Like all cultists, those who live and breathe vaping by telling everyone at every opportunity that it has saved them, embrace a set of fervent beliefs. Vaping theology is a set of sacrosanct, inviolable beliefs that all adherents repeat regularly at risk being cast out of the vaping temple by other true believers.

At the end of this blog are another 12 of these creeds, with many more in preparation. This one looks at a belief that is never far from the lips of those who patrol policy debates on vaping where proposals or evidence threaten in any way to inhibit their mission.

Common liability theory

The garlic-encrusted crucifix hoisted high at the first syllable of any vampire-like suggestion that vaping might act as training wheels for children and teenagers to take up smoking is known as the “common liability hypothesis”.  The hypothesis first gained modest prominence in debates about the “gateway hypothesis” in drug uptake research where crude post hoc ergo propter hoc (after, therefore because of)  reasoning has often insisted that  (for example) that those who try cannabis and then later start using narcotics because they first smoked dope

We all appreciate that if one thing follows another it often does not mean the first thing caused the second. Breast milk is often followed by infant formula, water, then by fruit juice and clamour for carbonated drinks, then later by alcoholic drinks.  So can we say meaningfully that breast feeding causes Coke and alcohol consumption? Obviously not.

But there are plenty of examples of where one thing very much does greatly elevate the probability of another, satisfying several causal criteria. Prison incarceration is followed so frequently by reoffending on release that responsible legal sentencing practice tries to avoid imprisonment whenever reasonable to prevent crime. Intoxication and speeding so greatly increase the odds of motor vehicle crashes that deterrence and penalties are set high in most governments’ policies.

The current national concern about untrammeled betting advertising is seeing huge concern about the Pied Pipers of betting trying to lure starting punters into seemingly benign little flutters whiile turbo-charing promotions for multi betting. A classic example of gateway engineering.

Yes, there are always confounding factors that can be highlighted in such examples, but these seldom exonerate the critical role of an earlier variable (eg incarceration, drinking before driving).

When it comes to vaping and smoking, there are some entirely relevant observations. Both involve inhaling and exhaling nicotine through cylindrical delivery systems. Both involve the often rapid onset of signs of nicotine dependence in users. Both share a word (cigarette/e-cigarette) that seems to point to a similarity. Both involve repeated hand-to-mouth movements and a richly semiotic repertoire of holding and gesturing. Both see clouds of smoke or vapour billowing from their users, sometimes in clever displays. Both very frequently start in early teenage years. For some, both are important accoutrements of the passage from childhood to early adulthood, richly signifying and often peer group reputationally rewarding rebellion against parental controls and school rules.

With vaping and smoking, the common liability hypothesis posits that those children who vape and then subsequently start smoking would have mostly taken up smoking even if vaping had never been invented. It argues that kids who smoke in today’s smoking denormalised social environments have a propensity to be rebellious and so are also likely to take other risks: they vape, they smoke, try illicit drugs, have sex early, miss school, graffiti walls and so on. With the vaping “leading to” smoking debate, common liability adherents point to these propensities for kids who vape to be more likely to smoke simply because smoking is one of a constellation of adult-disapproved behaviours that bring peer status and petty prestigious notoriety to those seeking such distinction.

Nobel prize winning (2000) neuroscientist  Eric Kandell and his wife have described a molecular basis for nicotine being a gateway drug whereby nicotine in one form primes the brain to make it more susceptible to using other forms of nicotine. 

As a schoolboy, I smoked, got older kids to buy me alcohol, got suspended from school for buying beer on a school drama tour in year 11 and was the first in my year to have sex. I would have probably scored high on any scale of risk-taking a social psychologist might have pushed in front of me.

There have now been at least four systematic reviews/meta-analyses of the fast-emerging research literature on whether vaping increases the likelihood of taking up smoking.

Soneji et al (2017) JAMA Pediatrics: (9 studies) “The pooled probabilities of past 30-day cigarette smoking at follow-up were 21.5% for baseline past 30-day e-cigarette users and 4.6% for baseline non-past 30-day e-cigarette users.” (ie 4.7 times higher)

Baenziger et al (2021) BMJ Open (25 studies) “comparing e-cigarette users versus non-e-cigarette users, among never-smokers at baseline the OR for smoking initiation was 3.19 (95% CI 2.44 to 4.16, I2 85.7%) and among non-smokers at baseline the OR for current smoking was 3.14 (95% CI 1.93 to 5.11, I2 91.0%). Among former smokers, smoking relapse was higher in e-cigarette users versus non-users (OR=2.40, 95% CI 1.50 to 3.83, I2 12.3%).”

O’Brien et al (2021) BMC Public Health (14 studies) “our meta-analysis calculated a 4.06 (95% confidence interval (CI): 3.00-5.48, I2 68%, 9 primary studies) times higher odds of commencing tobacco cigarette smoking for teenagers who had ever used e-cigarettes at baseline, though the odds ratio were marginally lower (to 3.71 times odds, 95%CI: 2.83-4. 86, I2 35%, 4 primary studies) when only the four high-quality studies were analysed.”

