February 18, 2020

Stanton A. Glantz, PhD

Journal of American Heart Association caves to pressure from e-cig interests

In June 2019, after peer review, the Journal of the American Heart Association published “Electronic Cigarette Use and Myocardial Infarction Among Adults in the US Population Assessment of Tobacco and Health” by Dharma Bhatta and me.  Based on a cross-sectional analysis of the Population Assessment of Tobacco and Health (PATH) dataset, we found an association between e-cigarette use and having had a myocardial infarction (heart attack) that was independent of the effect of any current or former cigarette smoking by respondents.

One potential problem with such cross-sectional studies of the health effects of e-cigarette use is that some of the heart attacks could have occurred before e-cigarettes were available.  We addressed this potential problem in the paper by conducting a subsidiary analysis limiting the data to people who had heart attacks after 2007, when e-cigarettes first became available.  This additional analysis is described in the published paper:

As discussed above, we cannot infer temporality from the cross‐sectional finding that e‐cigarette use is associated with having had an MI and it is possible that first MIs occurred before e‐cigarette use. PATH Wave 1 was conducted in 2013 to 2014, only a few years after e‐cigarettes started gaining popularity on the US market around 2007. To address this problem we used the PATH questions “How old were you when you were first told you had a heart attack (also called a myocardial infarction) or needed bypass surgery?” and the age when respondents started using e‐cigarettes and cigarettes (1) for the very first time, (2) fairly regularly, and (3) every day. We used current age and age of first MI to select only those people who had their first MIs at or after 2007 (Table S6). While the point estimates for the e‐cigarette effects (as well as other variables) remained about the same as for the entire sample, these estimates were no longer statistically significant because of a small number of MIs among e‐cigarette users after 2007. Note that this analysis does not capture reinfarctions occurring after 2007, whose risk could be increased by e‐cigarette use as it is for continued smoking conventional cigarettes [citations removed].

In addition, the fact that e-cigarettes were not on the market when some people had their heart attacks actually biases the estimate of the effect of e-cigarettes toward the null, i.e., makes it harder to detect an effect if one exists.

While the paper by Dr. Bhatta and me is not the only one reporting epidemiological, clinical, and biological links between e-cigarette use and heart disease, our paper has attracted considerable attention by pro-e-cigarette forces, initially led by Brad Rodu, a professor at the University of Louisville with a long record of working with the tobacco industry.    

Rodu and a colleague argued that the analysis described above was inadequate because the PATH restricted use dataset had the date of first heart attack and date at which people started using e-cigarettes and that we should have used these two dates to exclude cases rather than the approach we took. 

Indeed, one of the peer reviewers had suggested the same analysis.  As I detailed in a letter to JAHA, while there was some misunderstanding of the specific supplemental analysis requested by the reviewer, the analysis that we presented during the peer review process substantially addressed the question raised by the reviewer. As I wrote the editor, Dr. London:

In any event, it is important to keep in mind that this discussion is about a supplementary analysis, not the main analysis in the paper. As the paper states, restricting the data as we did substantially dropped the number of MIs and the supplemental analysis was not statistically significant. Reviewer 2 understood and accepted our supplementary analysis and, after we responded to the original comment, recommended publishing the paper as it is with primary analysis (which is based on the whole dataset) despite the issues discussed in this letter. [emphasis in original]

In addition, doing the additional alternative analysis will not change the main analysis in the paper, which the reviewers and editors accepted.

The normal protocol for raising a technical criticism of a paper would be to write a letter to the journal criticizing the paper.  If the editors find the criticism worth airing, they would invite the authors (in this case, Dr. Bhatta and me) to respond, then publish both letters and allow the scientific community to consider the issue.

Indeed, Rodu has published several letters and other publications criticizing our work, most recently about a paper I and other colleagues published in Pediatrics about the gateway effect of e-cigarette use on subsequent youth smoking.  (Such letter writing campaigns are a longstanding industry strategy.)

Rather than following this protocol, I first learned of Rodu’s criticism when USA Today called me for a response to his criticism.  I was subsequently contacted by the Journal of the American Heart Association regarding Rodu’s criticism.  I responded by suggesting the editors invite Rodu to publish his criticism in enough detail for Dr. Bhatta and I to respond, as well as accurately disclose his links to the tobacco industry.

Instead, the editors of the Journal of the American Heart Association demanded that Dr. Bhatta do additional analysis that deleted heart attacks before people may have used e-cigarettes as Rodu wanted rather than as how we did in the subsidiary analysis in the paper. 

Dr. Bhatta and I have no issue with doing such additional analysis.  Indeed, we prepared the statistical code to do so last November.  (I doubt that the results will be materially different from what is in the paper, but one cannot be sure until actually running the analysis.)

