category factcheck
score Pants on Fire 🔥🔥🔥🔥🔥
claim "Association of e-cigarette use to prediabetes risk"
author American Journal of Preventive Medicine / Shyam Biswal
tags ['third-party-fact-check', 'source-verification', 'popcorn-news', 'ergo-harmful', 'correlation-causality', 'pr-study']


Misattributing "prediabetes" to e-cigarettes

Study PR claims to have found statistical link between e-cigarettes and heightened blood sugar ("prediabetes"). What it mostly compared was smokers to smokers who (recently?) transitioned to NVPs. Who actually turned out to haved a lower likelihood than smokers. As the study notes, nicotine itself effects the observation. The vaping causality insinuation doesn't hold up (post-smoking levels).

  • Ignores smoking history confounders (apart for undercredited NVP-only stat)
  • Smokers often retain their heightened blood sugars after quitting/switching
  • It's just a survey summarization ("self-reported"), not based on actual blood tests
  • Including no previous or changed diabetary habits (but counting one-time vaping for categorization)
  • Summary omits that e-cig user have half the probability raise (1.25 vs 1.48) according to the disclosed stats nonetheless
  • Some of the survey respondends are still smoking alongside e-cig use (albeit model and conclusion/stats accounted for that)
  • "Prediabetes" is a bit of a smokescreen diagnosis
  • Research protocol/log is not published, so unclear how blood sugars became focus of this article.
  • In order to avoid misdiagnosis there, potential conflicts of interest find no mention for once.

Not checked if the stat analyses actually hold up. (Normally you would wait for a pubpeer or SMC review, rather than relying on the journals. Though this isn't from UCSD, and number fudging unlikely given the results.)

PR exaggeration

In typical BSoPH-fashion, the accompanying press release portrays the statistical non-finding as causal link. Whereas the paper itself said "BRFSS is a cross-sectional survey, so a causal relationship between E-cigarette use and prediabetes cannot be inferred."

Hence the gratuitous application of typical weasel phrases:

  • Evidence links
  • use is associated with increased odds
  • Our study demonstrated a clear association
  • With both e-cigarette use and prevalence of prediabetes dramatically on the rise (equivocation)
  • To determine the association between (the determination shows)

No reputable scientist questions the starkly diminished risk of e-cigs in comparison to combustible tobacco. But this is a Bloomberg-funded study of course:

  • challenges the promotion by some of e-cigarettes as a “safer” smoking alternative
  • because they are touted as a safer alternative, which we now know is not the case, …
  • The use of e-cigarettes is rising among younger population, which remains a public health concern. (→ It actually fell by >60% the last two years. And it's more of an astroturf than a public health concern)

And then of course culminating in contradicting the results per muddled attribution:

In this representative sample of US adults, e-cigarette use was associated with greater odds of prediabetes compared to those who did not use e-cigarettes or traditional cigarettes.

(E-cig users had half the likelihood difference.)


Preconditions were apparently accounted for in the survey selection. The numbers do not substantiate the worry about NVPs posing a significantly higher risk. Extrapolating the average differentials into "25% more likely than non-smokers" is also a bit misrepresentative.

"Smoking" status "Prediabetes"
NVP Never 10.9 (10.7, 11.1)
NVP Current 9.0 (8.6, 9.4)
NVP Former 9.2 (8.4, 10.1)
Combustible Never 9.3 (9.1, 9.6)
Combustbl. Current 13.4 (13.0, 13.7)
Combustbl. Former 10.2 (9.8, 10.7)
Sole NVP users 5.9 (5.3, 6.5)
Dual users 10.2 (9.8, 10.7)

It's evident the study task wasn't presenting the discrepancy in chances, but piling onto the e-cigarette fearmongering.



Prediabetes is not a condition per se, but perhaps useful as nutritional diagnosis. It's a marketable term, with some disputes as to its ambiguous semantics and threshold levels. (Sometimes lowered on assumed behest of pharmaceutical interests. Or at least speculative predictive treatment.)


Methodological and lingustic misrepresentations aside, this might still be an interesting article. Calling attention to potential issues is always welcome. The problem here is undermining confidence in the scientific process. This habitual slander won't get affected patients more receptable to real findings, once they emerge. Insofar the floated PR deters smokers from quitting attempts, it clearly fails the hippocratic oath.

The embellishments in the press release and unquantified conflicts of interest (Johns Hopkins is funded by an anti-harm reduction ideologue) are quite unsurprising. Hence "pants on fire".

Other potential issues in article


Study was spread in some news articles with varying degrees of review or adulation:

low-quality indicators

Related news posts were naturally shared by: PAVe (1499461180074639372) and Kulikov (1499417533237108736).
But interestingly even CTFK and TI abstained from this one however.