Matthew J. Carpenter and colleagues recently published what is probably the most informative randomized controlled clinical trial of the effect of giving current cigarette smokers free e-cigarettes on their nicotine consumption: Effect of unguided e-cigarette provision on uptake, use, and smoking cessation among adults who smoke in the USA: a naturalistic, randomised, controlled clinical trial. They recruited 638 people from 11 cities, then randomly assigned them to receive one month of free NJOY tank-style e-cigarettes or a control group who did not receive any product. To participate in the study, people had to be current cigarette smokers who had not used e-cigarettes in the past six months, were not currently using any smoking cessation medication, did not have cardiovascular distress and were not pregnant. Participants who received the free e-cigarettes were told, “you can use these these e-cigarettes however you like, as much or as little as you like. You can reduce your smoking entirely, or you can just use e-cigarettes in places where you cannot smoke … You are not required to use the e-cigarettes, but you might want to give them a try.”
They then recorded the participants’ cigarette smoking and e-cigarette use a month and six months later. In contrast to most earlier randomized clinical trials, they reported complete results, not just whether or not participants were still smoking cigarettes. Specifically, participants were classified as smoke only, vape only, dual users (both smoke and vape) and quitters (neither smoke nor vape). This full reporting of outcomes is a major strength of the study and should be the model for all future studies of the association between e-cigarette use and cigarette smoking, whether RCTs or population observational studies.
Consistent with earlier RCTs conducted in a clinical environment, they found a statistically significant increase in the odds of no longer smoking cigarettes among the people randomized to receive free e-cigarettes compared to the people randomized to receive no product: OR 1.8 (95% confidence interval 1.0-2.2). In absolute terms, at 6 months 14% of the smokers given free e-cigarettes had stopped smoking cigarettes compared to 8% of the control group. Thus, for every 100 people given free e-cigarettes, 6 more stopped smoking cigarettes that would have quit anyway thanks to the e-cigarettes.
Like all the earlier studies, these “stoppers” included “switchers” who stopped smoking cigarettes but were be consuming e-cigarettes and “quitters” who had stopped smoking cigarettes without continuing to consume nicotine from e-cigarettes.
Because Carpenter and colleagues reported their data so completely, I could compute whether being provided e-cigarettes was associated with being nicotine free — the usual endpoint for smoking cessation studies before e-cigarettes came along — at 6 months (i.e., quitters). The picture here looks quite different: There was no significant difference in having quit (stopped using nicotine: 8% of e-cig group vs 6% of control group; p=0.510).
The results for dual use were even more interesting. At 6 months, 45% of the people who received e-cigarettes were dual users compared to just 11% of the control group (OR 6.8, 95% CI 4.2-10.9, p<0.001). In other words, for every person who stopped smoking cigarettes (including switchers who continued using e-cigarettes and quitters who did not smoke cigarettes or use e-cigarettes) in the group that were provided e-cigarettes, 2.7 smokers became dual users. This compares with 1.8 in the control group.
This difference is even more pronounced for quitters (stopped smoking and stopped e-cigs): For every quitter in the e-cigarette group there were 5.7 dual users compared to 1.8 in the control group.
To their credit, Carpenter and colleagues discuss high levels of dual use (based on smoking behavior at 4 weeks), but dismiss this finding by noting that about one-third of dual users reduced their cigarette consumption by more than 50%. This 50% smoking reduction is a popular measure among such studies. The problem is that there is not good evidence that a 50% reduction has substantial health benefits. Indeed, a 2021 meta-analysis published by FDA scientists showed that, while a 50% cigarette reduction was associated with lower lung cancer risk, there was no significant mortality benefit associated with a 50% cigarette reduction.
In fact, dual use is likely worse than just smoking because of the fact that dual users are subjecting themselves to the toxicants in e-cigarette aerosol in addition to the toxicants in cigarette smoke. While some are the same, different exposures are created by the two products.
In addition, Carpenter’s study, like all the other RCTs (as well as the population observational studies), assumes that e-cigarettes, while not harmless, are minimally risky compared to cigarettes, despite the rapidly growing case that e-cigarettes pose substantial risks, including for heart and lung disease.
These biological findings are particularly important when assessing the likely individual and population health impacts of promoting e-cigarettes for stopping smoking (as opposed to stopping all nicotine) given the small increase in stopping smoking and the fact that it is swamped by increases in dual use.
In addition, for assessing the population health impact of authorizing the sale of e-cigarettes, the real-world population observational studies remain most relevant for regulatory decision making because they reflect actual use patterns of actual products.