The problem is that, during the process of revising the paper in response to the reviewers, we reported some sample size numbers without securing advance approval from the University of Michigan, who curates the PATH restricted use dataset.  This was a blunder on our part.  As a result, the University of Michigan has terminated access to the PATH restricted use dataset, not only for Dr. Bhatta and me, but for everyone at UCSF.

As part of our effort to remedy our mistake, we have published a revised version of the table in question (Table S6 in the paper) deleting the sample size numbers that had not been properly cleared with the University of Michigan. (Doing so did not materially change the paper.)  We have also worked with the UCSF administration to take other steps to remedy the mistake.  UCSF is continuing to work with the University of Michigan to restore access.

Now, under continuing pressure from e-cigarette advocates (link 1, link 2), the editors of the Journal of the American Heart Association have retracted the paper because, without access to the PATH restricted use dataset, we have not been able to do the additional analysis. 

The editors also gave Dr. Bhatta and me the option of retracting the paper ourselves.  We have not retracted the paper because, despite the fact that we have not been able to do the additional analysis Rodu is demanding, we still stand behind the paper. 

In their retraction notice, the editors detailed this unusual situation:

After becoming aware that the study in the above‐referenced article did not fully account for certain information in the Population Assessment of Tobacco and Health [PATH] Wave 1 survey, the editors of Journal of the American Heart Association reviewed the peer review process.

During peer review, the reviewers identified the important question of whether the myocardial infarctions occurred before or after the respondents initiated e‐cigarette use, and requested that the authors use additional data in the PATH codebook (age of first MI and age of first e‐cigarettes use) to address this concern. While the authors did provide some additional analysis, the reviewers and editors did not confirm that the authors had both understood and complied with the request prior to acceptance of the article for publication.

Post publication, the editors requested Dr. Bhatta et al conduct the analysis based on when specific respondents started using e‐cigarettes, which required ongoing access to the restricted use dataset from the PATH Wave 1 survey.1 The authors agreed to comply with the editors’ request. The deadline set by the editors for completion of the revised analysis was not met because the authors are currently unable to access the PATH database. Given these issues, the editors are concerned that the study conclusion is unreliable.

The editors hereby retract the article from publication in Journal of the American Heart Association. [original article URL: https://www.ahajournals.org/doi/10.1161/JAHA.119.012317]

Retraction is a very serious step, which is warranted when, according to the International Committee of Medical Journal Editors, there is “Scientific misconduct in research and non-research publications includes but is not necessarily limited to data fabrication; data falsification including deceptive manipulation of images; purposeful failure to disclose relationships and activities; and plagiarism.”  (http://www.icmje.org/recommendations/browse/publishing-and-editorial-issues/scientific-misconduct-expressions-of-concern-and-retraction.html). 

Dr. Bhatta and I did not engage in any of these behaviors.  Indeed, in response to a letter from me objecting to the handling of this situation, the editors of the Journal of the American Heart Association specifically stated: “The retraction notice is intentionally absent of any language suggesting scientific misconduct.

The results in the paper are accurately analyzed and reported.  That is why we refused to retract the paper.

As I said earlier, we are still hoping to regain access to PATH so that we can do the additional analysis and put this issue behind us.

Additional information on pressure on JAHA (added on 23 Feb 2020)

In addition to USA Today, which, as before, wrote a story sympathetic to the supporters of e-cigarettes, Vice News ran a story “A Major Study That Fueled National Vape Panic Has Been Retracted.”  While it was flattering that Vice considered our paper “a major study,” the reality is that it is one of many studies that link e-cigarette use to cardiovascular disease and heart attacks going all the way from basic science and clinical studies through epidemiology.

Vice did not contact either of the authors for comment before writing its story.  This was a little unusual (even USA Today asked for comment), until I was reminded that Vice has a deal with Philip Morris to promote its products.  And part of this deal is that Vice would not talk to health experts critical of Philip Morris products.

The Vice article also included a link to a letter David Abrams and 15 others to JAHA supporting Rodu’s position.  The letter is mounted on the Reason.com website.  Reason is a libertarian think tank with a long history of supporting the tobacco companies on a wide range of issues and which has received funding from Big Tobacco.

