A study from New Zealand (full study text pdf, news report) was published earlier today in the high profile medical journal Lancet which compared the effectiveness of electronic cigarettes to nicotine patches for quitting smoking. It found e-cigarettes ‘…with or without nicotine, were modestly effective at helping smokers to quit, with similar achievement of abstinence as with nicotine patches.’ Let’s take a look at it and see if we can put the results in context…
The new research was a randomised controlled trial (RCT) comparing e-cigarettes with conventional medicinal nicotine patches. Randomised controlled trials are a helpful research tool, because randomising study participants to ‘treatments’ (here meaning whether they got an e-cigarette or got a nicotine patch) helps minimise many of the biases inherent in other study designs where participants get to ‘choose their own’ intervention. In studies which don’t use randomisation you tend to find participants differ in other ways aside from the one you are interested in examining, which ends up being a problem. (The Cabinet Office paper Test, Learn, Adapt is great for a brief introduction to some of the advantages RCTs have in helping us figure out what works.)
The researchers took 657 New Zealand smokers, and randomly allocated 289 to nicotine e-cigarettes (this brand), 295 to patches, and 73 to no-nicotine e-cigarettes. The patch is a good comparison group, because we already know the patch works better than doing nothing, so if e-cigarettes are as good or better than the patch, they’re going to be better than doing nothing also.
6 months later, the researchers found 7.3% of those in the nicotine e-cigarette group were abstinent from cigarette smoking, 5.8% of those who were provided patches, and 4.1% who got the no-nicotine e-cigarettes.
So it seems that the quit rate in the nicotine-containing e-cigarette group was a bit higher. But, the results between the nicotine e-cigarette and the patch were said not to be statistically significantly different. This is a bit of scientific jargon: a quick explanation as to what it means in this case…
The researchers say that ‘the absolute risk difference for the primary outcome [nicotine e-cigarette versus patch] was 1·51 (95% CI –2·49 to 5·51)’.
This means that, between the two groups, there was a 1.5 percentage point difference in 6 month cessation rates between the e-cigarette and the patch (that’s just 7.3% minus the 5.8%) but that, given the play of chance which affects all comparisons like these, the true difference could reasonably be expected to be as much as 5.5 percentage points, or even minus 2.5 percentage points (the latter would mean the effect runs in the other direction, that is, e-cigarettes truly being a little bit worse at helping people stop smoking than patches). Because this range of possible values is slightly more leaning toward the positive side than the negative, and various alternate analyses the researchers conducted, they conclude that e-cigarettes are probably about as effective as patches for achieving smoking cessation at 6 months, under these conditions.
This seems a pretty reasonable conclusion, as it seems unlikely, given the results of the study, that the e-cigarettes used in the trial are very much better at helping people stop smoking than patches at 6 months, nor are they likely to be very much worse. So what can we take from it?
One thing is that it confirms that unfortunately, for many people, attempts to stop smoking fail often no matter what help they use – the overall success rates are low (in the case of this study, particularly low – which surprised the researchers themselves). Of course, this doesn’t mean that things that only help modestly are a waste of time, they can still be very worthwhile. It also shows that some of the more extravagant claims about the effectiveness of e-cigarettes are not supported. They are not a miracle product offering guaranteed conversion away from tobacco cigarettes first time, every time.
But related to this, one issue to be clear on is that the results of the study only really applies to the brand that was tested (which was a fairly low-performance and basic e-cigarette model). E-cigarettes take many different forms and improvements in technology are rapid. If a higher performance ‘second generation’ device that provided better nicotine delivery or better consumer experience were used, results could be different. Hopefully following this study, others will design research looking at this.
A final point is that we need to consider what these results mean ‘out in the wild’. That is, in the real world outside of the context of a scientific study that imposes artificial constraints on people’s behaviour (which are necessary to find out what works best, but are nevertheless artificial). Some notable secondary outcomes from the study were that e-cigarette users were substantially more likely to keep using their products throughout the study than patch users, liked the tactile and sensory qualities of it, and were more likely to recommend it to friends. In the real world people choose (and keep using – or not) their stop smoking interventions themselves, rather than being ‘allocated’ to them by a scientist. So at the big-picture population level one product could be more effective than another if it is more popular and used by a greater number of people than its competitor, even if, under controlled experimental conditions, they appear roughly equal.