Lessons for research going forward
There are also several lessons for best practice in future studies on the relationship between e-cigarette and cigarette smoking behavior, whether in RCTs or population observational studies:
- Authors should fully report results using Carpenter et al’s categories, including continued exclusive and dual or poly use of e-cigarettes
- Authors should report results for nicotine cessation as the primary endpoint (what Carpenter calls “quitting”) as well as analysis of just stopping cigarettes or other tobacco products
- Absent strong evidence that cutting cigarette consumption confers substantial health benefits, people should stop using cigarette reduction as a health endpoint
- Absent strong evidence that e-cigarettes reduce disease incidence (not just biomarkers of exposure), authors should stop assuming that e-cigarettes are dramatically less dangerous than cigarettes.
- Continuing nicotine addiction should be added as an “adverse event” in reporting RCTs.
A few technical points
I used the data in Carpenter et al’s supplemental Figure 6 to get the numbers for the 6 month levels of dual use and quitting. The 6 month numbers for stopping cigarettes are slightly different from the numbers in their Table 2. I wrote and asked Carpenter about the difference; here is what he told me: “Table 2, within the manuscript itself, uses an approach where missing is assumed to be non-abstinent (or not have made QA, or not have achieved 50% reduction in CPD). … For Supplemental Figure 6, we used Last Observation Carried Forward (dual use was not a primary focus of the study, nor a primary focus of this paper). We were transparent with that, stating LOCF within Statistical Analysis, though in retrospect we should have repeated this as a footnote within the table itself. It is our hope to dig into patterns of dual use a bit further and will consider other options beyond LOCF.” What they did was reasonable. Applying the approach in their Table 2 (if the data were available) would change the numbers a bit, but not the overall conclusions in this blog post.
Carpenter and colleagues present unadjusted ORs in their analysis on the grounds that randomization produced reasonably similar control and intervention groups. While it is a good thing that the randomization worked out so well, I would also have done an analysis including possible confounders (i.e., sociodemographics) at the individual level because doing so would produce more precise estimates and may have affected the results.
In their instructions to people who received e-cigarettes, they said, “you can just use e-cigarettes in places where you cannot smoke.” According to the American Nonsmokers’ Rights Foundation, e-cigarettes are already included in many smokefree laws.
Here is the abstract:
Background. As summarised in the most recent Cochrane review, the few clinical trials on e-cigarettes are largely focused on smoking cessation. We aimed to determine the naturalistic uptake, use, and impact of e-cigarettes among adults who may or may not want to stop smoking.
Methods. In this naturalistic, randomised, controlled clinical trial, adult smokers, across the motivational spectrum and with minimal history of e-cigarette use, were recruited online from the general community within 11 cities across the USA. Participants were randomly assigned (2:1) to either receive either a free 4-week supply of flavoured, tank-style e-cigarette, or not. E-cigarette group participants received a battery and device with up to 30 pre-filled tanks, offered among five flavours, with minimal instructions on use. The study’s primary purpose was to descriptively assess naturalistic uptake and usage of the e-cigarette, and to secondarily assess its impact on smoking behavior. The latter, assessed through six months of follow-up, included: a) self-reported 7-day point prevalence abstinence, b) incidence of quit attempts, and c) smoking reduction. This trial is registered at ClinicalTrials.gov, NCT03453385.
Findings. Between 5/2018 and 3/2022, 638 adult smokers were enrolled and randomly assigned (427 in the e-cigarette group and 211 in the no-product control group). Uptake of e-cigarettes was robust: approximately 70% of participants used the product, with average usage exceeding 4 days per week during the initial 30 days. Based on an intent-to-treat approach where missing data is imputed as smoking, almost all behavioral outcomes favored the e-cigarette group relative to no-product control, including point prevalence abstinence at six months (Odds Ratio [OR] = 1.8; 95% Confidence Interval [CI] = 1.0-3.1), cumulative incidence of 24-hr quit attempts (OR = 1.5; 95% CI = 1.0-2.2), and having reduced smoking by at least 50% since baseline (OR =1.8; 95% CI = 1.2-2.7). Results were similar under an alternative imputation.
Interpretation. Complementing cessation-focused trials, results suggest that unguided e-cigarette use also leads to smoking cessation, allaying the notion that causal effects of e-cigarettes on cessation are not reflective of real-world scenario of self-determined use. For smokers who may not be able to quit using existing pharmacologic approaches, e-cigarettes may be considered to achieve that purpose.
The full citation is: Matthew J. Carpenter, Amy E. Wahlquist, Jennifer Dahne, Kevin M. Gray, K. Michael Cummings, Graham Warren, Theodore L. Wagener, Maciej L. Goniewicz, Tracy T. Smith. Effect of unguided e-cigarette provision on uptake, use, and smoking cessation among adults who smoke in the USA: a naturalistic, randomised, controlled clinical trial. eClinicalMedicine 2023; 63: 102142. https://doi.org/10.1016/j.eclinm.2023.102142. It is available here.
(We published this point in a letter to the editor about the Carpenter paper; you can read our letter and Carpenter’s response.)
Stanton Glantz blog 