Dual Use of Cigarettes & E-Cigarettes: A Scientific Journal Exchange

With wider acceptance that e-cigarettes are a viable harm-reduction method among people who smoke, the scientific battleground has retreated to "dual use." A recent exchange captures this well.

Artwork by Pawel Bugucki

Background: The Harm Reduction Battle Has Shifted to Dual Use

E-cigarettes pose a small fraction of the harm of cigarettes and are highly effective for helping people stop smoking, whether if used as a pre-planned “medical treatment” for smoking cessation (Cochrane review) or naturalistically as an alternative consumer product without any specific intentions (Foxon & Niaura 2025; Jackson et al. 2025). After switching from smoking to vaping, people with COPD can reduce their symptoms (Polosa et al. 2020), myocardial infarction patients can reduce their chance of another major cardiovascular event (Gao et al. 2025; Supplemental analyses are a lot stronger than the main analysis), etc.

Evidence is strong enough that, in my experience, it’s become cautiously accepted in scientific circles that switching to e-cigarettes can benefit people who would otherwise continue to smoke. But any such statements are quickly followed by a caveat insisting that “it must be complete switching.”

The literal meaning of this is to convey that dual use (of both cigarettes and e-cigarettes) is not considered acceptable (in the eyes of most researchers). A level deeper, the meaning is that the opposition to e-cigarettes has retreated to the topic of dual use, now that the evidence is so strong that e-cigarettes are far less risky in a head-to-head comparison. Maybe it’s just my imagination, but this caveat also seems to serve as a social signal of taking the middle ground: “I support e-cigarettes for harm reduction, but only in certain circumstances — I’m not one of the extremists.”

Why the Opposition to Dual Use?

The simplest rationale for concerns about dual use is that “no amount of smoking is safe.” Which is true, but this doesn’t mean that the amount of smoking doesn’t matter at all. If someone cuts their cigarette consumption in half after starting dual use, the fact that there’s still some harm from smoking doesn’t negate that the person has reduced their total harm. Studies examining biomarkers of harm found that a 50% reduction in cigarette consumption reduces harmful exposures by approximately the same amount (Cohen et al. 2021).

The other main rationale for concerns about dual use is the large volume of flawed research. Case in point: the Glantz & Oliveira da Silva 2026 systematic review, which will come into play later. The following is a template that I’ve seen hundreds of times, where the paper:

1. Analyzes national survey data (usually cross-sectional, but even the longitudinal ones are not immune from flaws);

2. Groups people by what products they happened to be using at the time of the survey (usually exclusive smoking, exclusive vaping, dual use, and non-use);

3. Compare whether these groups have different rates of some health outcome measured in the survey (and research groups that engage in ‘salami publishing’ can go down this long list); and

4. Blame any differences between groups on a causal synergistic effect of vaping.

Briefly, the flaws in most research on dual use are:

• This is an “apples-to-oranges” comparison, because people who dual use are different from the other product-use groups in many ways. These pre-existing differences could explain any difference in health outcomes.

• Of many pre-existing differences between dual users and other groups that explains differences in health risk, probably the biggest is smoking behavior & history. This would require (at a bare minimum) adjusting for variables like pack-years and how long former smokers have been quit, but studies rarely do this.

• This confounding is exacerbated by the way these studies define product-use groups. Since most surveys are done once a year and they only ask about past-month use, a survey is less likely to “catch” someone dual-using if they were only doing so for 2 weeks out of the year. As a result, the people who do get classified as dual-using are more likely to have been in that state for longer, which by definition means more cumulative cigarette exposure.

• Many analyses don’t, or can’t, separate out people who developed the health condition before they ever started vaping or dual-using — in which case vaping/dual use cannot have been the cause. In fact, as Brad Rodu showed here, for many conditions that develop with age (e.g., COPD), it’s likely they developed the health condition first. This may have motivated them to switch to vaping (reverse-causality).

A better approach would be to follow individual people before vs. after they started dual using, to examine what their own trajectory is compared to their own baseline. This avoids the confounding and “apples-to-oranges” comparisons above.

(Finally getting to my original point in this post) That’s exactly what Pesola et al. did in their study on dual use.

Original Article: Pesola et al.

Pesola et al. recently published a secondary data analysis (link) drawing from their previously-published RCT on e-cigarettes for smoking cessation. In this secondary data analysis, these researchers did a deeper dive on what patterns of e-cigarette use were more often followed by successful cessation: specifically looking at dual use (the focus of this post) and flavors (which I won’t focus on here).