Chan et al (2021) Addiction (11 studies) “a significant longitudinal association between vaping and smoking [adjusted odds ratio (aOR) = 2.93, 95% confidence interval (CI) = 2.22, 3.87]. Studies with sample sizes < 1000 had a significantly higher odds ratio (OR = 6.68, 95% CI = 3.63, 12.31) than studies with sample sizes > 1000 (OR = 2.49, 95% CI = 1.97, 3.15).”

All of the above reviews found that non-smoking children who had vaped at baseline had significantly increased odds of smoking cigarettes at follow-up, compared with those who had not vaped.

Yet in a recent editorial in Addiction, Pesko et al say that the public and health-care professionals pointing to the evidence in these reviews must be “confused”, writing:

“significant evidence now exists that this association between vaping and smoking is not causal, which is a source of confusion for the lay public and health-care professionals. Survey data show youth cigarette use declining steadily despite vaping increasing. When past-30-day youth e-cigarette use rates were as high as 32.9% in 2019, youth smoking rates should have been rising if the SG’s statement that ‘e-cigarette use is associated with the use of other tobacco products’ represents a causal relationship. Instead, by 2021 the youth cigarette use rate fell to a record low 1.9%.”

So why not settle the question with a randomised controlled trial?

All the studies reviewed in the four reviews above were observational longitudinal studies. In the first week of epidemiology training, every student is required to write out 1000 times at pain of death that “association does not equal causality” and play the sport of finding unwarranted causal inferences in observational study reports like those reviewed above.  It’s only in randomised controlled trials that authors are  given a gold pass to start suggesting causality.

Randomised controlled trials (RCTs) are venerated in medical and behavioural science because an important goal of randomisation is to disperse biases randomly across trial participants. Randomisation  theoretically eliminates confounding factors that may play dominant roles in determining outcomes, even ahead of the influence of the key intervention variables of interest (eg vaping vs NRT or unassisted cessation in evaluating the effectiveness of ways of quitting).

Because variables like age, sex, education, smoking in one’s family and peers, or personality traits like determination, self-efficacy or parenting styles are seen as likely to be important in how a person traverses decisions to smoke, drink or take drugs, randomisation — particularly in large trials — is designed to randomly spread the allocation of such variables across different arms of a trial (ie those receiving  an active drug – here, nicotine vapes – and those not vaping), in theory thus eliminating their influence.

But of course there will never be a randomised trial of vaping in children. No research ethics committee is ever likely to consent to such a trial because it would mean that researchers would be requiring randomised trial volunteer minors to start using highly addictive nicotine vapes. With smoking rates in early teens fast approaching zero in some nations like the USA, Canada, Australia, the UK and New Zealand,  imagine the outcry if a research group wanted to risk addicting nicotine naïve children to nicotine.

Imagine further, the unlikely event that a study group was able to recruit a large number of parents who would give their full consent to their children being given vapes like this. Not even in the most totalitarian of political regimes would we find such behaviour condoned.

Knowing this, those who stridently insist that the available data on transitions to smoking in young vapers is associative but not causal, know that they will always be able to use this policy fire extinguisher and train their “association” hoses on the worrying fires of gateway claims. This is a devious game intended to perpetually dismiss concern about collateral damage to kids arising from policies that allow them very easy access to vapes.

Controlling for “propensity to smoke”

With RCTs out of the question (just as they are for example, with randomising drivers to get intoxicated to see if they really do have more crashes in real world conditions than those who’ve not consumed alcohol), the next best evidence available is when researchers control their analyses for the very ‘propensity to smoke’ factors gateway critics say are the real determinants of smoking uptake.

Here we have several studies which have set out to do just that. Let’s take two recent examples.

Using US PATH study data, Osibogon et al (2020) looked at 2 years follow-up of 12-17 year old non-smokers who were vaping and those who were neither vaping nor smoking.

They found that current e-cigarette users (cigarette non-current users) at baseline were 5 times more likely to become regular cigarette smokers at 1-year follow-up than non-vapers. However, this association was not significant at the second year of follow-up. In reaching this conclusion, the authors controlled  for variables known to be associated with progression to smoking in youth.

A second example is  a 2021 paper on the UK’s  huge Millennium Cohort. It found:  

“Among youth who had not smoked tobacco by age 14 (n = 9,046), logistic regressions estimated that teenagers who used e-cigarettes by age 14 compared with non-e-cigarette users, had more than five times higher odds of initiating tobacco smoking by age 17 and nearly triple the odds of being a frequent tobacco smoker at age 17, net of risk factors and demographics.” [my emphasis]

Most importantly, the paper also deflated the glib ‘kids who try stuff, will try stuff’ common liability theory dismissal of the concern that vaping acts as training wheels for later smoking uptake. In their analysis, the authors controlled for a rich constellation of ‘propensity’ factors that have been suggested to predict smoking uptake in youth. These included parental low educational attainment and employment status; parental reports of each child’s behaviour during the prior 6 months using the Strengths and Difficulties Questionnaire, with indicators of externalizing behaviours (i.e. conduct problems, hyperactivity, inattention; and internalizing behaviours (i.e. emotional symptoms, peer problems) parental smoking; whether a child spent time ‘most days’ after school and at weekends hanging out with friends without adults or older children present. Children, via confidential self-reports, indicated whether they had ever drunk alcohol (more than a few sips), ever engaged in delinquency (e.g. theft, vandalism) and whether their friends smoked cigarettes.

