Science Wounded and Evidence Shaken: Brad Rodu Dismantles Stanton Glantz
There are moments when scientific critique becomes an act of resistance: the study by Rodu et al. is one of them.
We know that when science bends under the weight of agendas, knowledge doesn’t just weaken—it becomes a public hazard. We also know that not all evidence enlightens; some, carefully manipulated, obscure what it should reveal.
In this context, marked by epistemological disputes and ideological tensions surrounding the risks of vaping, a new analysis published in Internal and Emergency Medicine by Brad Rodu, Nantaporn Plurphanswat, and statistician Jordan Rodu levels an accusation as direct as it is well-documented against an influential meta-analysis led by Stanton Glantz.
The study's title leaves no room for ambiguity: Inaccurate and Misleading Meta-analysis of E-cigarettes and Population-based Diseases. What follows—across dense, meticulously referenced pages—is a relentless dismantling of Glantz et al.'s methodology, assumptions, and conclusions, whose authority is severely undermined under the technical and ethical scrutiny presented here.
The criticized study: plenty of numbers, little precision
The meta-analysis by Glantz and his colleagues, published in NEJM Evidence, appeared to display the strength of a conclusive study: 124 odds ratios (ORs) drawn from 107 investigations, covering both cross-sectional surveys and longitudinal studies.
ORs, or odds ratios, are a statistical measure that compares the likelihood of an event occurring in an exposed group versus a non-exposed group—in this case, cardiovascular or respiratory diseases among electronic cigarette users and non-users.
At first glance, the data's scale and the publication's prestige seemed to shield its authority. Its main claim—that electronic cigarettes are no less harmful than traditional combustibles in terms of cardiovascular and respiratory risk—was received as a scientific verdict.
However, as Nantaporn and the Rodus argue, that conclusion does not rest on robust empirical truth, but on a methodological framework that is fundamentally unsound. Behind the statistical scaffolding, they argue, lies a tangle of conceptual errors, temporal omissions, and analytical choices that fundamentally compromise the study’s reliability.
The authors identify three structural flaws that, taken together, undermine the validity of the meta-analysis:
The first is the arbitrary mixing of heterogeneous diseases, grouped under common labels without clinical or methodological justification. The study combines conditions as disparate as erectile dysfunction, potentially fatal cardiovascular diseases, influenza, and chronic obstructive pulmonary disease (COPD), as if they shared the same etiological profile or a typical risk pattern.
This forced amalgamation distorts the meaning of the aggregated results and introduces a measurable bias. The inclusion of studies on erectile dysfunction—a multifactorial condition whose link to vaping is, at best, speculative—artificially inflated the risk estimates attributed to electronic cigarettes. The result: an alarmist statistical portrait that cannot withstand rigorous clinical scrutiny.The second structural flaw identified by the authors is the extensive and inappropriate use of cross-sectional studies that lack temporal information about exposure and diagnosis.
Most ORs included in the meta-analysis come from surveys like the National Health Interview Survey (NHIS) or the Behavioral Risk Factor Surveillance System (BRFSS), in which data on e-cigarette use and disease presence are collected simultaneously. This simultaneity eliminates the possibility of determining whether use precedes or follows diagnosis, making it impossible to establish even minimally reliable causal sequences.
In epidemiological terms, this is a grave error: without knowing the chronology of exposure and outcome, any causal inference becomes a mere guess. Despite this fundamental limitation, Glantz et al. treated these data as if they provided solid evidence, when in fact they amplify uncertainty and weaken the empirical foundation of their conclusions.The third structural deficiency lies in the poor use of longitudinal studies, which—in theory—should offer stronger causal insights due to their temporal tracking of participants. However, even the cohorts drawn from the Population Assessment of Tobacco and Health (PATH) survey were analyzed without rigorously adjusting for changes in consumption behavior over time.