The main findings were that participants who were dual-using (versus exclusively smoking) at Week 1 (pooled across both study arms, i.e. regardless of whether they were in the group randomized to receive e-cigs in the original RCT) were over 4 times as likely to have stopped smoking at Week 4. In terms of absolute numbers, this was 23.9% of participants who dual-used at Week 1 who had completely stopped smoking by Week 4, compared to only 5.0% of those who exclusively smoked at Week 1.

The study also had follow-ups at 6 months and 1 year. While the numbers were too small to analyze complete cigarette abstinent at 1 year (N=17 among the Week 1 dual use group and only N=1 among the Week 1 exclusive smoking group), the results went in the same direction (i.e., favored higher cessation rates after dual use). Moreover, there was also enough data to examine the secondary outcome of 50% reduction in smoking: those who were dual-using (vs. exclusively smoking) at 6 months were significantly more likely to have achieved a 50%+ reduction in cigarette smoking at 1 year (24.6% vs. 5.2%, respectively).

Authors conclude:

[These findings reassure] those worried about dual use that such use is associated with reduced smoking and a higher chance of stopping smoking altogether later on.

Criticism from Stanton Glantz

Longtime tobacco control advocate Stanton Glantz wrote a Letter to the Editor (link) criticizing Pesola et al.:

Essentially, Glantz objects to the fact that Pesola et al. draw a conclusion supported by their data that conveys something favorable about dual use. (It’s hard to summarize this in a way that doesn’t sound like I’m strawmanning it.)

The only objection strictly about the analysis of Pesola et al. is the lack of statistical testing for the 1-year outcome of complete abstinence (as I mentioned above, the numbers were too small to be analyzed).

The main focus of the criticism is instead on the framing:

The more important thing the authors ignore, however, is that providing e-cigarettes as a cessation aid produced 3.7 dual users for every person who stopped smoking cigarettes.

Among smokers who were provided e-cigarettes as a cessation aid, 17 of 107 dual users at 1 week had stopped smoking at 1 year (table 2 of new paper). That means that 180 were dual users at 1 year. That number compares with 79 smokers who had stopped smoking (from Table 2 of their earlier paper). That gives 180/79=2.3 dual users for every smoker who stopped cigarettes.

Sort of a Tangent, But I Think This Math Has 3 Errors

Here’s Pesola et al.’s Table 2 which Glantz bases the above numbers on:

1. Glantz’s first paragraph says 3.7 dual users created for every successful quitter, but the 2nd one says 2.3.

2. Glantz’s second paragraph says “17 or 107 dual users… had stopped smoking… That means that 180 were dual users.” Obviously 107 - 17 does not equal 180, but I think the 107 is a typo and should be 197 (i.e., the starting N of Week 1 dual-users in the 1st row).

3. The 17 number may also be wrong; it’s based on 17 of the Week 1 dual-users who were abstinent at 1 year; but all we know is that they were not abstinent, not that they were dual users. They could have gone back to exclusive smoking, and in fact the 2nd row shows that many of them already had at Week 4. (Credit goes to eagle-eyed Jukka Kelovuori for spotting this one).

Back to the Main Criticism on Reporting Favorable Results from Dual Use

In addition to not presenting this fact, the new paper minimizes the importance of continued dual use, “reassuring those worried about dual use that such use is associated with reduced smoking.” …

The authors ignored two systematic reviews of the health effects of dual use, one of 49 studies as of April 2021 that concluded “Existing studies indicate that dual use is at least as, or probably even more, harmful than [exclusive cigarette smoking] and another of 93 estimates as of October 2023 that concluded, “Pooled odds ratios for dual use versus cigarettes were increased for all [analyzed] disease outcomes (range, 1.20 to 1.41).”

One of the two systematic reviews referred to is Glantz’s own systematic review which is overwhelmingly based on the flawed “apples-and-oranges” type of papers I described at the beginning.

Essentially, Pesola et al. found, and accurately reported, their findings that dual use more often resulted in smoking reduction or smoking abstinence. Glantz is objecting to them not referencing his own research (which is based on a much more flawed evidence base), and more notably, not changing their own data-based conclusions on Glantz’s research which came to a different conclusion. (Again, I am not exaggerating; see the quote below).

He also ends with an inflammatory implication about “a failure of peer review:”

A failure of peer review?

