation was that due to the nature of the electronic medical record data, we could not quantify vaping and smoking as detailed as we had planned, nor the timing of vaping relative to smoking. ... Another limitation was the lack of information on potential effect modifiers such as exposure to secondhand smoke, air pollution, and genetic predisposition to lung cancer." Well if they acknowledge that they had no idea of the timing of vaping relative to smoking then how can they possibly draw a causal conclusion?

Even worse, the paper takes the opportunity to use this single flawed conclusion to condemn harm reduction. Although they acknowledged that they had no idea of the timing of the decision of these smokers to try e-cigarettes, and although they have no idea how long the patients vaped before their lung cancer diagnosis, the authors argue that: "Our results are in marked contrast to the 'harm reduction' approach that deems vaping to be less harmful than cigarette smoking, instead showing that exposure to aerosolized e-liquid may in fact promote lung carcinogenesis, especially when combined with smoking."  

So now they go even further in their conclusion than their study would allow them even if reverse causation was not a problem. Here, they conclude that vaping is no less harmful than smoking. It certainly has the appearance that there is investigator bias involved with that bold (and false) statement. They are literally making a plea to alter the entire strategy for smoking cessation throughout the country based solely on the finding that when smokers are diagnosed with lung cancer, they then are more likely to use e-cigarettes than smokers who are not diagnosed with cancer.

Perhaps the clincher here is that the authors, in the abstract and introduction to the paper, make a false statement -- three times -- revealing a massive misunderstanding of the nature of e-liquids. They state: "the inhalation of heated aerosolized vaping oil has now replaced cigarette smoking as the major source of nicotine among young people." Then, they state: "Notably, the inhalation of heated aerosolized vaping oil is currently the major source of nicotine among young people." Shortly thereafter, they state: "While the inhalation of aerosolized nicotine-containing vaping oil is known to expose the lungs to carcinogens, only a few studies have examined the potential role of vaping as a lung cancer risk factor."

E-liquids do not contain oils. Vaping oils are used in THC vapes, not nicotine vapes (e-cigarettes). Vaping oils are not used to deliver nicotine. They are used to deliver THC, CBD, or essential oils. To be clear, the use of vaping oils can be quite dangerous. If vitamin E acetate is used to increase the viscosity of the oil, vaping it can cause severe lung damage and even death. Even without vitamin E acetate, vaping oils can cause lipoid pneumonia, which can result in permanent lung injury. The excipients in e-cigarettes are alcohols: propylene glycol and glycerin. They are not oils. 

I'm not concerned that a particular study has a flawed conclusion. However, what is very concerning is that anti-vaping groups are using this shoddy research to support their false claims that vaping is just as dangerous as smoking. This is causing real public health harm because it dissuades smokers from quitting, may cause some ex-smokers to return to smoking, and leads to public policies that result in blocking a major off-ramp for smokers looking to quit.