This omission is no minor flaw: in a field where usage trajectories—initiation, cessation, relapse, dual use—are dynamic and decisive, ignoring those transitions is methodologically akin to watching a film by freezing the frame on a single still.
By treating as static phenomena that are, by definition, variable, the authors of the meta-analysis dilute the inferential power of longitudinal data, undermining the validity of their findings even in what should be their most promising form of evidence.
Statistical logic cannot rescue bad data
One of the most striking—and philosophically unsettling—elements of Rodu et al.'s article is its warning against the statistical mirage: the illusion of truth conjured by a number cloaked in significance. The statistically significant odds ratios reported by Glantz et al.'s meta-analysis may appear, to untrained eyes, as synonymous with solid scientific evidence.
But statistics alone do not generate truth: the strength of its conclusions depends on the quality of the data that feeds it. If datasets are misclassified, if structural biases—such as prior smoking history, which in many cases predates e-cigarette use—aren’t controlled for, the numbers don’t clarify; they distort. What appears to be a finding may, in reality, be an empty numerical construct—a sophisticated form of noise dressed in mathematical precision.
The authors of the critique invoke a warning long voiced by Egger et al. in their seminal work on meta-analysis: “garbage in, garbage out.” It’s a verdict as brutal as it is accurate. If the studies included in an analysis are methodologically flawed, no statistical model—no matter how sophisticated—can alchemize shaky data into reliable knowledge.
Apparent technical complexity does not redeem the fragility of the underlying material; on the contrary, it can conceal it behind a façade of rigor. Rodu and his colleagues emphasize that meta-analysis, far from being a neutral exercise in numerical aggregation, demands rigorous critical judgment about the quality of its sources. Ignoring this premise—as Glantz et al. arguably did—is not a minor oversight; it is a violation of the very principle that justifies the use of meta-analysis in scientific research.
Rodu et al.’s article doesn’t overlook a particularly telling detail: with sharp irony, it reminds readers that Stanton Glantz himself was forced to retract a previous study in 2020 for committing the very same methodological error he now repeats.
On that occasion, he had inferred a causal link between e-cigarette use and heart attacks, based on data that—once reviewed—revealed that many of the heart attacks had occurred before participants had even started vaping.
The chronology wasn’t merely implausible—it was impossible. This omission, which should have prompted more careful scrutiny in subsequent research, appears to have been overlooked. Thus, the current critique doesn’t merely highlight a technical flaw; it exposes a pattern of recurrence that undermines the author’s credibility and his commitment to scientific integrity.
The Bias of Time: Diseases That Take Decades to Develop
Another crucial point the authors firmly stress is the logical and epidemiological impossibility of detecting pathological effects of vaping within such short timeframes. Chronic diseases like COPD, myocardial infarctions, or strokes are not sudden or random events; in most cases, they are the cumulative result of decades of exposure to tobacco smoke.
To claim that effects of comparable magnitude could emerge after just a few years of e-cigarette use—especially in users who are, for the most part, current or recent smokers—is, in the words of the article itself, “epidemiologically unrealistic.”
The critique does not deny the need to study the potential risks of vaping, but it does demand rigor in timing and caution in interpretation. Confusing the possible with the probable, or the early with the definitive, is a form of distortion science should not tolerate.
Robust evidence, they argue, can only emerge from long-term follow-up of e-cigarette users who have never smoked combustible cigarettes. As of now, such cohorts are only just beginning to emerge.
A Technical Critique or a Scientific Reckoning?
Beyond the technicalities, the article by Rodu and his colleagues raises a concern that cannot be ignored or downplayed: How is it possible that such elementary—and profoundly consequential—methodological errors make their way into peer-reviewed studies published in high-impact journals?
The question is uncomfortable, but necessary.
Is this a case of scientific negligence, superficial editorial review, or are we facing something more troubling—the infiltration of ideology into the very processes by which evidence is produced?
When data are interpreted not to illuminate reality but to confirm a preexisting stance, the risk transcends academia and becomes societal. Science then ceases to be a tool for understanding and becomes an instrument of persuasion.