In a huge blow to common liability adherents the authors concluded:

“we found little support that measured confounders drove the relationships between e-cigarettes and tobacco use, as the age 14 e-cigarette and tobacco cigarette estimates barely changed with the inclusion of confounders or in matched samples. Furthermore, early e-cigarette users did not share the same risk factors as early tobacco smokers, as only half the risk factors distinguished e-cigarettes users from non-users, whereas age 14 tobacco smokers were overrepresented on almost all the antecedent risk factors. If there was a common liability, we would expect similar over-representation for users of both forms of nicotine.”

Pesko et al didn’t reference these inconvenient papers either.

Pesko et al’s paper was an editorial, not a systematic review, It cited none of the above reviews nor indeed any of the papers in those reviews, instead basing its glib dismissal of that evidence as “based on statistical association rather than clear evidence of causality”.

Instead it enlisted another old chestnut: that the gateway hypothesis conclusions are simply incompatible with the fall in smoking prevalence in youth when their vaping is rising. (“youth smoking rates should have been rising if the Surgeon General’s statement that ‘e-cigarette use is associated with the use of other tobacco products’ represents a causal relationship. Instead, by 2021 the youth cigarette use rate fell to a record low 1.9%”

This argument is frankly very feeble. I dealt with it in a 2018 paper with two colleagues in Nicotine & Tobacco Research. It relies on an assumption that the net population impact of any putative gateway effect of e-cigarette use would be larger than the combined net impact of all other policies, programs and factors which are responsible for reducing adolescent smoking prevalence (e.g. tobacco tax and retail price, measures of the denormalisation of smoking, exposure of children to adult-targeted quit campaigns, retail display bans, health warnings and plain packaging) and the important synergies between all of these.

Many nations have seen pleasing and continuing falls in adolescent smoking commence well before the advent of widespread vaping. Vaping is thus far from being the only factor responsible for declining smoking.

But this is the ridiculously  high bar that gateway critics demand that anyone suggesting gateway effects needs to jump over. If smoking is falling, the suggestion is that the uptake of vaping is causative. Note here too the ease with causal attribution from ecological data is invoked when it suits one’s purpose.

The combined impact of the abovementioned factors in preventing smoking uptake could easily mask considerable smoking uptake that might not have occurred in the absence of e-cigarettes. That uptake may not be big enough to reverse net falls in smoking prevalence which has seen hundreds of thousands of children and adolescents not take up smoking in nations where it has happened.

But the undeniable consistency in observational cohort studies, almost without exception, shows that if you don’t smoke and do vape, you are far more likely to smoke later, even when “propensity to smoke” factors are adjusted in studies which have done this.

To keep repeating these discredited slogans (“kids who try stuff, will try stuff”, “kids who are going to smoke, will smoke”) dignified by high-falutin’ hypotheses  like “common liability” that don’t survive first pass adjustment for their assumptions, discredits those who continue this narrative.

Other blogs in this series

Vaping theology: 1 The Cancer Council Australia takes huge donations from cigarette retailers. WordPress  30 Jul, 2020

Vaping theology: 2 Tobacco control advocates help Big Tobacco. WordPress 12 Aug, 2020

Vaping theology: 3 Australia’s prescribed vaping model “privileges” Big Tobacco WordPress Feb 15, 2020

Vaping theology: 4 Many in tobacco control do not support open access to vapes because they are just protecting their jobs. WordPress 27 Feb 2021

Vaping theology: 5 I take money from China and Bloomberg to conduct bogus studies. WordPress 6 Mar, 2021

Vaping theology: 6 There’s nicotine in potatoes and tomatoes so should we restrict or ban them too? WordPress 9 Mar, 2021

Vaping theology: 7 Vaping prohibitionists have been punished, hurt, suffered and damaged by Big Tobacco WordPress 2 Jun, 2021

Vaping theology: 8 I hide behind troll account. WordPress 29 Jun, 2021

Vaping theology: 9 “Won’t somebody please think of the children”. WordPress 6 Sep, 2021

Vaping theology: 10: Almost all young people who vape regularly are already smokers before they tried vaping. WordPress 10 Sep, 2021

Vaping theology: 11 The sky is about to fall in as nicotine vaping starts to require a prescription in Australia. WordPress 28 Sep, 2021

Vaping theology: 12 Nicotine is not very addictive WordPress 3 Jan 2022