Thursday, December 10, 2015

Is the Harvard E-Cigarette Buttery Flavor Study Credible?

A month ago, I warned vapers to avoid e-liquids containing buttery flavorings diacetyl (DA) and acetyl propionyl (AP, also known as 2,3-pentanedione) – chemicals associated with bronchiolitis obliterans, a serious, sometimes fatal lung disease seen in workers producing buttered popcorn. 

This week, faculty at the Harvard School of Public Health reported that DA, AP and/or acetoin (a similar flavoring) are present in popular e-cigarettes’ aerosols.  The study, whose lead author is Joseph G. Allen, appears in Environmental Health Perspectives (abstract here).

If true, the report would implicate major e-cigarette manufacturers, but there are irregularities in the researchers’ methods and findings. 

Allen used a custom lab device “that drew air through the e-cigarette for eight seconds at a time with a resting period of 15 or 30 second [sic] between each draw. Eight seconds was chosen to make certain that each draw had adequate time for the entire contents to be forced out of the smoking device and through the sampling media…The samples were collected until the e-cigarette cartridges or cartomizers were exhausted, determined by the lack of visible emissions in the chamber.  (emphasis mine)  In other words, the researchers used an aggressive procedure that vaped the samples dry.  Each “puff” was eight seconds long – an unrealistic length of time that could have produced higher-than-normal temperatures, combustion and smoke.

This is significant because at least two previous studies by Fujioka and Shibamoto (here) and Pierce et al. (here)  found that cigarette smoke contains both DA and AP.  Pierce reported that the smoke of a single cigarette has up to 1,037 micrograms of DA and 165 micrograms of AP.  She noted that DA is approved as a cigarette tobacco additive in Germany and the UK, and AP is also permitted in the UK.  However, according to Pierce, this “strongly suggests that pyrolysis products of tobacco and other cigarette components comprise a substantial fraction of the [DA] and [AP] concentrations in mainstream smoke.”  If the Allen e-cigarette methodology generated smoke, their results could have been compromised.

Another quirk of the Allen study was the resampling of “several of the same flavors from the same package (ie, testing two e-cigarette cartridges from the same pack)” for six of the flavors in the table at left.  It is troubling that the researchers failed to adequately explain why they ran these extra samples on only some products. 

More importantly, they didn’t offer a valid explanation for a wide divergence in values.  The researchers found no acetoin in one Pina Colada sample, but it was present in other samples at 17, 46 and 130 micrograms.  One Classic sample had no detectable acetoin, while another had 38 micrograms.  Allen suggested that these differences might be due to “variable chemical doses.” 

That explanation would be questionable if the sampled products were produced by major manufacturers with tight quality control systems, which seems to be the case here.

While, inexplicably, the Harvard researchers did not identify the e-cigarette aerosol brands under study, the brands can be inferred from the data. They are likely:

A – Mark Ten
B – Vuse
C – blu
E – White Cloud
F – South Beach Smoke
G – Volcano
H – Tasty Vapor
I – Unknown
Since Pina Colada is a blu e-cigarette from Lorillard, and Classic is a Mark Ten product from Altria, ingredients can be expected to be highly consistent among samples taken “from the same pack”, as was done in this study.  

Allen and colleagues also suggested that variability could be related to the fact the researchers “relied on a visual determination of emissions of the e-cigarette in the chamber” to determine that the e-cigarette was “fully spent.”  A maxim in toxicology is to use methods that have been validated for the products and agents under investigation.  Eyeballing a stop point is not a validated method.    

As I advised previously, vapers should only use liquids that are certified to be free of buttery flavors that are suspected respiratory toxicants.  However, laboratory investigations of e-cigarettes should use validated methods to assure credibility.  The results of the Harvard Buttery Flavor Study do not meet this standard.


Michael J. McFadden said...

Excellent and clear analysis Dr. Rodu! This is a piece of nonsense just about up there with the "Thirdhand Smoke" studies.

Two other good readings on it at Dr. Siegel's blog and Signs Of The Times at:


If you don't mind my sharing my SOTT comment here:

The study notes that the lowest level of exposure at which popcorn lung was detected in workers working for 8 hours per day was 0.2ppm. I believe that works out to 200 mcg/Liter. that's 200,000 micrograms per cubic meter. At about one meter average working air intake per hour, it was found that workers inhaling 1,600,000 micrograms of diacetyl every day for years began developing popcorn lung.