The letter to JAHA on the Reason website was signed by:

1.            David B. Abrams, PhD

2.            Kenneth Michael Cummings, PhD

3.            George Davey Smith

4.            Konstantinos Farsalinos, MD, MPH

5.            Jonathan Foulds, PhD

6.            Abigail Friedman, PhD

7.            Thomas Glynn, PhD

8.            Peter Hajek, PhD

9.            Martin Jarvis, DSc OBE

10.          Robert Kaestner, PhD

11.          Ann McNeill, PhD

12.          Marcus Munafò, PhD

13.          Raymond Niaura, PhD

14.          David Sweanor, JD

15.          David Timberlake, PhD

16.          Kenneth Warner, PhD

The letter includes the following statement at the end:  “The signatories write in a personal capacity and declare no competing interests with respect to tobacco or e-cigarette industries.”  This is surprising, given that Abrams and Niaura both serve on the Scientific and Technical Advisory Council for Philip Morris International’s Foundation for a Smokefree World, which is part of PMI’s efforts to promote IQOS and cleanse its image.  One would have thought Abrams and Niaura would have disclosed their engagement with FSFW under Question 5 on the ICJME form, which asks “Are there other relationships or activities that readers could perceive to have influenced, or that give the appearance of potentially influencing, what you wrote in the submitted work?” 

As of Feb 22, 2020, FSFW had pulled down the link listing the committee members, but it is available on the Internet Archive.  I happened to know about the FSFW link; I did not systematically search for other tobacco or e-cigarette industry connections for anyone on the list.

This is the first time I have seen the Abrams et al letter. (The Journal of the American Heart Association did not provide it to me or ask me to respond to it.)  Nothing in this letter undermines the paper as published. 

In particular, we properly described our analysis and accurately reported the results, including a sensitivity analysis that excluded all cases (regardless of e-cigarette use) where the heart attacks occurred before 2007, when e-cigarettes first came on the market.  This approach is superior to the approach Rodu (and Abrams and colleagues) suggest, which would selectively delete or change data only for e-cigarette users. 

In addition, the letter misrepresents what the paper actually says.  Whereas Abrams and colleagues write “the authors went on to conclude that the association was causal and made policy recommendations regarding smoking cessation practice on that basis,” there is not such claim of causality in the paper.  Indeed, we were careful not to make such a statement.

It is worth noting that several of the people who signed the Abrams letter have criticized our work over the years, including several who have written letters outlining their criticisms to journals, which were published with our responses. 

I remain puzzled as to why the JAHA did not simply follow this normal scientific protocol and publish the Rodu and Abrams letters and allowed us to respond. 

The scientific community could have then considered both their criticisms and our responses and made their own judgment.

Rodu and Plurphanswat publish their criticism in Addiction (added 26 Feb 2021)

While JAHA never showed us the specific criticisms that Rodu and Plurphanswat submitted, Rodu and Plurphanswat later published their criticism to a different journal, Addiction, which gave Bhatta and me a chance to respond.

Their analysis involved selective revisions to the dataset in a way that destroyed the cross-sectional design of the study.  Specifically, they changed people who had heart attacks before they started using e-cigarettes from “exposed” to “unexposed” while leaving the former e-cigarette users, current and former smokers and control groups unchanged.  This approach does not follow standard statistical procedures for cross-sectional analysis. This nonstandard approach led to the biologically implausible result that nondaily use of e-cigarettes is significantly protective (by a factor of over 5) against having a myocardial infarction. Their discussion does not address this bizarre result. 

After seeing the details of their analysis, we are more confident than ever that the appropriate way to conduct a sensitivity analysis to account for the fact that e-cigarettes came on the market around 2007 is to drop all cases, including the smokers and control cases, where the myocardial infarctions occurred before 2007.  That is precisely what we did in our JAHA paper (Table S6 and associated discussion in the main text).

Addiction then published another letter from Rodu and Plurphanswat that ignored the substance of the defense of our work and the criticism of their analysis, saying that our “response attempts to focus on statistical practices, but this is about scientific judgment and conduct.”  We agree that the whole discussion is about the proper formulation and interpretation of statistical analysis. That’s why it is so telling that R&P didn’t respond to our criticism of their nonstandard analysis.

Unlike their initial critique of our paper, Rodu and Plurphanswat’s letter provides the following disclosure: “Since 2005, B.R. has been supported by the Kentucky Research Challenge Trust Fund and by unrestricted grants to the University of Louisville from tobacco manufacturers (Swedish Match AB, US Smokeless Tobacco Company, Reynolds American Inc. Services, Altria Client Services, and British American Tobacco).  N .P. has been supported by these grants since 2013.”

We completely and accurately reported our results, including several sensitivity analyses and a longitudinal analysis of the association between e-cigarette use and incident heart attacks (which did not find a significant association, probably due to low power). Even after seeing the details of Rodu and Plurphanswat’s criticism, we continue to stand behind our paper. 

Again, it is unfortunate that JAHA retracted the paper rather than publishing the criticisms and our response so that the scientific community could judge for themselves.

Add new comment

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.