Not Available

We apologise but the Imperial Brands Science website isn’t available in your geographic location. Please visit our corporate site to find out more about the science we’re undertaking and the innovations we’re exploring.

Our Thoughts

//BTH #2 – “Vaping damages your lungs, but is still ‘safer’ than smoking”

Posted 23/12/2019 12:00am

Behind the Headlines is a regular feature taking a slightly deeper dive into a piece of recent media coverage of NGP research.

This edition’s story:

‘Vaping damages your lungs, but is still safer than smoking, suggests long-term study

On December 16, 2019 the Daily Mail stated: “E-cigarette users are 29% more likely to suffer emphysema and asthma than people who have never vaped or smoked tobacco”.

The story, also reported in Time magazine, drew on research published in the American Journal of Preventative Medicine conducted by Dr Stanton Glantz and Dr Dharma Bhatta. Controversially, the pair’s previous study, which associated EVP use with heart attacks, was recently retracted from its publishing journal over concerns the conclusions were “unreliable”.

How was the research conducted?

In Glantz and Bhatta’s new research paper the Population Assessment of Tobacco and Health (PATH) survey was used to determine the ‘longitudinal association’ between EVP (electronic vapour product) use and respiratory disease in 32,320 individuals.

What did the study find?

  • Among EVP users who did not report a respiratory disease at Wave 1 (September 2013 to December 2014), the authors found a statistically significant association between having an incident of respiratory disease at Waves 2 (October 2014 to October 2015) or 3 (October 2015 to October 2016)
  • Former EVP use had an AOR (adjusted odds ratio) = 1.31, (95% CI=1.07, 1.60)
  • Current EVP use had an AOR = 1.29, (95% CI=1.03, 1.61)
  • Current combustible tobacco smoking had an AOR = 2.56 ( 95% CI=1.92, 3.41)
  • The total odds of developing respiratory disease for a current dual user (those who used both cigarettes and an EVP) was 3.30 (with no statistical analysis available). However, this was not an AOR, so it doesn’t consider other variables which may have influenced the result.
  • The authors concluded that EVP use is an “independent risk factor for respiratory disease”.
  • They also state that dual use is “more dangerous than using either product alone”.

Our take on the research…

So, how ‘dangerous’ is vaping according to this study? Well, despite the negative headlines around this piece of research, the study actually suggests current EVP users are at a lower risk of developing respiratory disease than smokers. As Andy Tan (from the Dana-Farber Cancer Institute) stated in Time: “this study actually does support the harm reduction potential of e-cigarettes ”.


Did the study have any limitations?

In our opinion, numerous significant ones. They include:

  • It’s important to consider the odds ratio (OR) value.

While the authors report an overall increase in the OR for EVP use, the values were still below 2in clinical epidemiology, an odds ratio below 3 isn’t usually considered a significant finding, since most studies are vulnerable to the influence of subtle factors which might be difficult to control for.


The OR for EVP users was significantly lower than for combustible tobacco smokers, supporting the harm reduction potential of vaping than continuing to smoke.


  • 3 out of the 4 of the respiratory diseases (COPD, chronic bronchitis, emphysema) investigated take decades to become clinically apparent.

A study that investigates the development of these respiratory events across a two-year period is unlikely to be able to accurately define causation.

  • Smoking history was not appropriately considered.

The authors state that they have statistically controlled for combustible tobacco use. However, their method is simplistic. It doesn’t consider lifetime smoking duration and frequency. The risks to respiratory health associated with smoking are therefore not appropriately qualified within the study.

  • Product use was not biochemically verified, and disease not confirmed by a clinician.

The study relies on participants self-reporting product use. Respondents may not remember this information accurately or they could feel pressured to give answers that they feel are socially acceptable. We refer to this as recall or reporting bias.

  • The authors artificially calculate the total odds for dual users.

Dual users are not a uniform group of people. The category is complex and made up of multiple levels of use. Artificially multiplying the risk from current combustible tobacco smokers and current EVP users, as the authors have done, does not consider the nuances between user groups; neither does it appropriately represent any potential risk to this group.

In addition, dual use risk values are not adjusted meaning that factors which may impact the analysis are not accounted for. e.g. family history. It’s poor practice to compare adjusted values with unadjusted values.

  • The authors combined the number of events in Wave 2 and 3 to “increase the power of the study”.

This was significant for two reasons:

  • In general, data are usually pooled when there are not enough data points for an analysis to be conducted or be statistically significant. This could suggest the study lacks enough information to properly conduct a robust analysis.


  • The study essentially looks at respiratory events over a two-year period. COPD, emphysema and chronic bronchitis take decades to become clinically apparent. It would be very unusual to develop these diseases in a two-year period. In addition, these illnesses would have potentially been present, although undiagnosed, in many cases significantly before the study data was collected or even before EVPs became available on the US market (approximately around 2007).
  • The definition of ‘current’ EVP users is problematic.

The authors grouped all individuals who ‘ever used’, ‘ever used fairly regularly’ and ‘currently used every day or some days’ as ‘current users’. Grouping such diversity in vaping patterns together in one group is simplistic.

  • Confounders were not accounted for.

While the authors accounted for some confounders, they neglected others that could influence the results. Confounders are other variables which could explain an association. For example, whether individuals live within rural or urban settings, relevant occupational or home exposures (e.g. if they were regularly exposed to things like mould). Comorbidities, exercise and family history are also other important factors which were not considered.

  • Studies which use surveys to collect data inherently include certain limitations

For example, such questionnaires can be unreliable since respondents suffer from ‘survey fatigue’ in which, toward the end of the questionnaire, enthusiasm for finishing the questions reduces as does the quality of answers. In addition:

  • Reporting bias can also influence participants.
  • Answers are only taken from a specific point rather than over time
  • There’s no verification of product use and medical diagnosis.
  • PATH data was not designed to investigate respiratory health endpoints or to capture the complex information required for a detailed and reliable analysis of respiratory diseases potentially associated with EVP use, yet this is how the PATH data has been used.

You are free to share this content with credit to Imperial Brands under a Attribution-NoDerivatives 4.0 International (CC BY-ND 4.0) license.

Back to Our Thoughts