Dharma Bhatta and Stanton Glantz last year published a study claiming that e-cigarettes cause heart attacks. That assertion was based on 38 heart attack incidents, but the authors knowingly included people who had heart attacks years before first using e-cigarettes.
My economist colleague Nantaporn Plurphanswat and I described that fatal error in two letters to editors of the Journal of the American Heart Association. I applied pressure for the next seven months, and the editors retracted the study on February 18 of this year.
Although the incident was reported by USA Today’s Jayne O’Donnell, Ivan Oransky at Retraction Watch, Alex Norcia of Vice and others, the tobacco research community showed little interest. Despite the obvious research flaw, some insisted that we show the effect of correctly classifying vapers who had heart attacks years before starting e-cigarette use.
Dr. Plurphanswat and I have met that challenge, and we now report our results in the professional journal Addiction.
From the public-use dataset we know that at least 11 of Bhatta-Glantz’s reported 38 heart attacks occurred prior to e-cigarette use. But there are many more. For our re-analysis we determined the exact number of prior heart attacks, but we cannot disclose that figure due to restrictions imposed by the Inter-university Consortium for Political and Social Research. Following are our results after correctly classifying all of the prior heart attacks.
| Adjusted Odds Ratios For Heart Attack According to E-Cigarette Use | ||
|---|---|---|
| E-Cigarette Use | Bhatta-Glantz Original Results | Results After Correct Classification |
| Never | Referent | Referent |
| Non-Daily Current | 1.99 (1.11 – 3.58) | 0.18 (0.05 – 0.66) |
| Daily Current | 2.25 (1.23 – 4.11) | 0.69 (0.22 – 2.12) |
Our work reveals that Bhatta and Glantz’s doubled odds ratios disappear after correctly classifying the time-warped e-cigarette heart attacks.
Our article describes additional problems with the Bhatta-Glantz study, such as the fact that their definitions were internally inconsistent. They excluded current experimental e-cigarette users and smokers, but they included former experimental e-cigarette users and smokers, with no justification. Additionally, participants were not counted correctly. Bhatta-Glantz used all 32,320 survey participants in their analyses, but 7,183 of those had missing information on key variables; we don’t know what Bhatta and Glantz did with them, but they should not have been counted.
We note, “[Bhatta and Glantz] appeared to know about detailed information regarding when participants were first told about the heart attack and when they first used e-cigarettes. On page 9 of their [retracted] article, the authors said that they used those specific PATH survey questions to ‘address this problem’ of temporality with a secondary analysis ‘to select only those people who had their first MIs [myocardial infarctions] at or after 2007’ (n = 16, their Supporting Information, Table S6). Glantz also told a journalist that the secondary analysis reflected the timing of the heart attacks. However, despite this knowledge and claim, their secondary analysis did not address the temporality problem. We found that more than one-third of the current 16 e-cigarette users who had a heart attack after 2007 occurred before they first used e-cigarettes.”
We conclude, “…[Bhatta and Glantz] did not account for heart attacks that occurred before first e-cigarette use, even after the temporality problem was raised in pre-publication peer review…The analysis presented here supports the decision by editors of the Journal of the American Heart Association to retract Bhatta & Glantz’s article on 18 February 2020. It raises questions as to what analyses the authors may have undertaken that were not reported and why they did not conduct the analyses reported here or requested during peer review of the original article.”
As I noted previously, the retracted study was supported by $13.6 million from American taxpayers. Our new research warrants an investigation by federal officials to determine if those funds were misused.
No comments:
Post a Comment