The journal BMC Public Health has announced that it will be retracting a paper it published last October which concluded that the use of electronic cigarettes has helped accelerate the decline in cigarette smoking.
The study, entitled "Population-level counterfactual trend modelling to examine the relationship between smoking prevalence and e-cigarette use among US adults," analyzed U.S. population-based data on trends in e-cigarette consumption and smoking prevalence from approximately 8 years prior to when e-cigarettes became popular in the U.S. through 2019. The authors used adult cigarette prevalence trends from 1999-2009 to establish a baseline and then generated the counterfactual (what would have been expected in the absence of e-cigarettes) by continuing these trends through 2019. They then compared the predicted trend in smoking prevalence from 2010-2019 with the actual trend.
They found major discrepancies between the predicted and observed prevalence of smoking, with smoking rates dropping much more rapidly than expected. The magnitude of the "excess" decline in smoking correlated highly with greater prevalence of e-cigarette use. Furthermore, discrepancies between expected and observed levels of smoking were greater among subgroups with higher levels of e-cigarette use: young adults, adult males, and non-Hispanic White adults. The article concluded that: "Population-level data suggest that smoking prevalence has dropped faster than expected, in ways correlated with increased e-cigarette use."
The data were obtained from the National Health Interview Surveys, which are publicly available. The authors are affiliated with Pinney Associates, which does consulting work for Juul, an e-cigarette company in which the tobacco company Altria has a 35% stake. The authors fully disclosed the funding for the study--which was provided by Juul--as well as their conflicts of interest (i.e., their consulting work for Juul) in the paper.
Apparently, the journal received a letter to the editor complaining that the paper's conclusions were invalid and calling for its retraction for two reasons:
1. The assumption that e-cigarette prevalence was zero in 2010 is incorrect, as e-cigarettes were widely available in the U.S. going back to 2007.
2. The paper was funded by Juul, which has a financial interest in the results of the paper.
However, the authors provided a supplemental table in the paper showing that even if one eliminates the assumption that e-cigarette use was zero in 2010 and simply rely on survey-measured e-cigarette prevalence values, there is still a strong and significant correlation between the discrepancy between predicted and actual smoking and the prevalence of e-cigarette use for two subgroups: males and young adults (the two groups with the highest levels of e-cigarette use).
Additionally, the authors fully disclosed their funding and conflicts of interest which were known to the journal's editorial staff and the reviewers prior to the decision to accept the manuscript for publication. Moreover, this particular journal does not have a policy that precludes consideration of papers that are funded by the tobacco industry or for which the authors report conflicts of interest.
Despite these two critical facts, the journal nevertheless wrote a letter to the authors stating that it made the decision to retract the article. On July 18, the journal told the authors that their response to the letter to the editor was not sufficient to address the letter writer's concerns. The reason given by the journal for the retraction was as follows:
The assumption that e-cigarette prevalence was zero in 2010 is not supported, and although the sensitivity analysis showed that there was still a significant correlation between the magnitude of the discrepancy between predicted and observed smoking prevalence and the level of population e-cigarette use for males and young adults when this assumption was eliminated, the relationship between cigarette use and the smoking prevalence disparity among young adults is most likely due to e-cigarette marketing.
The pivotal argument in the letter is as follows (I bolded it because of its importance): "In fact, bodies of literature are emerging globally about the effect of the tobacco industry’s marketing campaigns aimed at younger generations, using social media and influencers, which are associated with an increase in uptake of e-cigarettes (and dual use of e-cigarettes and tobacco products) in younger age groups. In my opinion, the effect of this strategic marketing seems a much more likely explanation of the association between younger age and increased e-cigarette use found by Foxon et al. than that e-cigarettes are particularly effective at helping younger age groups quit smoking."
The letter stated the final reason for the retraction as: "After careful consideration, and in light of the EBM’s feedback, the journal has taken the decision to retract the article in line with COPE guidelines. Our investigation has concluded that since the paper’s conclusions are based on assumption of ‘0’ prevalence of e-cigarette use in 2010 the results are non-significant."