In this context, the critique by Rodu et al. is not merely a methodological reckoning—it is an urgent and necessary defense of the principles that should govern public health research.
For decades, Stanton Glantz has been an influential and polarizing figure in the field of tobacco control. Renowned for his relentless activism against the tobacco industry—a battle that has earned him admiration and notoriety—his career has also been marked by repeated methodological controversies.
His insistence on equating the harms of vaping with those of combustible tobacco has come under increasing scrutiny—particularly in recent years, as multiple toxicological studies have shown that while e-cigarettes are not harmless, they carry a significantly lower risk profile. This stance has sparked tension even among some of his closest peers in the field of tobacco control, including internationally respected figures like Neal Benowitz, Nancy Rigotti, Jamie Hartmann-Boyce, Steven Cook, Jonathan Livingstone-Banks, and Michael Cummings, who have publicly challenged his interpretations of the evidence.
Dr. Brad Rodu and his collaborators position their text squarely within this fraught landscape, not merely as a targeted technical critique but as a call to raise the standards of epidemiological research in a domain where bias can carry real public health consequences. Also—and no less importantly—as an explicit counterstrike against a narrative that, in their view, weaponizes science for political ends.
In that sense, the study demands not just methodological precision—it demands scientific integrity.
Science and Public Health: A Fragile Balance
This clash between studies is not merely an academic disagreement—it reveals a deeper, enduring tension at the core of contemporary public health policy. On one hand, the precautionary principle—vital in contexts of uncertainty—drives action in the face of potential risks, even when evidence remains inconclusive. On the other hand, the demand for scientific precision warns against turning conjecture into policy or weak correlations into assumed causality.
In the case of electronic cigarettes, this tension has become particularly acute. Fears that they may serve as a “gateway” to traditional smoking have led many regulators and experts to downplay—or even disregard—their potential as a harm reduction tool for adult smokers.
Thus, fear of the possible eventually eclipses the verifiable, and prevention, when divorced from rigorous analysis, risks becoming dogma.
But as the Royal College of Physicians in the UK clearly warns: “Nicotine, while addictive, is not the primary cause of smoking-related diseases; it’s the smoke.” Ignoring this distinction is more than a conceptual mistake—it is a form of political blindness that can have devastating consequences.
The risk, then, lies not only in misinterpreting data or applying flawed statistical models. The real danger is ethical: condemning millions of smokers to continued exposure to combustible cigarettes—the deadliest form of nicotine consumption—for fear of the wrong remedy.
In the name of misguided caution, dependence on known harm is prolonged, while the door to potentially safer strategies is shut. At that threshold between evidence and decision, science cannot abdicate its critical responsibility—nor its duty to speak, with data and nuance, what is uncomfortable yet urgent.
A Passionate Defense of the Scientific Craft
The article by Rodu et al. is more than an academic critique—it is a manifesto on how science must be conducted in contexts where data coexist with ideological tensions.
It reminds us—with the calm of rigor and the urgency of context—that epidemiology is as much an art as a science. Its power lies not in the accumulation of numbers or the complexity of models, but in the precision with which questions are posed—and, above all, in the honesty with which the answers are interpreted.
In times when evidence can be shaped by interests, pressures, fears, or algorithms, this study stands as an act of resistance: a meticulous defense of critical thinking against the lure of premature certainty.
The debate over vaping remains open—and must remain so. But if it hopes to serve the public interest rather than hidden agendas, it requires more than statistics: it demands rigor, humility, and a healthy dose of skepticism. Because health—like truth, as the Rodus teach us—is not reached without doubt, but it cannot be built on falsehoods.
Rodu, B., Plurphanswat, N. & Rodu, J. Inaccurate and misleading meta-analysis of E-cigarettes and population-based diseases. Intern Emerg Med (2025). https://doi.org/10.1007/s11739-025-03956-w