It is not surprising that the authors ignored these papers; e-cigarette advocates have a habit of ignoring inconvenient evidence. It is disappointing, however, that the reviewers and editors of Nicotine and Tobacco Research did not catch these problems in peer review and at least require authors to remedy these problems and modify their conclusions accordingly prior to publications. [emphasis mine]

Reply 1: From Editor at the Journal

Glantz’s inflammatory remarks about “a failure of peer review” prompted not one, but two replies from the journal. The first (link) is by Angela Attwood, who I assume was the handling editor on Pesola et al.’s article, offering justification on why some of the points raised did not warrant correction or action during peer review.

On the criticism that statistical testing was not performed at the 1-year follow-up for outcomes of complete abstinence (due to the small number of cases):

When numbers are low, presenting descriptive outcomes rather than conducting underpowered statistical tests is an accepted approach.

On the lack of citation to Glantz’s meta-analysis:

Crucially, the body of literature that these reviews draw on is immature with a notable lack of high-quality studies. Pisinger and Rasmussen identify several weaknesses in the available evidence…

Many individual studies are observational, and pre-existing differences between individuals who dual use, exclusively vape or exclusively smoke are often not acknowledged or inadequately adjusted for in analyses… Causality is difficult to determine; these differences could precede use in some cases…

It remains a clear possibility [that for some people], dual use may have long-term positive outcomes either through lower exposure to smoking and/or using e-cigarettes as a pathway out of smoking.”

Reply 2: From Editor-in-Chief at the Journal

Editor-in-Chief Caitlin Notley also responded to address the “failure of peer review” accusation (link):

On the allegation of ‘a failure in the review process’, the NTR Editorial Board wish to assert our unwavering commitment to scientific integrity…

Our editorial decisions are grounded in independent peer review, methodological scrutiny, and adherence to established ethical standards. That reviewers and Editors did not include reference to the two reviews cited in this letter should not be taken to infer that they are ‘e-cigarette advocates..[with] a habit of ignoring inconvenient evidence’. For example, it could reflect that those reviews, one of which is the letter authors own publication, have widely discussed limitations (inclusion of many cross-sectional studies, potential confounding from former smoking history and interpretation of the results).

Reply 3: From Original Authors

Original authors also replied (link), with a title that perfectly and succinctly captures their rebuttal: “Selective contrary references do not change empirical data.”

On the criticism that for every successful quitter, multiple dual users are created:

All existing stop-smoking interventions yield more treatment failures than successes at one year. If among smokers unable to quit immediately, dual use increases their odds of subsequent cessation and significantly reduces their intake of tobacco toxins compared to continued exclusive smoking, that is a positive move.

On the lack of citation to Glantz’s meta-analysis:

Regarding our use of citations, we cited one example of concerns and one example of reassurances for both the effect of dual use on smoking cessation and on health risks. Multiple citations exist for both positions but adding more references would not add much to a paper that reports actual empirical data.

My Own Take

After having read hundreds of flawed studies over the years on dual use, all with the same main problems I outlined in the beginning, I think the Pesola et al. study is enormously valuable and advances the current evidence base in important ways:

• They follow a preselected group of people over time. This allows “complete capture” of their trajectories and avoids the problem I described above where population surveys “miss” people who dual used briefly and successfully switched.

• The outcomes are within-person changes, so they are in reference to participants’ respective baselines, which avoids the apples-to-oranges comparison.

• They show harm reduction even within the category of dual use: dual use often involves substantial reduction (of 50% or more) of cigarette smoking, which pushes back against the overly simplistic idea that dual use is inherently risky.

Health outcomes were outside the scope of this smoking-cessation RCT, but obviously, the quadrupled rate of stopping smoking (assuming it persists) in participants who dual-used will extend to improved health down the line, compared to people who exclusively smoked.

Of course, this is only one fairly small study, but it’s a different (and much-needed) type of study design than most studies on dual use (e.g. analyses of cross-sectional survey data). Since those studies are all wrong in the same way, we don’t need any more of that evidence (which makes us “precisely wrong” as Marcus Munafò puts it).

Not only is Pesola et al. a stronger study design in its own right, but it’s also a different study design from other research, which is important for triangulation of evidence. If a different study design is producing contradictory conclusions, then that means the prevailing conclusion is incorrect or incomplete and scientists need to re-evaluate or conduct new research.

What contradictory evidence across different types of studies certainly doesn’t mean, is that empirical conclusions should be overturned because a different body of research came to a different conclusion.