As opposed to this concern about Vapers inhaling about 10 micrograms.

Actually, if we reasonably dropped the weird high-level outlier of 238 micrograms supposedly measured for the kid-loving-candy-flavored "Peach Schnapps" variety (Golly but those kidz DO luv their Schnapps, now don't they?) the average exposure drops to below FIVE micrograms.

So the ordinary vaper of these weird flavored fluids would have to sit around vaping for roughly 160,000 days (i.e. FOUR HUNDRED AND THIRTY-EIGHT **YEARS**) to get just a one day equivalent exposure to the workers who need to work for years on end to get the dreaded popcorn lung. How many years? It's not clear what the minimum might be, but from the information supplied let's *guess* we're talking about popcorn workers generally working at least ten years before they're buttered up and put away.

So how many days of e-cigging would it take for e-ciggers to get that sort of nasty worker dose? 160,000 x 10yrs x 300 workdays per year equals:

584,000,000 days of puffing away on an ecig (about one and a half million years) before the typical e-cigger might get popcorn lung.

Feel free to check my figures: I *have* been known to accidentally drop/add a decimal point here or there as I do a lot of this stuff in my head... but realize this: even if I'm off by a full order of magnitude we'd still be looking at 160,000 years of puffing. If I was off by THREE orders of magnitude, it would STILL take the better part of 2000 years of constant puffing.

The jury is still out as to whether ecigs might be somewhat harmful or somewhat beneficial to health, but if they're truly beneficial enough to extend the average life span to over fifteen hundred years...

Well hell's bells on a trampoline, I'd take my popcorn lung at that point with a cherry on top and a big silly smile!

EDIT: Heh, I *knew* this study reminded me of something! If you look at my analysis of the diethylene glycol study at (The study that spurred all the "Antifreeze In E-Cigs!" headlines around the world), you'll find the same pattern of researchers finding VERY low results in just about everything they tested except for ONE absurdly high result in ONE sample. Of course in their case emphasizing that sample only meant ignoring 17 other results. In the present study we're being asked to ignore over FIFTY other results!
EDIT TWO: I tracked down my 2007 article on popcorn fumes if you'd like to read about parents murdering their children by exposing them to popcorn fumes at the movies. See "Fumes 'n Fun, Popcorn 'n Sun" at (second article)

Odd that the ecig exposure wasn't compared to those sorts of exposure, eh?


John walker said...

Brad is it possible that the prolonged heating of the liquids tested turned other compounds in the 'flavor mix' into Diacyetyl or its precursor? Seems plausible, but am only a layman. Does seem that they have once again, set up a 'test' that proved that if you heat organic compounds, enough for long enough ,you can get smoke.

Bill Godshall said...

Perhaps Brad can explain why he believes that the trace levels of DA, AP and/or acetoin found in various vapor flavorings pose a risk for vapers to get Popcorn Lung (when far greater levels of those same constituents are in cigarette smoke, which has never been found to be associated with Popcorn Lung).

I consider warning vapers to not use flavorings that contain trace levels of DA, AP and/or acetoin (despite no evidence that vaping trace levels of those constituent pose any harm) basically the same speculative boycotting as warning smokeless tobacco users to stop using American made smokeless tobacco products (because they contain trace levels of nitrosamines, despite no evidence that using American smokeless tobacco products containing trace levels of nitrosamines pose harm to users).

Brad Rodu said...

In a previous post (, I wrote about “decades of definitive epidemiologic evidence for the relative safety of smokeless tobacco.” Although moist snuff products contained moderate levels of nitrosamines in the 1970s, these products were never associated with increased cancer risks in contemporary epidemiologic studies. Nitrosamines are contaminants in the production of moist snuff, and their level in today’s products is minuscule.

In contrast, buttery flavors are nonessential additives to e-vapor. I cannot explain why cigarette smokers don’t develop bronchiolitis obliterans, even though they are exposed to these agents. However, I have already linked to several scientific studies documenting the link between exposure to vaporized buttery flavors and that disease ( ; ; ; and

I remain convinced that vaping is far less hazardous than smoking with respect to cancer and cardiovascular diseases. But I am much less confident that long-term vaping will not be associated with any airway problems. Rather than simply wait for problems to develop, I think it is prudent to request that e-cigarette manufacturers list all ingredients for their liquids, and they should not add suspected or known respiratory toxicants. My advice to vapers is to avoid products with buttery flavors.