The Rest of the Story
There is no valid scientific basis for the retraction of this paper. The journal states that it is retracting the article because the assumption that e-cigarette prevalence was zero in 2010 renders its results invalid. However, the results do not rely on the assumption that smoking prevalence in 2010 was zero. In fact, when the authors made no assumptions at all about e-cigarette prevalence and simply used the survey data on e-cigarette prevalence, they found a gaping difference between predicted and observed smoking prevalence among young adults and among males and in both cases, the magnitude of this gaping difference was highly and significantly correlated with the prevalence of e-cigarette use. The authors also documented to the journal editors that even if they changed the main assumption so that e-cigarette prevalence was 0 back in 2006 or 2007 or 2008 or 2009, the results were essentially unchanged.
Moreover, the validity of the paper's findings does not rely upon any assumptions about e-cigarette prevalence in the first place. The main finding of the paper is that there is a large discrepancy between predicted and observed smoking prevalence in the past decade, with observed smoking prevalence being substantially less than would have been expected based on pre-existing trends. And the magnitude of the discrepancy is greatest among males and young adults. There needs to be some explanation for these two findings. It is beyond refutation that one critical change that occurred during the past decade was that there was a dramatic increase in adult e-cigarette use and that this increase was most dramatic among males and among young adults. Thus, even without calculating any correlation between e-cigarette prevalence and the magnitude of the smoking prevalence discrepancies one can reasonably posit that the use of e-cigarettes is a likely explanation for the greater than expected declines in adult smoking. This is especially true given data from the NHIS suggesting that during the past decade, no fewer than 4 million adult smokers quit smoking successfully using e-cigarettes.
The journal editors simply dismissed the sensitivity analysis showing that the paper's findings are not dependent upon the assumption of 0 e-cigarette use in 2010. That analysis is presented within the paper itself, yet the editors dismiss it. And the primary reason they dismiss it is because they believe there is an alternative explanation for the correlation between e-cigarette use and the discrepancy in observed vs. predicted smoking prevalence among young adults. Specifically, the editor argues that this finding is explained by e-cigarette marketing, which resulted in increases in youth e-cigarette use. But that is no explanation at all. The only thing that e-cigarette marketing might explain is why there was an increase in youth vaping. In no way does that invalidate the finding that there is a gaping discrepancy between actual and expected smoking prevalence among young adults which correlates with the magnitude of e-cigarette use.
Moreover, the paper also found that without making any assumption about 0 e-cigarette use in 2010, there was a gaping discrepancy in observed vs. predicted smoking among males in the United States that was correlated strongly and significantly with the magnitude of e-cigarette use. The journal editors do not provide any refutation of, or alternative explanation for this finding.
Importantly, the journal editors are also incorrect when they state that they are retracting the article "in line with COPE guidelines." The relevant COPE guideline states that a paper may be retracted only if the editors "have clear evidence that the findings are unreliable, either as a result of major error (eg, miscalculation or experimental error), or as a result of fabrication (eg, of data) or falsification (eg, image manipulation)." None of those is present in this case, and the editors do not contend that a major error, miscalculation, experimental error, fabrication of data, or falsification of data is present. Thus, the editors are not adhering to COPE guidelines at all.
Essentially, the paper is being retracted because the editors do not agree with the conclusion reached by the authors given the findings presented in the paper. This is not a valid reason to retract a paper. There are always alternative explanations for a paper's findings and if editors started retracting every such paper, they would be purging about 95% of the publications in their journals.
You, like me, may find this an egregious example of an attempt to purge the scientific literature of a paper that the editors, after the fact, have decided that they don't like the study findings. You may be wondering what is really behind this because the scientific explanation given falls flat on its face.
Well, the editor has given us a clue in another statement they make in the letter to the study authors explaining that a letter to the editor has been received. That statement, which quotes a WHO document, reads:
"At the time of writing, the evidence is insufficient to recommend the use of ENDS as cessation devices at the population level."
This conclusion is not only irrelevant to the issues at hand, but it has the appearance of revealing the true reason for the editor's disdain for this paper which they accepted for publication. Apparently, the editor sides with the WHO in believing that electronic cigarettes should not be recommended for smoking cessation. In other words, the editor appears to have a disdain for electronic cigarettes themselves, and this appears to have translated into a disdain for the article reporting the dramatic and unprecedented effects of e-cigarettes on adult smoking cessation.
I therefore read this story as indicating that the paper was retracted for essentially political, rather than scientific reasons.